AGARD-LS-182

AGARD
ADVISORY GROUP FOR AEROSPACE RESEARCH & DEVELOPMENT 7 RUE ANCELLE 92200 NEUILLY SUR SEINE FRANCE
Processed / not arocessed by DIMS

"

AGARD LECTURE SERIES 182

................................

signed.................. date

Fundamentals and Special Problems of Synthetic Aperture Radar (SAR)

Les Aspects Fondamentaux et les Problkmes Spkcifiques aux Radars ii Ouverture Synthetique ( S A R )

NOT FOR DESTRUCTION

This material in this publication was assembled to support a Lecture Series under the sponsorship of the Avionics Panel of AGARD and the Consultant and Exchange Programme of AGARD presented on 5th-6th October 1992 in Bad Neuenahr, Germany, 8th-9th October 1992 in Gebze-Kocaeli (nearIstanbul), Turkey and 26th-27th October 1992 in Ottawa, Canada.

Published August 1992

Dislribulion and Availability on Back Cover

AGARD LECTURE SERIES 182

Fundamentals and Special Problems of Synthetic Aperture Radar (SAR)

Les Aspects Fondamentaux et les Probkmes Specifiques aux Radars a Ouverture Synthetique (SAR)

The Mission of AGARD

I

According to its Charter, the mission of AGARD is to bring together the leading personalities of the NATO nations in the fields

of science and technology relating to aerospace for the following purposes:

- Recommending effective ways for the member nations to use their research and development capabilities for the common benefit of the NATO community:
~

Providing scientific and technical advice and assistance to the Military Committee in the field of aerospace research and development (with particular regard to its military application);

- Continuously stimulating advances in the aerospace sciences relevant to strengthening the common defence posture; - Improving the co-operation among member nations in aerospace research and development; - Exchange of scientific and technical information; - Providing assistance to member nations for the purpose of increasing their scientific and technical poteniial; - Rendering scientific and technical assistance, as requested, to other NATO bodies and to member nations in connection with research and development problems in the aerospace field. The highest authority within AGARD i s the National Delegates Board consisting of officially appointed senior representatives from each member nation. The mission of AGARD is carried out through the Panels which are composed of experts appointed by the National Delegates, the Consultant and Exchange Programme and the Aerospace Applications Srudies Programme. T h e results of AGARD work are rcported to the member nations and the NATO Authorities through the AGARD series of publications of which this is one. Participation in AGARD activities is by invitation only and is normally limited to citizens of the NATO nations.

Published August 1992 Copyright 0 AGARD 1992 All Rights Reserved ISBN 92-835-0683-9

Prinred by Speciuliscd Printing Semica Limired 40 ChigweN Lune, Laughton, Essex IG103lZ

Abstract
T h e Lecture Series will cover the field of airborne and spaceborne SAR with respect to its technical realization in order to convey the participants ideas and know-how on SAR, on its capabilities and on the technology necessary for the successful construction and application of airborne and spaceborne SAR systems. T h e basic principles of SAR will b e explained and its peculiarities, ambiguities and special effects will be highlighted especially in comparison with airborne and spaceborne Radar with Real Aperture (RAR). T h e influence of speed and altitude variations. lateral motions o n respective compensation possibilities will he presented. T h e antenna is a system related SAR component. Therefore, the influence of the antenna parameters o n specification and capabilities of SAR will b e considered Polarization effects and multi-polarization SAR are presently key-points for SAR development and SAR application as well a s qurstions on absolute SAR calibration. T h e advantages. necessities and limits of these topics will b e included.

Digital SAR processing is for SAR indispensible. Theories and special algorithms will be given along with basic processor configurations and different processing techniques on hardware and software bases. T h e simulation of SAR-systems as well as SAR-products will also b e a topic of the Lecture Series. This includes the simulation of S4R-techniques and SAR-components as well as the simulation of SAR images. Special SAR-methods like squint-, stretch- and spotlight-techniques for example will be presented in addition to the inverse SAR-techniques using the motion of targets instead of the motion of the radar. A presentation of the state of the art giving examples of presently planned and up to now realized airborne and spaceborne SAR with its application foreseen will conclude the Lecture Series. T h e Lecture Series should appeal to technically and technologically oriented engineers concerned with development of SAR and to scientists, who have to work with SAR for different applications as well as to students of both specialities who have already attained a high degree of knowledge in techniques and of remote sensing applications. This Lecture Series, sponsored by the Avionics Panel of AGARD, has been implemented by the Consultant and Exchange Programme.

111

...

Abriigii
Cc cycle dc conferences traitera du domainc des SAR airoportes et spatioportes du point d e vue de leur realisation tcchnique. Les conferenciers mettront a profit leur compitence pour presenter leurs idees concernant les SAR, leurs possibilitis, et Ics technologies qui sont i utiliser pour reussir I'industrialisation et la mise en oeuvre des systimes SAR a e r o p o r t b et spatioportes.

Les principes de base d e ces systemes seront exposes, ainsi que leurs particularites, ambiguitb et specificitis, en particulier par rapport aux radars aeroportes et spatioportes a ouvertiire reelle (KAR). L'influencc des deplacements latiraux ainsi que des variations d e vitesse et d'altitude sur les possibilitis respectives d e compensation sera egalement presentee.
L'antenne est un composant du systeme des SAR. Par consequent, I'influence des parametres d'antenne sur les ;specifications techniques et les capacites dcs SAR sera prise en consideration. A I'heure actuelle, les effets d e polarisation et de multipolarisation SAR doivent i t r e consideris comme des points c l i pour le developpement et les applications des SAR, d c msme que leur etalonnage absolu. Les avantages, les necessith et les limites de ces donnees seront examines.

Le traitement numerique est indispensable aux SAR. Des theories et des algorithmes specifiques seront proposis; ainsi que des configurations d e processeur d e base et differentes techniques d e traitement suus les aspects materiel et logiciel.
La simulation des systemes SAR et des produits SAR constitue un autre sujet de ce cycle de confircnces. Ce sujet comprend la simulation des techniques et des composants SAR, ainsi que la simulation d e l'imagerie SAR. Des methodes specifiques au SAR, telles que le deport antenne (squint) les impulsions etalees (stretch) et le mode telescope par exemple, seront presentis en complement des techniques SAR inverses faisant appel aux mouvement des cibles au licu du mouvcment du radar. Le cycle de confirences se terminera par une presentation d e I'etat d e I'art dans ce domaine, avec des exemplcs d e systimes SAR aeroportb, spatioportes existants et projetis et des applications previsibles pour lesquelles ils ont et6 concus.

Ce cycle de conferences est susceptible dinteresser les ingenieurs travaillant sur IC developpement des SAK et les scientifiques a p p l e s 6 travailler avec les SAR pour divcrses applicatkins ainsi que les itudiants ayant dija dcs connaissances avancees dans les deux domaines des techniques en question ainsi que des applications en teledetection.
Ce cycle de confCrenccs est prisenti par le Panel AGARD dxvionique; et organis6 dans le cadre du prcgramme des Consultants ct des Echanges.

List of Authors/Speakers
Lecture Series Director: Dr Wolfgang Keydel DLR Institut f i r Hochfrequenztechnik 8 0 3 1 Oberpfaffcnhofen Germany

AUTHORS/SPEAKERS
Dr John C.Curlander
Jet Propulsion Laboratory 4800 Oak Grove Drive Pasadena, CA 9 1109 United States D r Anthony Freeman Jet Propulsion Laboratory 4800 Oak Grove Drive Pasadena, CA 91109 United States MI Jean-Philippe Hardange Thomson-CSF 178, Boulevard Gabriel P h i 92242 Malakoff Cedex France MI David Hounam DLR Institut fur Hochfrequenztechnik 8031 Oberpfaffenhofen Germany D r Herwig Ottl DLR Institut fiir Hochfrequenztechnik 8031 Obcrpfaffenhofen Germay D r R.Keith Raney RADARSAT Proiect Office Canadian Space Agency 110 O C o n n o r St, Suite 200 Ottawa,OntarioKlA 1 A l Canada

Contents
Page Abstract Abrege List of Authors/Speakers iii
iv
V

Reference Basic Principles of SAR

by W.Kcydel

1 2

SAR Peculiarities, Ambiguities and Constraints

by W.Keydel

Motion Errors and Compensation Possibilities

by D.Hounam

3
4 5

The Real Aperture Antenna of SAR, A Key Element for Performance

by H.Ottl

Polarization Effects and Multipolarization SAR

by A. Freeman

Radiometric Calibration of SAR Systems

by H.6ttl

SAR Simulation
by D.Hounam

7
8

Multi-Frequency Multi-Polarization Processing for Spaceborne SAR

by J C C u r l a n d e r and C.Y.Chang

Inverse Synthetic Aperture Radar

J:P. Hardange

9
10
11

Special SAR Techniques and Applications

by R.K. Raney

Review of Spaceborne and Airborne SAR Systems

by R.K.Raney

Bibliography

vi

BASIC PRINCIPLES OF SAR

hv
~~

W. Keydel Deutsche Forschungsanstalt fiir Luft- und Raumfahrt e.V. Institut fiir Hochfrequenztechnik 8031 Oberpfaffenhofen G e & "

SUMMARY The basic principles of SAR will be explained. Equations f o r geometric and radiometric resolution and their inter-relations will be given in addition to a range equation. The difference between focussed and unfocussed SAR and the conception of beam sharpening will be explained.
1. RESOLUTION, KEY TO REMOTE SENSING

overcome by the here considered Synthetic Aperture Radar ( S A W techniques.

2. The SYNTHESIS OF AN APERTURE [ l to 51

Radar techniques are principally one dimensional measurements. For image-construction the second dimension has to be added. This is done principally bv movino the radar olatform of side lookina svstems in aircrafts 01 satellites. The geometric range resolution of such systems is dependent on the bandwidth which estimates the shortest processed pulselength. The azimuth resolution is dependependent on the antenna aperture 1 which determines the half power beamwidth of the radar beam.
~1
~~ ~~~ ~ ~~~~

_ _

For remote sensing purposes the resolution of

the respective sensor is one of the main factors. Resolution In the wide sense is defined as the degree to which a sensor can discriminate two closely spaced targets, having similar properties (geometry, colour, velocity, etc.) However, angular resolution is a matter of dominant concern. It is defined as the minimum angular separation between two items which can be distinguished by a system. Note that for all systems using elecromagnetic waves, the laws of optics apply. The angular resolution of an optical system is principally limited by both the aperture diameter and the turbulence of the wave propagation medium 1i.e. the earth's atmosphere). Quantitatively the angular resolution of an aperture of a given size 1 is found by the ratio of wavelength L over this size.

The basic idea of the synthetic aperture radar (SARI is the construction after data collection of a very long antenna along the flight path (here assumed as strongly linear) by means of data processing. Along the flight path are the measuring points for amplitude, phase and frequency of the backscattered signal. In this way the real antenna acts as individual elements of the large (synthetic) array antenna. The stored echos are combined through data processing, and the SAR image is produced. The Fig. 1 and 2 show the geometry and the respective terminology of SAR.
P . .I..& "0ll.O"l.l B..n"ldlhll.lnull P*, .5"",h.I* n~ke",., 8..nribl* P. * ,I " ., B.."Md,h ,Il.*.llonl
,AI,">

r = h -

(1)

The human eye, for example, is an optical system, the resolution r ofwhich is limited by the pupil diameter. under mean illumination mm the puwith a wavelength of about 5 x pil diameter varies between 1 mm and 8 m and a n angular resolution power between about 2 arc m i n u t e s and 1 0 arc seconds resu1t.s. The mean resolution power of 1 a r c minute corresponds to a ground resolution of 3 m for a distance of 10 km. (This value can be experienced (approximately!) by looking down to the ground from an aircraft). For optics with diameters roughly a factor 100 larger and at altitude of about 1 0 0 km, a resolution power in the order of 10 cm results. This corresponds excellently to values known or guessed for military satellites which are now operational.
~~~~~ ~~~ ~ ~~~ ~ ~~ ~~ ~~

Diffraction limiting laws of optics apply to microwave remote sensing systems, too. Here, the antenna is the lens of the System and the antenna aperture, the diameter or length determines the aperture size. The antenna diameter 1 determines the so-called antenna pattern. The halfpower beamwidth of this pattern determines of conventional radars the angular resolution following in good approximation equation (11, In order to obtain the same resolution as the human eye in the visible region tremendous antenna diameters are required. A wavelength of 5 cm for instance (corresponding to 5 GHz) would require an antenna diameter of 175 m. This illustrates the principal disadvantage of conventional microwave systems in crbital application. However, these disadvantage can be

R"

+ . . ,
R."-

R.-Eround
T. I<.0

mns.

-n_

Fig. 1 Geometry and terminology for SAR in 3-dimensional representation.

Due t o t h~. e inherent velocitv u Of the radar, t h e .~~ .~~~ .~~
~~ ~ ~~~ ~ ~

j~

~~~~~

frequency of the received signal is Doppler shifted against the transmitted frequency. F o r a target seen under small angles p against the crosstrack direction (less about 30) these Doppler frequency is:

T i s simply the time necessary to shift the This , Doppler frequency from the absolute highest value to the absolute lowest value within the bandwidth of the filter. From (31 and ( 4 ) results f o r the change of f, with time

For bfD = Bmi, and AT = , , , T ( 7 ) by multiplication

results froni ( 6 1 and

1-

- -Swath Width- -

With Af,

Bmin results from (5)

In connection with ( 9 ) results herefrom:

This is the optimum resolution obtainable by use Of a fixed filter for a certain distance R (Slant range). Remarkable is the angular resolution which is dependent an the square root of R and not R as is the case for a normal SLA8. The resolution is independent of the antenna dimensions. Principally, for image construction it is necessary to use for each range bin an other bandwidth (Fig. 1, 2). The use of a filterbank following these equations within the swath from near range to far range (R, s R 5 Rf) allows a relative simple real time processing. If a tracking filter will be used than a so-called matched transmit receive system results. Here, the total time of target illumination can be used: this is the total data acquisition time of the system T , . During this time the aircraft flies a length Lp, which is the largest obtainable synthetic aperture. This time is also called dwell time,
L p = " . T
U

Fig. 2 Twodimensional scheme of SAR

If a filter is used with the bandwidth Af, at the output of this filter a singal will be ehowing up which is obtained from a small angle range A p with the bandwidth AiD. TO this A P belongs an azimuthal distance A X which is the azimuth resolution raf f o r this simple case. In formulas written holds:

Bp

Together with (3) follows

By use of the well known general valid relations between real aperture length 1 , half power beamwidth Phr, slant range and wavelength i (?ig. 1, 2 ) , (L, = Phr ' R , Bhr = i l l ) follows: This is the first and simplest method to built a synthetic aperture by means of a filter with the bandI rap = width Afo, it is called Doppler beam sharpening. It (13) 2 ' is remarkable, that (4) is independent on the real antenna aperture. This is the remarkable formula for the oprimum theoretical reolution of a so-called focmsed SAR. Following (11 to this resolution corresponds a synRemarkable, that it becomes as better as :smaller thetic antenna aperture L ' the real antenna size 1 is. This is opposite to fD ' real aperture radars and most optical systems. The remarkable fact is that for SAR the theoretical limit of azimuth ground resolution is given by the half antenna length in the fl.ight direction of the In order to optimize this method, the Smallest posradar. sible bandwidth Bmin has to be estimated. kin depends on the maximum observation time ,,T availabThe independence from range and wavelength is a le. further remarkable attribute of SAR resolution. A look on Fig, 1 and 2 shows: For each distance, each

1-3

range bin, holds an other synthetic aperture and these synthetic aperture increases with increasing distance as well as the geometric resolution of any optical aperture decreases with increasing distance. The range independence of the resolution is a reason f o r the possibility to extend SAR results gained with airborne systems to spaceborne systems. As allready mentioned the construction of the synthetic aperture for a SAR can be considered as the artificial construction of an array by means of computer techniques. The signals will be stored correctly with respect to amplitude, phase and the appropriate positions. During the image processing procedure the stored signals will be added up correctly and processed to a SAR-image. However, it will be a difference between a real array aperture length and a synthesized aperture of the-same length. Whilst for a conventional arrav the one way beamwidth p h estimates the 1 thetic array due to the coheiciiL to be taken i ift is factor 1 / 2 must be included in the formula (1) for the of way pattern. The time during which a point target is illuminated , (11). by the radar beam is called the dwelltime T The required dwelltime for a specified azimuth resolution is:

tude signals must be stored as well as in the unfocussed case, before the addition Of the various signals a phase correction equalizing the phase difference R , R has to be made. Fig. shows the impulse response for a point target for different. 150a), The signal degradaphase errOrS ( 3 0 0 , tion with increasing error evidently can be seen, The consideration of SAR as a synthesized array leads to the same equations and results with respect to resolution, synthetic aperture length etc. as the Doppler consideration does. However, the meaning of focussed and unfocussed becomes more understandable in the case of the synthetic aperture consideration. Also the name SAR becomes here more evident.

3. THE PHASE INFORMATION, KEY TO SAR 11 to 51

3.1 THE PHASE REFERENCE

u = flight velocity and t = observation time (the , is the maximum observation time is the dwelltime), R shortest distance between platform and target. With R t R , = 2R, follows:
R.R o .L.K?...e - 2 R, (21)

h R %=r,.u.

(141

, can be provided with an An illumination time T oneway antenna beamwidth

To this oneway range difference corresponds a phaseshift R - R, u2 t2 = 2% - = II (22)

h h R,

ha=

(15)

The two way phase change will be twice:

q = 2 n - b.R,
2

t2

There is principally a need for some phase compensation during the whole dwelltime. The distances R , of the far end of the synthetic aperture to a point target are larger as the distance of the middle R,. (Fig. 1). A so-called unfocussed SAR ignores these differences. This case limits the observation angle to the area, where the differences in distance are smaller than h l 1 6 . which is identical with n l 4 radian in phase /Re - R,/
5

(231

This equation is very important as a reference -function for digital SAR processing. The phase shift is a ouadratic function of time. This is t h e so-cailgd phase~historyof a point target, the phaseshift versus time is a parabola.
~~ ~

16

116)

TO the quadratic phase function of time belongs a linear frequency shift, The Doppler shift f , of the signal due to the inherent constant platform motion is given by

If this focus condition is fullfilled the occuring degradations and the reduction of the result is not very large in comparison with the ideal case. However, the dwelltime is not as large as it could be and the resolution is degraded as well as the maximum aperture length. From fig. 1 results with R, t R, 2 R , :
F

For a transmitted signal Vt ceived sional has the form

V ,
=

= VI

sin(2n ft) the re-

V2 sin(2n ftt 2 n f ,
2
u2

($)

e- e

2R,(Re

- R,)

. t) ,
(25)

(171

V , =

The limitation (16) leads to the maximum aperture length of a unfocussed SAR:

t V2 sin[2n(f t It1 . 1R ,

This aperture is responsible for the resolution obtainable and leads to

raun =

This is strongly equivalent to a linear frequency modulation. The received signal of a SAR is linearly frequency modulated. This modulation will be considered a s a code which designes all points with respect to their azimuth angle during flight time. The steepness of the so-called chirp is
1 R , . The bandwidth required for this linear modulation , : results from the dwell time T
2 v2 -

d$

R ; I

(191

In the focussed case the incoming phase and ampli-

1-4
2

B
f ,

= Z L T , .

Ro

126)

3 . 2 DEPTH OF FOCUS

Under the assumution that a maximum uhase error of 118 ma" be allohed at the end of the'auerture the so-calied~"Depth~of Focus" (DOFI can be defined. DOF is the accuracy with which a given quadratic phase reference function must be matched to the considered range, it defines the number of diffe rent reference functions required over the whole swath considered. It holds per definitionem:
~~ ~~

.~

~~

DOF

2A R , .

(27)

The maximum allowed one way phase error caused by improper positioning of the phase reference function can be written as IAR - A R , , ! = 1 1 8 . From (20) follows :

~ A R - AR,~ = A R ~ (-12 8 R ,

1 L

1 8 '
Fig. 4 Example for focus influence of an image quali ty. The lower image is unfocussed (errors result from velocity) the upper imape is focussed.
4.

With (12) results 8 r: DOF = . L The depth of focus becomes smaller as I . is made smaller. This is a very important fact for image processing. For example, if L = 5 cm, ra = 3 m the DOE is about 1.44 km and, therefore, the processing of a 4.3 km swath requires 3 different reference functions. For a resolution of 50 cm under the same conditions the DOF is 40 m and 10 references are reauired. This increases the urocessina comulexitv. ~. Fig. 3 and Fig. 4 show the iniluence of focussinq effects on SAR images exempiarically.
1 .e

HIGH RESOLUTION IMAGING BY USING THE REFERENCE FUNCTION

PHASE

The phase of the received signal 125) contents the required information an an observed target. These information can be extracted by a quadratic demodulator which eliminates the terms with the carrier frequency and leaves the socalled '"inphase component I" and the "quadrature component a" and a joint amplitude factor A containing amplitude or signal to noise informations on the target.

(30)

1.1

Therefore, the complex signal relevant for the information on the target is (with A = 1) :

1.2

S,

Ilt)tjQ(t)

"2t explj 2n(-)tl

LR,

(31)

g
2

0.8

0.8

0.4

0.2

Herein, the target is positioned at t = 0 (where f , = O!) on the variable t, the dwelltime runs from T , l 2 to tTD/2. The further information can be obtained by correlation with a known reference function sr for a point target. The appropriate reference function f o r a point target at this location is the same as ( 3 1 1 . The outuut of the co:rrelator is: tTl2
So =

Fig. 3 Point target impulse response degradation for unfacussed prozessing with different phase deviations ( 3 0 " , 90", 180').

S,(t).s:lttt,) -TI2

dt ,

(321

1-5

T.

The sinc function in (33) delivers the image of the point target. Remarkable are the sidelobes. The correlator output has a maximum for t , = 0, when the reference function and the target aligned with each other. Fig. 5 shows the unprocessed amplitude of a point target as well as it gives the processed image of a point target in azimuth direction.

zes the vector character and the signal delay time The received radar signal is of the kind of N = 1 ti J(wi ti + qi) (351

i=l Here is N the total number of scatterers contributing to the signal, Eithe vector-field scattered by the ith scatterer, E, the totally received field, wi = 2 nfi the circular frequency and ti = t t r i the observation time with the respective delay time ti. The radar receiver in connection with the data processing part has the task to transform this complex signal into an observable signal proportional to lgr12 and to extract the information on distance, velocity, behaviour, shape, quality and other uantitative aspects of the observed target. Ic,T2 is equivalent to the observed power. A very important part plays the term "coherency" which will be used for techniques, instruments and for scattering mechanisms as well. Coherency means the phase stability and the statistical behaviour of the phase of electromagnetic waves Over the observation time. F o r complete incoherent scattering this means if there are many independent Scatterers within the beam and for a pure statistical phase distribution between the scattered signals - the performance of radar can be described with good validations by the radar equation. 6.1 THE RADAR EQUATION FOR SAR The radar equation is given by Hall fo1 a point target (which consists per definitionem of one single scatterer) 111 :

time

---->

Fig. 5 Amplitude of a pulse response (upper curve) from a point target and SAR-prozessed impuls response of a point target (lower curve) following (33) in comparison with point target response for'a real aperture radar.

5. MODULATION PRINCIPLES FOR RANGING

The main considerations up to now went into direction of the azimuth resolution capability of SAR, which is obviously the main advantage: The azimuth resolution of SAR can be equivalent to the resolution of optical systems, despite of the fact that small antenna apertures will be used. However, the range resolution should have the same order in order to produce high quality imagery. The range resolution ry of a pulsed radar is equal to half of the pulslength zP.

Herein is: , P = average transmitted power, G = antenna gain, h = wavelength, 0 = radar cross-section, R = distance radar-target, k = Boltzmann constant, To = noise reference temperature usually 290 K, F = receiver noise figure, L,,, all losses. For the dwelltime T , Results from (11) and (12):

ry=p.

(34) h.R TD -- z -r , " '

137)

High resolution would require extremely short pulsi, and this would require extremely small bandwidth. However, this would entail considerable technological problems and would rapidly lead to the border of SAR possibilities. A solution is here to use pulse compression-methods [ l ]. There are mainly two methods: the commonly used frequency chirp modulation and the digital pulse code modulation with Barker or Pseudo-Noise codes respectively. Both methods require large time bandwidth products and a r e well developed for RAR with ranging tasks. However, it should be mentioned here, that the principles of S A R can be considered without special ranging procedures. But looking at the results of the previous Doppler shift considerations during the azimuth measurements a SAR uses principally a linear frequency modulation for azimuth coding, which is formally iclentical with the frequency modulation principle for ranging. For both, ranging and azimuth measuring the same receiver unit (and the same processor) can be used and this is the main reason why always the FM-chirp for ranging will be considered i r r connection with SAR in the literature.
6. POWER CONSIDERATIONS AND FADING STASTISTICS

This leads to the SAR equation for point targets: P a , , G2 h3 o SIN = l 4 n ) " ' (38) R (kT,F) .2u.r,.LtOe
For area targets the radar cross section per unit

area eo has to be taken into account:

Remarkable is here the dependence of the cube of the range even for point targets. This is a significant difference to real aperture radar. This equation holds if all scatterers within the resolution cell ra'ry are independent that means for incoherent summation of the hackscattered Dower. Fio. 6 shows as an example the image of an oiispill on-the seasurface taken with a SAR in C-Band.
~~~ ~

Principally, radar uses information on electromgnetic Waves characterized with amplitude E, phase q , frequency f, the polarization which characteri-

Fig. 1 shows the signal to noise ratio as a function of range (71. The dependence of R3 and of the antenna diagram becomes evident as well as the socalled amplitude line-echo (ALE) which results from

1-6

F i g . 8 S i g n a l i n t e n s i t y curve of t h e imaqe i n Fig. 6 a f t e r radiometric c o r r e c t i o n u s i n g t h e c u w e i n F i g . l b . (660 m f l i g h t a t t i t u d e , 30 deg depression a n g l e . ) F i g . 6 SAR image of an o i l s p i l l on t h e sea surf a c e taken with DLR E-SAR, C-band [ 7 1 . but many s c a t t e r i n g c e n t e r s which c o n t r i b u t e t o t h e radar response s i g n a l ; i n d i v i d u a l s c a t t e x i n g cent e r s may be seen with very high r e s o l u t i o n systems lexeptians are very simple t a r g e t s only 1.ike spheres, corners, z y l i n d e r s , p l a t e s ) . Fig. 9 shows as an examole t h e d i s t r i b u t i o n of such c e n t e r s on a c a r I8 I .

-so

1-

zm
nMD.

=a
ml.1L.r.l

. c a

i
300

->

Fig. l a Signal t o n o i s e r a t i o over t h e whole scene i r F i g . 6 a s a f u n c t i o n of range. ( 6 6 0 m f l i g h t a l t i t u d e , 30 deg depression angle I l l . ) The A l t i t u d e l i n e echo (ALE) c l e a r l y can be ident i f ied.

11

F i g . 9 D i s t r i b u t i o n of r e f l e c t i o n c e n t e r s on a c a r i n a twodimensional r e p r e s e n t a t i o n . Resolut i o n i s about 15 cm Y 15 cm (rreasurement Graf DLR [ Z ] ) .

Under t h e assumption of s t a t i s t i c a l d i s t r . i b u t e d amplitudes E, and phases 0 , (35) becomes ,i s t a t i s t i c a l phasor t h e q u a l i t y of which i s dependent on t h e r e s p e c t i v e d i s t r i b u t i o n f u n c t i o n s . Fo.7 uniformly d i s t r i b u t e d phases q i and Rayleigh d i s r i b u t e d amp l i t u d e s r e s u l t s ap. exponential d i s t r i b u t i o n f o r t h e amplitude square of ( 3 5 ) . That means i n g e n e r a l i f homogeneous a r e a s of t h e e a r t h a r e i l l u m i n a t e d 0 l bv a coherent radar t h e backscat-tered siorial amoli0 100 ZOO 300 a0 BOO 000 tudes of t h e s i n g l e observations i p i x e l s ) a r e s t a aanp. 6111.t.lli -> t i S t i . c a l l y d i s t r i b u t e d and t h i s i s t h e reason f o r F i g . 7b Signal i n t e n s i t y curve of F i g . l a a s a funct h e speckle t y p i c a l f o r each r a d a r image a l s o f o r t i o n of range i n t e g r a t e d over 50 s . I 6 6 0 m images of very homogeneous a r e a s . There is no posf l i g h t a l t i t u d e , 30 deg depression a n g l e . ) s i b i l i t y t o c h a r a c t e r i z e a radar t a r g e t hy measurement of one p i x e l only. P r i n c i p a l l y t h i s speckle r e f l e c t i o n s d i r e c t l y underneath t h e a i r c r a f t . T h i s can be reduced by image averaging w i t h t h e use of c u r v e can be used f o r image c o r r e c t i o n . m u l t i looks f o r i n s t a n c e , t h i s reduces t h e s t a n d a r d d e v i a t i o n p r o p o r t i o n a l t o t h e r o o t of t h e look num? i g . 8 g i v e s t h e s i g n a l i n t e n s i t y curve of Fig. 7 a f b e r and smoothes t h e s p e c k l e . Speckle can be seen t e r c o r r e c t i a r . The image i n F i g . 6 i s c o r r e c t e d ( ! ) i n a l l t h e radar images presented h e r e . by using t h e r e s u l t s o f Fig. 7 .
~~~~~ ~ ~ ~ ~~~~ ~

~j

6.2

-STATISTICS

The P.ayleigh d i s t r i b u t i o n f o r the envelope Ti, longing t o ( 3 5 ) i s

be-

However, equation 135) shows t h e s t a t i s t i c a l n a t u r e of t h e radar response s i g n a l due t o t h e t a r g e t and s c a t t e r s t a t i s t i c s . Normally complexe radar t a r g e t s l i k e c a r s , t r e e s and a r e a s e t c . have not o n l y one

The phase is uniformly distributed

P(Q) =
1

for

0 < 9 < 2n

Mean value are:

v,

140h) of 140a)

and the second moment : V

The variance is by definition:

The variance represents the fluctuation of the registered voltage around the mean value, this is principally an inherent noise and it can be written as a mean squared AC component %. Together with (41) one obtaines the inherent signal to noise ratio S,, for Rayleiqh fading

s,,

= -=

Fig. 10 Histogram of the filtered SAR image of Fig. 6. (Oil slick A2, 660 m flight altitude.) Both curves follow a Rayleiqh distribution.

3.66 2 5.6 dB .

This occurs, even in the absence of any other additional noise. That means, the Rayleigh fading is equivalent to addition of noise that results in a signal-to-noise ratio of 5.6 dB, even if the ratio of the average signal level to actual noise is very high. Thus, the best equivalent signal-to-noise ratio that can be obtained when Rayleigh fading exists is 5.6 dB. This is the reason that fading distributions are so important in radar problems. One often would like to have a high signal-to-noise ratio, but the effective signal-to-noise ratio cannot exceed 5.6 dB unless multiple independent fading samples are added together, regardless of the ratio of the mean received power to the thermalnoise power in the system [51. The measured signal P, which is considered in (361 is the power, which can be written as
P , =

vL

~N

Vri.

i=l For the mean holds principally

v,

v,

C"

But for the variance one obtaines

E
= = ,

0.429

c :

With these results for the optimum equivalent S , , results instead of 141) :
S ,
=

3.66 . N .

144)

: v

The maximum obtainable signal to noise ratio increases rapidly with increasing N. The time T during which a point target remains within the real beam is the time for observing this fading signal, This time can be obtained from the consideration of the extension of the footprint in flight direction which is the azimuth groundresolution ra of the real antenna.
ra
= 2;2

From distribution (40) for V, results the exponential distribution for P , :

1
P ,

P(P,)

'= - expl-

20;

-1 20s

for

0 < P ,

. (42)

For the mean value

i,

and the second moment $ holds

= 80:

P, =

20;

Phr .

= VT

1451

.
The number N,, of independent samples in azimuth direction is the product of bandwidth and ohservation time.
N , ,
= Afo
'

With these results the Rayleigh discibution 140) can be rewritten:

,
(46)

Fig. 10 shows as an example the histogrames of Fig. 7 171. If N independent pixel values are added than foll o w s from the laws of probability theory that the mean value of the sum is identical with the mean value of the single element whereas the variance for the sum of N independent samples is the variance of the single sample divided by N. That means for the case of a linear detector if N numbers of samples are averaged and if Vri is the envelope voltage of the ith sample than is the average received voltage VL:

N , ,

b .

Pir R

This leads to the simple equation

The equations 1461 and ( 4 7 ) are very fundamental. They apply to SLAR and to SAR as weIl. Equation 147) shows the possible tradeoff which can be made between large resolution and small ra and high specle reduction (large N ) . The number Ny of independent samples in range d i rection results from considerations of range resolution. The slant range resolution for a pulse

1-8

radar ry is given by:

r y

c ' Tp = 2 _ - -C

28

'

If N y returns with thi resr than a poorer resolution wi: the number of independent ~i nal to the reduced range re!

n will be averaged ult. Therefore, xels is proportia2n

The equivalent pulselength rpeq associated with averaged returns is:

rpeq
= Ny

Ny

<p

.
' ReOolution : 11.4 m

Using these results, one can express the number of independent range samples: (48)
In ( 4 8 ) is Be, the equivalent bandwidth which be longs to rpeq.

The total number of independent samples is given by:

N =

N ,

'

Ny

1491

Equation ( 4 8 ) shows as well as equation ( 4 7 ) in connection with (45) the possible tradeoff between high geometrical resolution and high speckle reduction which corresponds to high radiometric resolution. Radiometric resolution describes the possibility to consider two areas with two different speckles lor medium grey tones) as different. one possibility is to take the mean values X, and X2 of the probability density functions of the pixels of those two fields and use the ratio of both mean values for a measure of the separability:

Fig. 11 Impulse responses Tlms l(Zdimensional o O ims) D SODO image) of a

I
' PSLR :.28.1 tlB ' ISLR : 13.9 d B

point target showing the interconiection of resolution and siderlobe level.

X2
rrad =

TI
XI

150a)
AX

91 t
rrad = XI

7 .

150b)

This ratio is called radiomezric resolution. For SAR the angular resolution is closely connected to the radiometric resolution. The radiometric resolution is the minimum brightness contrast necessary for the discrimination of two targets. The radiorretric resolution in radar images is dependent on the image statistics, the speckle. Therefore, by increasing the integration time 1i.e. the observation time of a certain area represented through the number of l o o k s on it) the speckle will decrease, the radiometric resolution increases (the image becomes sharper), but the angular resolution decreases. Fig. 11 shows the processed radar ansver of a point target with different geometric resolution (8.94 m and 11.4 m). The decrease of the sidelobes with increasing resolution is evident. A quality measure is the peaksidelobe ratio PSLR as well as the integrated sidelobe ratio ISLR which are defined as follows: PSLR
=

ower within the sidelobes power within the main beam

'
'

ISLR =

power within the sidelobes power within the total diagram

Fig. 12 SAR image of DLR-Research Center and airfield in Oberpfaffenhofen taken at 5 cm wavelengths from an aircraft at 914 m altjtude. Scene dimension 2700 m x 3510 m, resolution 2 m x 3 m [azimuth x range), 8 looks.

1-9

In Fig. 12, 13 and 14 this tradeoff between geometric and radiometric resolution as well as the influence of the speckle is illustrated. The images from the DLR-Dornier airport at Oberpfaffenhofen have been taken by the airborne SAR of the DLR at 5 cm wavelength 19, 10, 111. Fig. 12 shows an 8 look image with low speckle and high contrasts. This image with its dimensions of 2.700 m Y 3.510 m (range Y azimuth) has a resolution of 2 m in range and 3 m in azimuth: with its many details it is comparable to a r e a l photography. Fig. 13 shows enlarged a part of Fig. 12: the scene dimension is 828 m in range an 1.079 m in azimuth. This one look image has an azimuth resolution of 0.5 m and a range resolution of 2 m. (The resolution in range here can not be increased due to the limited bandwidth of the system.) The increase of the speckle against the 8 look image in Fig. 12 is evident. Impact of high resolution becomes clear by looking at the large black platform in the upper left corner. There an aircraft clearly can be seen. This aircraft together with its background is shown enlarged in Fig. 14 with the Same resolution. The image size corresponds here to 60 m x 60 m. The differences in range and azimuth resolution are evident. This same image was used for estimation of the dimensions of the aircraft. A comparison with the known dimensions of a DO 228 aircraft leads to the conclusion, that the image shows with a high degree of probability a DO 228. The estimated dimensions l w i t h the real values of a DO 228 in paranthesis1 are: Total length 14.3 m (15.041, totHl wingspread Fig, 13 SAR image, part of the airport in Oberpfaffen- 16.3 m (16.97 m), wingspread of the elevator unit 7 . 1 m (6.45 m). This points out, that for airborne hofen. The scene corresponds to a section of the image in Fig. 12. Scene dimension is 826 m SAR a resolution in the order of dm is the present x 1019 m, wound resolution 0.5 m x 2 m (ari- state of the art [12, 131 However, for spaceborne SAR the present state of the art is poorer due to muth x range), 1 look, The aircraft on the different reasons (atmospheric turbulences, powerplatform in the middle will be enlarged relimitations, data-rates). Present spaceborne SAR oresented in Fig. 14 (Moreira. DLR) . principally can have a ground resolution power in the order of one or a ew meters
~

7. SAR IMAGES 7.1 IMAGE GEOMETRY SAR-pictures are different from normal photographs in many details. They look like an aerial photograph although taken from a sidelooking perspective. This side looking perspective leads to shadoweffects and to the well-known characteristic plasticity of SAR-pictures. The shadows can be used for heiaht estimation of trees, rocks, and other shadow producing elements in a SAR image as in Fig 22 for example.

RADAR-IM4GE PLANE

RADAR-IM4GE

Fig. 14 Enlargement of the aircraft on the dark platform in the upper left corner of Fig.13. Scene NEAR RANGE FAR RANGE dimension is 60 m x 60 m, ground resolution is o b c 0.5 m Y 2 m (azimuth x range), The unsymetry in the resolution pixels caused by the unsymLEVEL SURFACE metrical ground resolution area clearly can be seen. The aircraft is probably a Do 228; this is a conclusion of a geometrical evaluation of Fig. 15 Radar-image format showing slant-range prethe aircraft image in comparison with the well sentation a l b, cI relative to ground-range known real dimensions. The different reflecabc, assuming a level surface long-dashed tion centers typical for radar imaging a s well line represents radar-image plane. Look angas the Speckle clearly can be seen (Moreira, le is inclination of the radar beam off verDLR) . tical [141.
3 '

1-10

An incorrected SAR inage is allways a so called slant range image it represents basically the distance from the radar to each of the respective Surface elerr.ents in the scene. Therefore, in its raw s t w e it shows some oeometric distortion due to the differences between slatrange and groundrange explained in Fig. 15 (141. Fig. 16, 17 and 18 show raw data a slant range image as well as a finally processed image of the same scene
~~

Fig. 18 Azimuth processed image - only the totally focussed region. Processed from saw data in Fig. 16 and slant range image in Fig. 17.
Foreshortening [14]

Slopes inclined toward the radar appear compressed relative to slopes inclined away from the radar. The effect is illustrated in Fig. 19. The foreshortening factor F is approximately:
F = sin(*-a)

Fig. 16 Raw data: RCM corrected, range compressed image before the azimuth processing.

where the look angle B is the angle between the vertical plane and a line that links the imagingradar antenna to a feature on the ground, and is the slope angle of the surface. Alpha is positive ( a t ) where the slope is inclined toward the radar (foreslope), and negative ( E - ) where the slope is inclined away from it (backslopel

SLOPING SURFACES

Fig. 17 Slant range image processed from raw data in Fig. 16. Azimuth processed image including the partially processed region. A further set of geometric relief dispalcements that is characteristic of all imaging radars is the apparent variation in length of equal terrain slopes when the respective slope lengths are imaged at different incidence angles. The displacements result in foreshortening, layover, and shadowing. These effects are feature-dependent and cannot be easily corrected.

Fig. 19 Radar foreshortening of slope ab, which is projected as al bl, relative to s:lope bc, which is projected as bl cl, show:.ng look angle 61, foreslope angle a+, backslope angle 3 . Long-dashed line represents radar image plane.
Layover [14]

Layover is an extreme case of foreshortening that occurs when the l o o k angle B is smaller than the foreslope a+ (8 < E + ) . This is illustrated in Fig 20. In this case, a mountain is laid over on its side.

1-11
..... ,.,
SAR RADAR BEAM RADAR-IMAGE PLANE

,.L-li

... . i . < .

: i.

....

, I

".L\..l/rl

....

i <

.. . . .. .
' j

RADAR-I MACE

SLOPING SURFACES

Fig. 20 Radar layover of slope ab projected as bl al on radar image, where look angle 0 is Smallel than slope angle a+. Long-dashed line represents radar-image plane 1 1 4 1 .
RADAR BEAM rRADAR-IMAGE PLANE

Fig. 2 1 Radar shadow of surface bcd projected as b, dl on radar image. Long-dashed line represents radar-image plane 1 1 4 1 . Shadowing I 1 4 1 Shadowing is essentially the opposite of foreshortening. Slopes inclined away from the radar are in shadow when the look angle 19 plus the backslope anole are than 90' r ~ a- . . iireater .i.d - a ' ) > d 2 1 . Sha,~ dows are caused by ground features that obstruct the radar beam and prevent illumination of the area behind them. This effect occurs on Seasat SAR images whenever the backslope in the radar viewing direction exceeds about I O " . It is shown diagrammatically in Fig. 21. Fig. 2 2 shows clearly the shadows of different features
~ ~~ ~ ~ ~~

Fig. 2 2 Iller river and a channel taken by DLR-SAR (C-band) from 1 8 2 0 m altitude. Resolution 2 m x 2 m, scene dimension: 7 6 8 m Y 7 6 8 m. The altitude of the trees beside the road in the upper left corner can be estimated from knowledge about the flight geometry and the shadow length's. For the first 4 trees from right to left just left beside the left bridge result: shadow length's between 35 m and 40 m and tree altitudes between 2 0 m and 2 4 m (image: Moreira, DLR) [ 1 2 1 .

8. BASIC HARDWARE CONSIDERATIONS

8.1 PRINCIPAL CONSIDERATIONS

A simplified block diagram for a SAR is shown in Fig. 2 3 . The SAR is mounted on a platform moving at a COnstant velocity. The PRF must be sufficiently high to avoid azimuth ambiguities. This criterion requires that the radar platform displacement cannot exceed one-half the antenna size between successive transmit pulses. In a SAR, phase stability is exceedingly important. The prime oscillator which provides the signal for the transmitter as well as the reference for the receiver must be very stable. The timing of the transmit pulses must be very precise with respect to the prime oscillator. If the radar platform velocity is not constant, the deviations must be measured and this data used to compensate either t h e incoming signal or passed on to the signal processor a s a correction. Fig. 2 3 Block diagram of a SAR (schematically).
A typical SAR transmitter is designed to overcome

limitations of peak power in components and to satisfy stringent azimuth and range resolution requirements. High resolution in azimuth requires stringent phase stabilities from pulse-to-pulse and over the integration time. High range resolution requires wide RF bandwidths. To meet the signal-to-noise ratio and target detection requirements correlation principles and pulse code modulations can be used respectively with special wave forms. Electronic circuits using a voltage controlled oscillator can provide the desired transmit pulse. The output fr@m the high power amplifier passes through a circulator and is radiated by the radar antenna. The received signal passes through the same circulator, is amplified (and normally pulse

1-12

compressed). (the pulse cor.pression circuit can use a freqilency dispersive delay line which convert!; a wideband linearly swept FM iong pulse into a short pulse signal with same bandwidth.) The phase of the short pulse is measured by the phase coherent d e tector, and the resultant signals are delivered to the signal processor. 8.2 THE DLR E-SRR AS AN RERLISATION EXAMPLE The DLR airborne experimental Synthetic aperture radar System E-SRR, designed and manufactured at the DLR Institut fuer Hochfrequenztechnik, is a research tool to elabozate SRR related Pro& lems concerning both system performance and data analysis. The instrument is installed on board a DLR Dornier DO 228 aircraft, which is a Small STOL aircraft (STOL: short take-of and landing), offering the advantages of low costs and operation from airstrips in any part 31 the world. All SAR results shown here have been obtained with this instrument. RF centre frequency, L-band: C-band: X-band: IF centre frequency: System bandwidth: SAW chirp, signal bandwidth: expanded pulse length: compressed (analogue) p u h e length: Digital chirp, signal bandwidth, Applications narrow swath mode: wide swath mode: The E-SAR is a high-resolution SAR operating in r,-, super wide swath mode: C- and x-band with either horizontal or vertical expanded pulse length: polarization. Although being developed mainly for Antenna gain, L-band: use by the research corrmunity, commercial lease C-band: opportunities are a s well anticipated. The Sensor X-band: is versatile, with many options for flight and E Antenna 3 dB beamwidth, dar configurations and image products. It provides azimuth, L-band: the opportunity to image areas, wether flat or C-band: mountaineous terrain, ocean or ice, with excellent X-band: image quality. It can be used for monitoring resourelevation, L-band: ces, renewable such as agriculture and forestry, 01 C-band: nonrenewable such a s geological resources. Changing X-band: characteristics such as urban growth, deforestation Transmit peak power, L-band: or ocean waves a l s o can be rronitored. C-band: X-band: The system Platform Receiver noi5e figure, L-band: C-band: A Dornier DO 228 aircraft eqJipped with modern naX-band: vigation systems like a lase1 inertial reference system ( I R S ) and a GPS receiver carries the E-SAR Receiver dynamic range with sensor. Its maximum take-off weight equals 5980 kg. AGCISTC: Nominal pulse repetition freThe maximum operatirg altitude above mean sea level IMSLI is 8000 m. The maximum cruising speed is quency (PRFI : about 440 km/h. For SAR operation the nominal ground Variable PRF ranae: speed of the aircraft is 70 mls, which corresponds Quantization (I or Q) : to 252 kmlh. A/D converter dynamic range (at 35 MHz) : The On-board segment Sampling rate, narrow swath mode: wide swath mode: super wide swath The radar sensor is a modular designed system, which mode: contains three different RF-segments in L-, C- and X-band. Pulse generation and IIQ-detection a r e loca- Echo buffer memory capacity ted in the IF-section. A sinale diaital conversion iI or 0 1: recordina Svstem used to store store the SAR raw da- Nominai'data rate on high density and recording system is used-to ta on high dgnsity density digital tape (HDDTI formatted in in^ t.aoe: the SAR 580 HDDT format. Maximum recording time per tape (14 inch tape reel) : The On-ground Segment Spatial resolution, range and azimuth, narrow swath mode: wide swath mode: The ground segment consists of the following super wide swath mode: units: Number of statistically inde- Radar Raw Data Transkription pendent l o o k s : A SAR 580 High Density Tape Transcription Radiometric resolution ( 8 looks) : system (HTS) transcribes the raw data from Geometric distortion: HDDT to computer compatible EXABYTE (Video 81 tapes. A further data transfer to conTable 1 E-SAR specifications. ventional CCTs is possible. This operation Since the beginning of 1989 the E-SAR system has been flown many times in preparatory campaigns far the European Remote Sensing Satellite ERS-1. The German/Italian X-SAR, which will f l y with SIR-C on three Shuttle Radar Lab missions, and the French Radar 2000, both spaceborne SAR projects, are supported with E-SAR image dats.
~~~ ~ ~~~ ~ ~~~~ ~~

Fia. 24 E-SAR Sensor SvStem blackdiagram (RF-electronicsl 1.3 GHz 5.3 GHz 9.6 GHz 300 MHz 120 MHz 100 MHz 4.98 us
17

ns

90 MHz 50 M3z 18 M . ~ z 5.0 p,; 14 dBi 17 d13i 17.5 cdBi 18 19 17 3s

33

Dt?g Deg Deg oeg Dcig

30 D f i q 500 W 90 W 2500 Vi 8.5 dB 4.0 d E l 4.5 d i l

L

40 d B

952.3E Hz t/-30 %

6
25

bit

dE, 100 MHz 60 MHz

20 MHz
2560 words 28 MBPS 15 min 2.5 m x 2.5 m 4.5 m Y 4.5 m 11.5 m x 11.5 m
8

L~

< 2 dB
les t h m one resalutim cell

1-13
I 6 1 Elachi, Ch.

Spaceborne Radar Remote Sensing Applications and Techniques. IEEE, New York, 1 9 8 7 . Entwurf und Ergebnisse des DLRECht2eit-AzimutDrozessors fur das E-SAR-System. DLR-FB 8 9 - 3 0 .
~~ ~

[ 7 ] Moreira, A.

[81 Graf, G

High Resolutin Imaging of Radar Targets with Microwaves. Conf. P ~ O C . Military Microwaves ' 7 8 , London/Engl., 2 5 . - 2 7 . 1 0 . 7 8 Microwave Exhibitions and Publ. Ltd., pp. 295.302.

Fig. 25 E-SAR Sensor system blockdiagram (digital electronics). also converts the SAR 580 HDDT format into signal CCT format and D r O vides the full G t i Gate in a single channel transcription mode.
t h e SAR . 5RO video . ~

L91 Horn,

R.

C-Band SAR Results Obtained by an Experimental Airborne SAR Sensor. Proc. IGARSS, IEEE, 1989, pp.
2213-2216.

[lo] Moreira, J.

- E-SA8 standard SAR Processing

The E-SAR standard SAR processor consists of thre basic modules, auxiliary data processing, offline motion compensation and focused multilook SAR processing. The development was carried out in DLR. The processor output is calibrated and available in form of a standard image product. Motion Compensation The DO 228 aircraft is fairly sensitive to air turbulence and, therefore, not very well suited for carrying a SAR sensor. A SAR is a coherent system and flight instability causes phase errors, which, in turn, defocus and distort the image geometrically. This problem can be overcome by measuring the dynamic behaviour of the platform and correcting the SAR data, either on board the aircraft or on ground. TWO methods fo1 compensating platform motions are implemented with the E-SAR, one using an inertial Measurement Unit (IMU), the other, the "Reflectivity Displacement Method (RDM)", extracting true forward velocity and Line-of-Sight (LOS) accelerations out of the radar raw data. This guarantees that the E-SAR achieves good image quality with high spatial and radiometric resolution.
9. REFERENCES

Estimating the Residual Error of the Reflectivity.Displacement Method for Aircraft Motion Error Extraction from SAR Raw Data. ICCC Intern. Radar Conf., May 7- 10, Arlington, USA, 1 9 9 0 , pp. 70-75. A New Method of Aircraft Motion ~rror Extraction from Radar Raw Data for Real Time SAR Motion Compensation. 12th Canadian Symposium on Remote Sensing, IGARSS '89, Vancover, Canada, 10- 14 July 1 9 8 9 . Proc. IGARSS, IEEE, 89, pp. 2217-2220. Microwave Sensors for Remote Sensing of Land and Sea Surfaces. Geo Journal 24.1, 1 9 9 1 , pp.
7- 25.

: 1 1 ] Moreira, 3 .

I121 Keydel, W.

(131 Keydel, W.

Verification Using Spaceborne Microwave Imaging. IEEE, Technology and Society Magazine, Dec. 1990lJan. 1 9 9 1 , pp. 5 3 - 6 1 . Seasat Views North America, the Caribean and Western Europa with Imaging Radar. JPL-Publication 80-67, 1 9 8 0 .

[ 1 4 ] Ford, J.P.

[ I ] Skolnik, M.I.

Radar Handbook, McGraw Hill Book Company, 1 9 7 0 .

Blom, 8 . Daily, D. Elachi, X.

121 Hovanessian, S.A. Introduction to Synthetic Array and Imaging Radars. Artech House, InC., 1 9 8 0 .

131 Tomiyasu, K.

Tutorial Review of Synthetic Aperture Radar ( S A R ) with Applications to Imaging of the Ocean Surface. Proc. of the IEEE, Vol. 66, No. 5, May 1970, pp. 563-583. Synthetic aperture radar. The Artech radar library. Artech House, Inc., 1 9 7 6 . Microwave Remote Sensing Active and Passive. Vol. I1 Radar Remote Sensing and Surface Scattering and Emission Theory. Addison-Wesley Publ. Comp. Advanced Book ProgramIWorld Science Devision Reading, Mas.,
1982.

I 4 1 Kowaly, J.J.

(51 Ulaby, F. Moore, R.K. Fung, A.K.

2-1

SAR PECULIARITIES,AMBIGUITIES AND CONSTRAINTS by w. Keydel Deutsche Forschungsanstalt fiir Luft- und Raumfahrt e.V Institut fiir Hochfrequenztechnik 8031 Oberpfaffenhofen Germany ABSTRACT A synthetic aperture radar (SAR) is basically a coherent scatterometer that employs a coherent real aperture radar with highly sophisticated data evaluation and image processing capabilities. Therefore, the coherence of the system is very important; furthermore, the keypoints for SAR are data storage, evaluation and processing. These facts entailpeculiarities of SAR and special ambiguities which are different from those arising with real aperture radar (RAR). The objective of this paper is to point out the specialpeculiaritiesand ambi' . . ' gurties of SAR in comparison to the corresponding properties of RAR. Main topics in this connection are: basic peculiaritieslike range dependency of signal to noise ratio, azimuth resolution and influence of platform velocity. Furthermore, range and azimuth ambiguities, pulse repetition frequency limitations, velocity effects and phase errors influence on SAR-image that cause motion compensation problems. All these effects will be explained together with different contrast-equations between the target and clutter signals of SAR and RAR. These facts entail peculiaritiesof SAR and special ambiguities which are different from those arising with real aperture radar (RAR). The objective of this paper is to point out the specialpeculiarities and ambiguities of ZAR in comparison with similar facts of RAR. Basic peculiaritieslike range dependence of signal to noise ratio, azimuth resolution and influence of platform velocity. Furthermore, range and azimuth ambiguities, pulse repetition frequency limitations, velocity effects and phase errors influence SAP-images and cause motion compensation problems. 2. BASIC SAR-PECULIAPITIESIN COMPARISON WITH REAL APERTURE RADAR
For complete incoherent scattering, this means if there are many independent scatterers within the antenna beam and for a pure statistical phase distribution between the scattered signals the performance of radar can be described with a good validation by the radar equation. The decisive factor is here the signal to noise ratio. For the detection of a point target not only the radar signal itself but also its background, the clutter, and the respective contrast between signal and clutter is most important. However, each imaging radar measures and uses the clutter as primary signal. Therefore, the clutter to noise ratio, CIN, is for imaging radars decisive, whereas for point target estimation the contrast between signal and clutter, S I C as well as the signal to noise contrast SIN is essential.

1. INTRODUCTION
A Synthetic Aperture Radar is essentially a coherent scatterometer or real aperture radar with sophisticated data evaluation and image processing. Important is the coherence of the system. However, storage, evaluation and processing of the data are the key points for SAR. The use of a synthetic aperture in connection with extremely high range resolution methods like pulsecompression requires a high degree of coherency and frequency stability. This holds also if pulse compression techniques for scatterometer will be used. However, the combination of synthetic apertures with pulse compression entails requirements for frequency adjustable oscillators for example and other high sophisticated components and this makes evident: SAR requires extreme effort not only with respect to software but also to hardware.
~

Table 1 s h o w the different equations for area and point-targets and for real and synthetic apertures and the respective contrasts. For contrast computation the losses in all equations have been assumed as identical and.equal. However, this is normally not the case and it shall be remarked that normally special system losses are dependent on the special applied techniques; but for simplicity it shall be allowed here to take all losses out of consideration in order to reach the simple compaArea Target (CIN)reai = Contrast (Smreal=
'

Point Target Real Apertur (FAR) (SIN) real P a , ,

=

G2 X2

P a , ,

G2 A3 oo

ry

( 4 ~ R4(kT,F).B ) ~

f , I ~ . L (471)~ ~ ~ R3 ~ (kT,F) f , LtOt KIN1 SAR

=

0.i a,.R.h.ry.B.r,
(S/CISAR =

Synthetic

(SIN) SAR

Factor

B.rp A R f ,
I " '

2"

Table 1 Radar- and contrast equations for real aperture radars and SAR.

2-2

rison and contrast equations in table 1. ICINI = received clutter to noise ratio, ( S I N ] = received signal to noise raiio, , , P = average transmitter power, G = antenna gain, k = wavelength, R = distance, (i = radar cross section, oo = radar cross section per unit area, k = Boltzmann constant, To = 290 K, B = receiver bandwidth, F = receiver noise figure, f , = pulse repetition frequency, T~ = pulselength, , , L = losses, u = SARvelocity, 1 = real azimuth diameter of SAR antenna, ry = ground range resolution, ra = prozessed azimuth resolution. Table 1 shows for point and area targets as well the peculiarities of SA8 against conventional radar. It shows evidentlv the imoortance of the m -

1. The maximum-azimuth resolution is independent of wavelength and distance.

2. A better azimuth resolution can be reached with smaller real antennas and not with larger antennas as it is the case with RAR and optical systems respectively.
3 . The SIN for a SAR is inversely proportional to the platform velocity. 4. The SIN for point target detection is inversely proportional to the third power of range and inversely proportional to the prozessed azimuth resolution.

5. The SIC for point targets is independent of distance R.

3 . AMBIGUITIES

length and distance becomes evident while for area targets the real aperture length plays an important part. Especially the expressions in the "1mpTo"ement-Factor-Line" and in the "ContrastROW" show factors, which a r e most important for a comparison between RAR and SAR. In order to increase the understanding of table 2. it should be mentioned, that there are basically two SAR-techniques called the focussed and the unfocussed techniques. Table 2 compares the maximum azimuth resolution and maximum synthetic aperture length for imaging radars with real aperture, focussed and unfocussed synthetic aperture for a real azimuth antennalength l.
Real Aperture Radar

Ambiguities play an important part in all radar considerations. For a l l pulse-Doppler radars almbiguities exist due to the periodical structures of the signals. This illuminates the fact, that the pulse repetition frequency PRF, f , , will be the decisive factor. However, ambiguities can also be responsible or the choise of the basic radar frequency. The principal Doppl e r frequency system received with period.ically pulsed Doppler radar for moving configurations i:r shown in Fig. 2 schematically,

,+

AmOlilYde I squm1

max. synthetic aperture length azimuth resolution

Unfocused SAR

1
X.R.11

max. synthetic aperture length azimuth resolution Focussed SAR

G
X.R/l

max. synthetic aperture length

azimuth resolution

112

Table 2 Comparison of RAR, focussed and unfocussed SAR . Fig. 1 represents the maximum azimuth resolution of different radar systems versus height for a 10 m antenna length 1 typically used for spaceborne SAR as it is installed on ERS-1 for instance.
Le/"..ed

Af
f

PRF

Azimut Ambiguity, Pubespectra

IlS

Ynl*L"..Id

I*"

so., hD~,>",*

Fig. 2 Scheme of a pulse spektrum and of azimuth ambiguities caused by PRF. 3.1 BASIC AMBIGUITY CONSIDERATIONS FOR RAR The distance between main spectral lines corresponds strongly to PRF and, therefore, holds for the unambiguous Doppler range

o..o,",,o",md

,$

>

203

,a4

.. :ig. 1 Azimuth-resolution of radar systems versus

The Doppler frequency fD results from the relative velocity ur between radar and radar target f3llowing the , = 2 uric f. The following basic peculiaritiesof a SAR, especidlly relation f of a focussed SAR, in comparison with a conventional This leads to the unambiguous range f o r the velocity real aperture radar WAR) result from table 1: height, r e a l antenna length 10 m, incidence angle: 4 5 a .

2-3

Urunam

c a

fp

(2)

sampling. With bandwidth B = &fD = 2f, follows from (1) the so-called oversampling ratio
fP >
Af ,

The unambiguous range for the distance measurement R , is given by

Runam

1 .

(7)

c 1 1 . P

(3)

(This holds for a periodically pulsed radar and for small tP.) A combination of (2) and ( 3 ) leads to the ambiguous product

Figure 2 shows the spectral parts coming from the first and second ambiguous band and contributing to the desired SAR band schematically [ 5 1 .

The described ambiguity is mainly PRF conditioned. Howeve=, a main influence has the antenna pattern which is partly responsible for the spectrum shape, 2 Fig. 2. Especially improper (that means higher) side(4) uunam Runam 5 8 . f ' lobes enable the SAR to receive power from positive and negative squint angles, which is within the deThis equation shows that the choice of frequency for a sired frequency band. These ambiguities are called pulse doppler radar limits principally the possibility dopler squint angle ambiguities. for the simultaneous measurement of distance and velocity of a radar target. On the other side a frequency Fig. 3, 4, 5 and 6 show the influence of PRF on azilimit is fixed if the unambiguous values of u n n , , and muth ambiguities. Fig. 3 shows a 2-dimensional representation of a radar image of 5 corner reflectors R , are given. Example: and Fig. 4 gives the same image in 3-dimensional representation taken in X-band with a PRF of 952 kHZ. uunam = 300 m s ' , R,, = 40 km, Fio. 5 and 6 show the same scene taken with a PRF of 23i kHz. The ambiguities can be Seen evidently frequency requirement: f s 1 G H z ,

'

These relations derived for a simple RAR hold principally for SAR also. The equations show the importance of the PRF for all radar considerations. In fact, the PRF is the central radar parameter far SAR. The appointment of PRF has deep going consequences for the whole SAR, and the discussion of PRF limitations is widely identical with the discussion of the ambiguities. Range ambiguities will result if the PRF is too high, following in principle equation ( 3 ) . Azimuth ambiguities will result if the PRF is to low, so that the reflected signal phase changes by 2n radians ore more between two successive pulses. However, the SAR ambiguities are not only controlled by the waveform % image of 5 corner reflectors (X-band), PRF Fig. 3 S (represented by PRF for instance) but also bv the an"-- > tenna pattern.
1 1 ~

3.2 AZIMUTH AMBIUITIES FOR

SAR

Basic equation for azimuth ambiguity consideration is equation (l), which gives a lower limit for the PRF and which is identical with the requirement of the sampling theorem. f, is the maximum considered doppler frequency:

f ,

2u

sinp

2u

u is the platform velocity of the SAR and p the angle against the antenna mainlobe direction. (Normally the mainlobe of a SAR antenna is vertical to the velocitv-vector: however. but , souint anoles are ~ossible not considered here for simplicity reasons.) From (1) and (5) follows for small antenna beams ( 0 < 30) :
~~~~

~~

~1

~~~

~~

azimuth [mi
f rig.

4u

4 3dimensional representation of the SAR image in Fig. 3, PRF 952 kHZ. The contours on top represent -3 dB values.

fp

FOI the half power beamwidth phr = 2 p and because of the validity of phr = 111 follows:

-s

f ,

(6)

This equation defines the lower limit of PRF, it implies that the transmitter must be pulsed before the radar platform moves a distance equal to one half the real antenna length. The basis f o r the sampling theorem represented by (1) is the use of ideal low pass Fig. 5 S A R image of the corner reflector configuration limitation. However, this condition is normally not of Fig. 3 and 4 taken in X-band with 238 kHz fullfilled and this entails ambiguity levels within PRF, the azimuth ambiguities clearly can be the used doppler band. This level can be minimized by increasing the PRF, that means by a so-called overseen.

2-4

3 . 3 RANGE AMBIGUITIES

In order to guarantee a range unambiguity an equation similar to ( 3 ) holds:

f ,

1
Run
'

.rpt2-

This equation defines principally an upper limit for the PRF. However, this condition cannot be fullfilled

Fig. 7a SAR image of Zurich airport disturbed by sidelobe ambiguities caused by an antenna squint angle of 6.8" (Bhr = 12&1; near range 2805 m, far range 5355 m, u = 70 m s-', X-W (DLRE-SAR measurement Horn. Moreiral.

azimuth

Fig. 6 3dimensional SAR image of Fig. 5. The contours on top represent the 3 dB values for the mainlobes and the - 20 dB values of the azimuthal ambiguity sidelobes. at any time, it can be incompatible with the condition for the azimuth ambiguity and in this case one makes allowance for range ambiguities and tries to suppress these ambiguous signals with proper antenna pattern desion or with soecial orocessino resoecti_arocedures . vely. Fig. l a Shows exemplarically an X-band SAR image with ambiguities. The ambigcities are eliminated in Fig. 7b. Fig, lb SAR image of Fig. 7a, correctly prozessed to the squint angle 6 . 8 O ; all ambiguities are 4. PRF-CONSIDERATIONS eliminated (measurement Horn, Moreira).
~~~

viewing geometry is fixed other (given orheight Of satellite, fixed incidence angle bit a tuned to selected values. Fig. 8 show the PRF must be allowed PRF-bands for a satelliteborne SyStem with a fixed depression angle, taking into account the variation of the local orbit height r71. The white ranges the allowed PRF-ranges,-m is the numb'zr of the There are principal limitations due to azimuth arnbi- are respective range ambiguities. The lower limit for the guities, range ambiguities, swathwidth, complete co- PRF with respect to azimuth ambiguity is ,also given in verage. There are also unallowed PRF-bands due to geo- Fig. metIical variations like earth curvature, orbit excentricity, height variations, and altitude line The result is: echos. - with increasing range ambiguity decreases the re The ambiguity equations (6) and ( 8 ) lead to the iiollospective width of the allowed PRF-bands. wing PRF limitations: This can lead to the request of switchable PRF for 2" 1 different purposes (shown in Fig. 9 as necessary due <fpS to ALE-influences). Substituting (9) into ( 6 ) (consiRun ' dering the equality in both relations) onti ohtaines r p t 2 C together with the maximum azimuth resolution relation The choice of PRF estimates the maximum swath width R , an important relation between swathwidth, azimuth reor vice versa. Principally, an impulse needs for c ~ o s solution and SAR velocity: sing the swath the time T = 2%/c. Herefrom results: 2u Rg - c . (10)
~~

It has been already mentioned, that the PRF is a decisive factor for SAR-ambiguities. However, from other point of view the PRF has deep going consequences f o r the effectives of a SAR also. The definition of PRF becomes difficult due to different other conditions and limitations which have t.0 be fullfilled.

fp -' 2 % L -

(91

r.

A pulsed radar using a single antenna is normally unable to receive during the duration of the transmission pulse, it is blird at certain slant ranges. If the

2-5

E t

platform.
PRF-Bands
Satellite X - b n d

For a pulsed radar ALE c o n s i d e r a t i o n s w i l l be i d e n t i c a l with a l t i m e t e r - c o n s i d e r a t i o n s . An example f o r PRFl i m i t a t i o n due t o ALE shows Fig. 9 i n comparison with F i g . 8 . For m = 12, t h e i n f l u e n c e of ALE l e a d s t o t h e request of PRF-switching as mentioned i n t h e previous section. 5 . ANTENNA PECULIARITIES Equation ( 6 ) g i v e s a r e l a t i o n between t h e r e a l azimuth diameter of t h e SAR-antenna, f l i g h t - v e l o c i t y and t h e
PRF:

G.1,
fP

A minimum v e r t i c a l dimension of t h e antenna I,,,,

'
@Fig.

will be e s t i m a t e d due t o t h e n e c e s s i t y t o focus t h e beam i n t o t h e swath width %. From Fig. 2 i n l e s s o n 1 and i n conncction w i t h (10) t h e following r e l a t i o n can b e derived f o r a small h a l f power beamwidth:
10
20

30

40

50

60

70

80

I\h,*rn -

--

SO

8 Allowed PRF-bands v e r s u s a l t i t u d e v a r i a t i o n s f o r a s a t e l l i t e , X-band.

From I 6 1 and 1 1 0 ) follows i n connection with (1.1):

SATELLITE
Y-Band

Equation ( 1 2 ) g i v e s t h e lower bound f o r t h e antenna area.

6 . PECULIARITIESAND CONSTRAINTS CAUSED BY PHASE ERRORS

E s s e n t i a l f o r t h e e f f e c t i v i t y of a SAR system i s i t s phase coherency. High q u a l i t y images can be produced only i f t h e t a r g e t s phase h i s t o r y i s observed along a p r e c i s e l y known r a d a r t r a n s l a t i o n , However, t h e s e phas e h i s t o r y i s normally not known e x a c t l y . Phase e r r o r s u s u a l l y occur. P r i n c i p a l l y t h e r e a r e two d i f f e r e n t t y pes of phase e r r o r s : More o r l e s s d e t e r m i n i s t i c e r r o r s , caused by well known geometry e f f e c t s and d e f i n e d i n s t r u m e n t a l i n fluences. S t a t i s t i c a l e r r o r s caused by i n s t a b i l i t i e s of t h e rad a r i t s e l f , of t h e p l a t f o r m o r of t h e propagation p a t h f o r example.
6 . 1 DETERMINISTIC GEOMETRY EFFECTS, W G E CURVATURE

m=

m=

I n l e s s o n 1, Cap. 3 . 2 t h e "Depth of Focus" has been considered and i n Fig. 3 an example f o 1 t h e i n f l u e n c e of defocussing on a SAR image has been shown there. The same c o n s i d e r a t i o n s can be used f o r t h e computat i o n of t h e s o - c a l l e d "Ranqe Curvature E f f e c t " . The range c u r v a t u r e (RC) i s given by

Fig. 9 PRF l i m i t a t i o n s due t o a l t i t u d e v a r i a t i o n s Ah, example f o r PRF switching.

The A l t i t u d e Line Echo

RC = Re - R,

(13)

The a l l r e a d y mentioned a l t i t u d e l i n e echo (ALE) i s t h e r a d a r s i g n a l coming from Nadir. The time d u r a t i o n of ALE i s with s u f f i c i e n t accuracy i d e n t i c a l with t h e t r a n s m i t t e r p u l s e d u r a t i o n rP. ALE appears e a r l i e r than t h e d e s i r e d swath echo. On t h e one s i d e it w i l l be more a t t e n u a t e d t h a n t h e d e s i r e d s i g n a l p r o p o r t i o n a l t o t h e s i d e l o b e l e v e l of t h e antenna p o i n t i n g i n Nadir d i r e c t i o n and on t h e o t h e r s i d e i t w i l l be l e s s a t t e n u a t e d due t o t h e s h o r t e r d i s t a n c e , where ALE i s r e s u l t i n g from. so has normally a higher value f o r Nad i r d i r e c t i o n than i n any o t h e r d i r e c t i o n s and t h i s w i l l i n c r e a s e t h e ALE-signal l e v e l . P r i n c i p a l l y , t h e ALE can be used a s a r e f e r e n c e f o r geometrical c a l i b r a t i o n as well a s f o r a l t i t u d e e s t i m a t i o n of t h e

For high azimuth r e s o l u t i o n a p p l i c a t i o n a long i n t e g r a t i o n time i s r e q u i r e d and image degradation may be caused if t h e time d e l a y v a r i a t i o n corresponding t o (13) reaches t h e same o r d e r or exceeds t h e range resolution c e l l (T~/ZI f o r a p u l s e r a d a r . From 113) follows

? . ' R, RCm-.

16

12

(151

2-6

By this the importance of ra is evident. In comparison to ry folliws:

RC=-.
'Y

L2 R ,
(161
1 6 ry
1 :

Other so-called deterministic phase changes can occur due to orbit excentricity of Satellites m d each rotation effect, which can normally be estimted exactly or due to antenna influences like diagranune deformation or anqle switchinq. Both effects car, be measured and estimated exactly.. 6.2 PLATFORM INSTABILITY EFFECTS Fig. 11 shows measured displacements in the line of sight of a n airborne SAR, extracted by a RDM motior compensation equipment.

This equation gives the number of resolution cells through which a point migrates during the formation of the synthetic aperture. Ratio values in excess of 0.3 will normally cause image degradation and this must either be avoided by design or compensated during signal processing 121. In Fig. 10a the range curva:ure clearly can be seen. However, here the curvature is less enough to avoid image degradation. Fig. 10b shows the influence of a range curvature ef fect together with ambiguities in the sidelobes on the pulse answer of a point target. Here, a degradation clearly can be seen.

a
Fig. 11 SAR platform line of sight displazements extraction from SAR raw data by R ) M (DLRE-SAR, L-band, measurement Horn, liloreira) . Fig. 12 shows as an example the degradation of a point target imppulse response due to turbulences up to 2 m 5 '. (RMS 1 m s-') for an airborne SAR.
0

azimuth

-->

SAR row dofa

: : 7 m

-20

-40

-60

I I

m m c

-80 -1.5

-0.5

0
Time

0.5

I .5

[I]

Fia. 10a The raw data received from a point target (lower image) show a s well as the range compressed data the influence of range c u r v a -

s-',, peak value 2 m s-', tween -2 m upto 4.5 m)

displacements be-

azimuth Fig. 10b Impulse response of a point target disturbed due to range migration caused by range curvature. The sidelobes are caused by ambiguity. The Zdimensional representation on top sliows the estimation of deviations of the aircraft from evidently a smearing effect (DLR-E-SAR, fiarn, an ideal path as shown in Fig. 11 Moreira) .

In principle these effects can be neutralized by motion compensation by measuring the dynaniic behaviour of the platform and correcting the SAR data eighter on board of the aircraft or an grcund. TWO methods are principally possible, eighter the measurement using an Inertial Measurement Unit (IMU) ore a procedure, which extracts the true forward velocity and line of sight changes and acceleration out of raw data called "Reflectivity Displacement Method" RDM. RDM, recently has been developed at DLR [9, 1 0 1 . This procedure has been used to exclude the errors caused by aircraft motion and velocity instabilities during the processing of the images shown here. In the DLR approach only the information in the radar sianal has been used f o r mo-

2-1

Obviously, phase errors caused by atmospheric and ionospheric turbulences in principle cause the same motion errors as the aircraft instabilities. Therefore, the same method can be applied for the correction of phase errors due to propagation. This implies that for spaceborne SAR the negative influence of atmospheric and ionospheric turbulences to a large extend can be neutralised when they occur [lo]. This results in an increase of the resolution power and in image quality. Therefore, it seems in principle that for spaceborne SAR a dm resolution can be reached. However, special techniques are required in order to fulfill power and data requirements in this case. The tolerable platform velocity variation Au can be computed from (18) in lesson 1. The radar range R at the ends of a synthetic aperture (after the dwelltime TD) is:
R =

J
n : .
t (-12.

The differentiation with respect to u delivers dR= _ = _ _ du 4 R Au

=

"

AR Au Fig. 14 Example for errors in SAR images caused by phase distortions and the effect of motion compensation. The upper image is distorted due to a 12 % velocity variation, the lower image shows the same scene after a motion compensation. Scene dimensions are 1254 m x 1344 m, ground resolution 3 m x 3 m (measurement Moreira, Horn, DLR-SARI .

-AR

u T$ FOT a tolerable range error of k / 1 6 follows a tolerable velocity component Au (for R = R - ) : 4 R , Au=-.
u

u r;
_ = -

(17) R , A . Ti Remarkable is: The tolerable A" becomes more critical for a moving target at a distance R the displacement if the azimuth resolution is improved. AX
16

Fig. 13 gives the measured forward velocity variations of an airborne SAR during a flight time of about 60 seconds.

AX = i R

Ut r U Sin*

(18)

< t

!L
6r'-.~

. . -

.-

I 1

I ' ]
20
40 Tsme

1
i

The sign in (18) depends on the direction of the target velocity with respect to the radar. An example io1 image shift due to the radial component of target motion is shown in Fig. 1 5 . Ships with velocity components radial to the Seasat SAR orbit plane have an image displacement. This effect can be used for the estimation of target velocities if the distance between the SAR and the targets as well as the velocity of the SAR is known.

?
Y

.,~--.L.Li--J
0

1
60

80

[SI

Fig. 13 Changes in forward velocity of E-SAR extracted from DLR-E-SAR raw data, L-band, by RDM (measurement Horn, Moreira) . Fig. 14 shows as an example errors in a SAR image due to a 12 % velocity variation and its corrections with a motion compensation procedure.

6.3 TARGET MOTION EFFECTS

~

Usually a SAR signal processor locates the position of a non-moving target in the image place where its doppler frequency is zero. If the target is moving with a radial velocity component Ut=, then it imposes a Doppler shift on the signal. Thus a target moving towards the radar will cause a shift Ax in target location in the flight direction of the radar (and vice versa).

Fig. 15 SAR image of a sea surface with 2 ships taken by SEASAT. The displacement of the ships against the satellite clearly can be seen. The opposite displacement against the wakes corresmndinas to omosite velocitv directions (prbcessing: DLC-iJT-DA). Mean displacement is about 450 m.
In addition to radial velocity, radial acceleration

I f the radar is moving with a velocity U than results will also

image distortions such as defocus in

2-x

azimuth and range, range etc, XOWeYer, s e n s i t i v e e f f e c t i s azimuth defocus ( 3 1 .

linear
'Onsidered

the

' array is to be as long a s t h e d e v i a t i o n s from a s t r a i g h t l i n e a r e

lessthan a For a synthetic array t h i s f o r t h e two way case must be halved due t o t h e phase coherency c o n d i t i o n s . Therefore, f o r t h e l i n e of s i g h t d e v i a t i o n holds: AR s 114. An a c c e l e r a t i o n d 2 R , d t ~ acting for the dwelltime TD corresponds t o a r a d i a l p o s i t i o n e r r o r AR = d2R/dt2 (if a constant acceleration is assumed), In the two way t h i s c o e f f i c i e n t a l s o m u s t be halved. From t h e s e c o n s i d e r a t i o n s r e s u l t s with equation (14) from lesson 1:

L i k e o t h e r electromagnetic systems, r a d a r s employ rad i a t i o n w i t h p o l a r i z a t i o n , t h e alignment. ( r e l a t i v e t o v e r t i c a l o r h o r i z o n t a l ) of t h e e l e c t r i c v e c t o r i n t h e wave. P o l a r i z a t i o n on both t r a n s m i t and r e c e i v e is determined by t h e antenna. S c a t t e r i n g o b j e c t s , such as t e r r a i n f e a t u r e s or hard t a r g e t s Such a s v e h i c l e s , have r a d a r c r o s s s e c t i o n s t h a t r e f l e c t c ! i f f e r e n t l y i n response to the incident polarization. In p r i n c i p l e a complete d e s c r i p t i o n of a r a d a r t a r g e t can be given only if a l l l i k e p o l a r i z e d and cross-pol a r i z e d amplitudes and t h e r e s p e c t i v e phases of t h e radar s i g n a l a r e known. Such a p o l a r i m e t r i c r a d a r g i ves a l l p o s s i b l e information on a t a r g e t w i t h i n t h e r e l a t i v e small bandwidth of t h e modulated r a d a r c a r r i e r frequency.
A t p r e s e n t , technoloqicaL l i m i t a t i o n s i n t h e e f f i c i e n -

cy of r a d i o frequency power g e n e r a t o r s a r e important ( 1 9 ) design f a c t o r s . The '"radar equation" sho*S an i n c r e a s e dt2 2Tg R2.1 of t h e average power necessary f o r higher frequency ranges. The dependence is l i n e a r : t h i s mi?ans t h a t an This is t h e maximum t o l e r a b l e a c c e l e r a t i o n a t a r g e t X-band SAR needs about twice a s much pow?r as a Cmay have i n l i n e of Sight d i r e c t i o n of a SAR o r t h e band SAR, and about 7 times a s much a s an L-band SAR. radar platform may show v i c e v e r s a . ( 1 9 ) i s a b a s i c In addition, other technological d i f f i c u l t i e s increase equation f o r motion compensation. For turbulence d i with frequency. Phased a r r a y antennas ( d e s i r a b l e f o r sturbances i n Fig. 1 2 mainly a c c e l e r a t i o n e r r o r s a r e beam s t e e r i n g 1 a r e much e a s i e r t o r e a l i z o i n lower responsible. than i n higher frequency bands. I n L- and C-band a r e a l i z a t i o n with m i c r o s t r i p technology i:; s t a t e of I n Table 3 s e v e r a l sources of phase errors a r e l i s t e d t h e a r t , whereas i n X-band a r e a l i z a t i o n of a l a r g e t o g e t h e r with t h e r e s p e c t i v e e r r o r s caused w i t h i n SAR m i c r o s t r i p antenna with s u f f i c i e n t e f f i c j e n c y f o r imaoes. The Fioures 16a t o 16d show a s examoles t h e space a p p l i c a t i o n s seems t o be extremely d i f f i c u l t . i n f l u e n c e of d i f f e r e n t phase e r r o r s on t h e impuls r e S t a t e of t h e a r t f o r X-band use s l o t t e d a'aveguide sponse of a point t a r g e t . a r r a y s 111, 121. This IS t h e main reason f o r t h e singl e p o l a r i z a t i o n of X-SAR a g a i n s t t h e m u l t i p o l a r i z a t i o n 7 . REALIZATION CONSIDERATIONS c a p a b i l i t y of SIR- C.

-s-=-.

d2R

2u2

Atmosphere and ionosphere produce frequency dependent distortians, These effeCtS set an upper frequency lim i t due t o a t t e n u a t i o n f o r a i r b o r n e r a d a r labout 9 0 GHZ) and spaceborne radar labout 15 G H Z ) and a lower l i m i t f o r spaceborne SAR due t o ionospheric g r a n u l a r i t i e s labout 1 G H z ) . The S t a t e of technology s e t s upper limits a s well. The frequency bands a v a i l a b l e f o r rad a r s u r v e i l l a n c e a r e a l s o l i m i t e d due t o internat.ional agreements.

The mean t r a n s m i t t e r power determines t h e d i s t a n c e from which a radar observation t o a c e r t a i n t a r g e t can be SuCCeSSfUllY made and, therefore, the orbit altitude of spaceborne is power limited,

Table 3 E f f e c t s of phase e r r o r a c a u s e d by platform o r t a r g e t motions r e s p e c t i v e l y ( t h e r e l a t i v e mut i o n i s from relevance o n l y ) , a s well a s caused by phase j i t t e r Arp i n t h e propagation p a t t . and i n t h e SAR e l e c t r o n i c .

2-9
0 -10 0

-10 -20
I

2 . 8

-20

-30 4 0
-50

2 .

-30
-40

= -

s s
<

E"

-50

-60

-60

-70
-0.6

-0.4

-0.2

( 1 . 2
Tlma [SI

-70
0.. 0.6 -0.8

-0.I

-0.2

0.7.

0.4

0.6

rima[.I

Fig. 16a Impulse response, shifted by a linear phase error.

0

Fig. l6c Impulse response, degraded by a cubic phase error.

Tlmo

[SI

11mo

[SI

Fig. 16b Impulse response, degraded by a quadratic phase error. Principally, the power required increases with the 3rd power of the radar distance. (This means a doubling of the distance, i.e. the orbit height, requires a n 8-fold multiplication of the required power). A larger antenna beams more power to a required area (expressed through the so-called antenna gain) than a small one, and, therefore, a large antenna would seem to be favourable. But the ground resolution of a SAR is improved for smaller real antennas. These considerations lead to a tradeoff, which must be made carefully in order to fulfill a satellite SAR's requirements. In any case the transmitter power is a limiting element for the design of a SAR as well as the antenna. Todays state of the art are a few hundreds watt (mean power). This implies requirements for the primary power supply of a satellite. State of the art is about 6 kW to 1 0 kW. These requirements can be fulfilled with solar power generators and atomic generators as well. The antenna dimensions for spaceborce SAR have at present values of about 2 m x 15 m.

Fig, 16d Impulse response, degraded by a random high frequency phase errors. capacity. A reasonable goal would be 250 Mbit s-'. New data transmission systems with splitted data links to data relay satellites or ground stations respectively are under preparation. Data storage capability has to be increased also A S a first step new recorders with capabilities exceeding 100 W i t sK1 are qualified f o r use in space. The present approach is to use more than one recorder, i.e. one recorder for each channel in multipolarization and multifrequency SAR, as on the SIR-C mission. Image data processing capabilities also are limited, However, this processing time will decrease rapidly during the next few years and at the end of the decade real time processing with excellent image quality will be possible. Requirements for high resolution and wide swath in continous strip map mode are in conflict with data handling capabilities, and with requirements for a large antenna to conserve power. Therefore. new SAR techniaues have to be introduced a n d developed which 'allow electronically steered beams. For this purpose the spotlight mode and Scan-SAR modes are under consideration. The length of a synthetic antenna corresponds to the section of the flight path from which one target stays within the antenna beam. This fact leads to the requirement for wide beams and, therefore, small antennas for conventional high resolution systems. The same effect, however, can be reached if a small antenna beam (from a large antenna) can be continuously pointed at the target. This allows also a longer synthetic array and, therefore, a finer azimuth resolution [ l l ] . However, the gain of azimuth resolution entails a loss of coverage due to the fact that during the continuous spotlight illumination of one small surface

8. FUTURE SAR TECHNIQUES

SAR sytems produce a tremendous amount of data (examples for data-recorder bitrates on the ground: ERS-1 102 MBPS, X-SARISIR-C 2 45 MBPS per channel ( ! ) , DLR airborne E-SAR 28 MBPS). Requirements for simultaneous high resolution and large swath widths make the data rates higher. All requirements for extensions of SAR to multifrequency and multipolarizatian capability respectively entail a multiplication of the data rates and this would exceed the present limitations of data handling. This seems to be a key problem in all high resolution imaging systems. Therefore, different requirements have t o be fullfilled in order to handle or reduce the data stream of future systems either by means of onboard processing or with development of advanced SAR systems like spotlight SAR. One solution is to increase the data links

2-10
a r e a , t h e sensor passes o t h e r p a r t s of t h e swath which a r e not i l l u m i n a t e d . Therefore, t h e s p o t l i g h t mode can be used f o r the e n l a r gement of a sector of t h e observed swath s i m i l a r t o t h e zooming with an o p t i c a l camera. The Scan SAR Mode can be used f o r an extension of t h e swath i n r a d i a l d i r e c t i o n using more than one beam generated i n a time shared manner. This method i n c r e a s e s t h e swath a t t h e expense of a z i m u t h r e s o l u t i o n I 1 6 t o 2 0 1 . A s p o t l i g h t SAR i n o r b i t , however, would be a r a t h e r expensive enterprise. The combination of a l l modes allows v a r i a b l e r e s o l u t i o n and swath widths a s well; t h e t r a d e o f f between r e s o l u t i o n , swath width, power e t c . l e a d s t o optimised c o n f i g u r a t i o n s .
9 . REFERENCES

German P a t e n t P 4 1 2 4 0 5 2 . 6 .
Ill1 BuckreuR, S .

Motion E r r o r s i n an Airborne S y n t h e t i c Aperture Radar System. ETT Vol. 2, No. 6, NovIDec. 1 9 9 1 , pp. 555-554. V e r i f i c a t i o n Using Spaceborne Microwave Imaging. IEEE Technology and S o c i e t y Magazine, Dec. 1990/Jan. 1 9 9 1 , pp. 53-51. C-Band SAR R e s u l t s Obtained bv

1121 Keydel, W .

I131 Horn, R

[ I ] Skolnik, M.I.

Radar Handbook. McGraw H i l l Book Comp., N e W York, 1 9 7 0 . I n t r o d u c t i o n t o Synthetic Array and Imaging Radars. Artech House, Inc., 1 9 8 0 .

:141 Moreira, A.

I21 Houanessian, S.A.

Improved Multilook Techniques Applied t o SAR and SCANSAR Imagery. IEEE Trans. on Geo!;c. and ~ e mote Sensing, Vol. 2 9 , N O . 4, July 1991.
A New Subaperture lipproach f o r Real-Time SAR P r o c e s s i n a , ETT, V O 1 . 2 , N O . 6, NOv.-DeC 1991.

I151 Moreira, A .

L31 Tomiyasu, K .

T u t o r i a l Review of Synthetic A p e r t ' x e Radar (SARI with App l i c a t i o n s t o Imaging of t h e Oceans Surface. Proc. of t h e I E E E Vol. 65, No. 5, May 1 9 7 8 . Synthetic Aperture Radar. The Artech Radar Library. Artech House, I n c . , 1 9 7 6 Microwave Remote Sensing Vol 11, Radar Remote Sensing and Surface S c a t t e r i n g and E m i s sion Theorv. Addison Wesley Publ. Comp., 1982. Spaceborne Synthetic Aperture Imaging Radars: Applications, Techniques and Technology. Proc. I E E E V O 1 . 70, NO. 1 0 , Oct. 1 9 8 2 , pp. 1174-1203.

[151 Brunner, A .
Langer, E . Ottl, H . zeller, K

I 4 1 Kovaly, J . J .

:51 Ulaby, F . T . Moore, R . K . Fung, A . K .

Concept f o r a Spaceborne Synt h e t i c Aperture Radar (SARI Sensor Based on Act.ive Phased Array Technology. AGARD Conf. Proc. No. 459 on Hiqh Resolution A i r - and Spaceborne Radar, Papers p r e s & t e d a t t h e Avionics Panel SymDosium h e l d i n The Haaue. The Netherlands, 8-12 Pug: 1589, pp. 23 A 1 - 23 A l O . Concept of an X-Band S y n t h e t i c Aperture Radar f o r Earth Observing S a t e l l i t e s . Journ. of Electromagnetic Waves and Applications, May 1 9 9 0 . Taking a Broader View: Radarsat adds Scansar t o i t s Ooerations. Proc. IGARSS ' 8 8 Symp.., Eding burgh, Scotland, 13-15 S e p t . 1988, pp. 1027-1032. Canada's RADARSAT. Remote Sensing Yearlmok 1 9 9 0 , (personal comunicarion) The Radarsat S y n t h e t i c Aperture Radar: A F l e x i b l e Imaging System. Proc. 11. Canadian !iymp. on Remote Sensing, Water.Loo, 22.-25.6.1987.

161 Elachi, C h . Bicknell, T . Jordan, R . L . Wu, Ch [7l Schlude, F .

(171 JatSch, W . Langer, E . Ottl, H.

[ 1 8 1 Luscombe, A .

Imaging Radar Systems. Proc. of a n ESA-EARSeL Workshop h e l d a t Alpbach, A u s t r i a , 16- 20 March 1981 on "Coherent a n d Incoherent Radar S c a t t e r i n g from [ 1 9 1 Raney. K Rough Surfaces and vegetatod Areas", ESA-SP-155. Estimating t h e Residual Error of t h e R e f l e c t i v i t y Displacement Method f o r A i r c r a f t MOt i o n Error E x t r a c t i o n from SAR Raw Data. ICCC I n t e r n . Radar Conf., A r l i n g t o n , USA, 1990, pp. 10-.75.
A New Hethod of A i r c r a f t Motion Error E x t r a c t i o n from Radar Raw Data f o r Real Time SAR Motion Compensation. 1 2 t h Canadian Symposium on Remote Sensing, IGARSS '89, Vancouver, Canada, proc. IGARSS, I E E E , 89, pp. 2217-2220.

I81 Moreira, J.

I 2 0 1 Luscombe, A . P

[9] Moreira J.

:IO] Keydel,

W.

Moreira, J.

Verfahren zur Extraktion "on durch d i e Atmosphare verursacht e n Phasenfehlern des Ricks t r e u s i g n a l s e i n e s Abbildungsradarsystems aus Radarrohdaten.

3-1

MOTION ERRORS AND COMPENSATION POSSIBILITIES hv -, D. Hounam Deutsche Forschungsanstalt fur Luft- und Raumfahrt.e.V. Institut fiir Hochfrequenztechnik 8031 Oberpfaffenhofen Geimany
1. S W R Y

The synthetic aperture radar (SAR) technique relies on knowledge of the relative motion between the sensor and the target. If the flight path of the sensor is not accurately known or the SAR processor is limited in its ability to take the flight data into account, the SAR image will be degraded. Motion errors are particularly critical for SAR sensors on small, low-flying aircraft, due to turbulence, and where high spatial resolution is required. The lecture discusses the effects of motion errors on image quality and the requirements on the sens o r and processor to compensate for motion errors. The DLR airborne sensor, E-SAR, and associated image processor will be used as examples. Techniques using a priori knowledge of the flight path from independent sensors, e.g. inertial navigation Systems (INS), and by extracting the flight data from the SAR data, e.9 autofocus and reflectivity displacement method (RDM), are treated. The author would particularly like to thank J. MOreira and S. BuckreuR whose work, referenced belaw, was used extensively for the lecture.
2. INTRODUCTION

along-track position and spatial resolution will not be influenced, by attitude errors. However, if the antenna beam points to a different part of the Doppler spectrm than that which is processed (filtered), the image intensity will be influenced. With serious mismatch, defocusing will occur, due to distortion of the matched filter weighting function and errors in the range migration correction. Also ambiguous responses can occur. Antenna pointing needs to be known to process the SAR data, o r the processor has to derive it by analysing the Doppler spectrum. The latter can be considered normal processing practice and, therefore, in the following, these effects will be neglected. Whereas, with the above reservations, the focusing of a SAR is insensitive to angular deviations, deviations in the path of the sensor will lead to displacement and defacusing of the target in the final image. The path of the sensor does not need to be a straight line but it must be known with sufficient accuracy so that deviations can be taken into account in the processing of the SAR data to achieve a sharp image. This process is often referred to as motion compensation. When discussing the path of a sensor, the question arises as to where the reference point within the sensor is located. Displacements of the sensor are clearly only effective on the propagation path of the radar signal, Hence, the reference point within the sensor is the phase centre of the antenna, i.e. origin of the spherical far-field wavefront. The phase centre is the point from which the antenna effectively radiates. For many antenna types, the phase centre is outside the Structure of the antenna, e.g dish antennas, where it is located behind the dish. Hence, by the path of the sensor is understood the path of the phase centre of the antenna. If the path is not measured directly at the phase centre, roll, pitch and yaw angles need to be known to correct for the offset. Errors in these angles will then lead to degradation in image quality, further contradicting the generalisation above that the focusing of the SAR is insensitive to angular variations. For the sake of simplicity, these angular effects have been neglected below. If needed their influence can be derived from the formulas provided.
Slnarlv. a - st-able fliaht. Dath areatlv eases the
j.
~~~ ~ ~ ~~ ~ ~ ~~~ ~

Every amateur photographer is familiar with the blurring of his photograph if the camera is not held steady while the film is being exposed. The action of the lens focusing the scene on photographic film results in the scene being resolved in angular units. Rotation of the camera in any plane will cause the scene to move across the film and blur the resulting image. In contrast to this well known effect, the focusing of a SAR sensor is in principle insensitive to angular deviations. This is because the sensor resolves the scene in terms of displacements rather than angles. In the plane orthogonal to the flight path, i.e. in t h e range direction, the position of a target is determined by measuring the delay of the tacget's echo (range delay). The spatial resolution is determined by the length and bandwidth of the transmitted pulse. If the sensor is rotated in this plane (roll angle), the influence of the nntenna gain pattern may cause the strength of the echo, i.e. the image intensity, to vary but the delay and, hence the spatial resolution will remain unaffected. Parallel to the flight path, i.e. in the azimuth direction, the scene is resolved by matched filtering the Doppler spectrum, there being a fixed correspondence between Doppler frequency and the relative position of the target on the ground. The spatial resolution is determined by the spacing of the lines of constant Doppler and the bandwidth of the matched filter. Rotation in azimuth produced by pitch and yaw will have no influence on the lines of constant Doppler nor on the matched filter characteristics. In principle, therefore, the

task of motion compensation. Satellite platforms fall into this category even though in Some cases, e.g. space shuttle the orbital parameters as well a s the attitude can be poorly defined. The real challenge for motion compensation is presented when SAR sensors are flown OD light aircraft due to the influence of air turbulence and the often frantic effort of the pilot to combat it. The airborne SAR sensor of the DLR (ESAR) is such a case 111 and will be used as an example in the f o l l o wing analysis. Fig. 1 shows an image of the DLR centre in Oberpfaffenhofen using this Sensor which exhibits typical degradations due to motion errors. The aircraft was deliberately flown to produce motion errors which can be seen a s blurring of the image (top right1 and geometric distortion

~.~~

3-2

of the runway and taxiways

In the following analysis, the influence of motion errors on the azimuth imaging properties will be discussed first a s this is far greater than the influence in range. The an31ysis closely folloiims the approach of Buckreui) ( 2 1 . These properties are discussed in terms of the impulse response, which is the response Of the SAX sensor, including processing, to a point target.

t .

Fig. 2 SAR geometry, where P is a target 3" the ground, H is the altitude of the sensor, D is the distance of the target f m m the ground track, Ro is the slant range at closest approach, 0 , is the azimuth angle and 0 , is the depressj~onangle of ithe target. distance between the target and the ground track. Alternatively,

where Rg is the range at the point of cloz:est approach of the sensor of the target. 'ig. 1 SAR image of the DLR centre in Oberpfaffenhofen exhibiting strong motion errors.
2 . 1 Impulse Response in the Azimuth Direction

Integrating and defining t = 0 when the sensor is at the point of closest approach, we obtain the slant range r:

Let us consider first the azimuth channel of a SAX system, where the platform flies in a straight line (Fig. 2). The coherent integration can be expressed by the convolution of the backscattered signal Slt), with a reference function H ( t 1 , where Zit) is the impulse response;.

Ignoring the constant phase expression, the two way phase change q l t ) , mentioned in Eq. 121 becomes: B(t)
Or,
i

/Zit)

IS(t)

H(t) 1

(1)

4n t2 . v2 h 2.R0
~

where A is the radar wavelength.

(71

The retuined signal S(t), corrupted by a phase e r ror O ( t ) IS given by

?(t)

.t' ,

where

= 7 i,.Ro '

4n.v2

(8)

were A. is the amplitude of the returned radar si.gnal and q(t) is the nominal. phase history of a point target. According to the geometry (see Fig. 2) the velocity in line of sight (LOS), is: VLOS(t)
=

Hence, the nominal phase moduiation is a qiadratic function of time, i.e. linear frequency modulation. Let us now consider deviations from the nominal path where the platform is displaced from the nominal position A to the position A ' (see Fig. 3). The range r' becomes: r'lt)
=

V,sin@,(t)

(3)

where V is the forward velocity of the aircraft and 0* i s ther angle between the LOS vector and cross-track plane (azimuth angle). Far small values of OA, we can put:

Ro t V2 t2 t A X ( t )

RO

'eI,(9)

- AyltIc0~0~ + A ~ ( t ) s i n 0, ~

B , ( t )

where H is the height of the sensor and D is the

w'

Vt -

141

where Ax(t) is the displacement in the flight direction, A y l t ) and Az(t) a r e the displacements of the platform in y and z direction, respectively, and OD is the depression angle of the target.

3-3

the phase error Q(t!. If the weighting function is a rectangular window w(t)= WR(t), with

and Q ( t ) = 0, the ideal impulse response IZo(t)I becomes :

It will be seen below that the phase error function m(t) caused by motion errors can lead to considerable distortion of the azimuth impulse response.
2.2

Impulse Response in the Range Direction

Differentiating, Eq. (10) we can obtain expressions for the angular frequency error Q(t1, due to velocity components along the three axes.

Fig, 3 Deviation from the nominal flight path, du e to turbulences. The phase error Bit) is, therefore: Qit)
=

4n

IAX(t!.Qa(t)

Ay(t!.coS%

t AZ (t) sinQ,l.

(10)

From equation (10) it can be seen that the influence of the motion errors in the x direction is independent of the depression angle Qa and, therefore, of the position of the the target within the swath, Whereas, in the other two directions (y and z ) , QD needs to be known. Also, it is evident that the variations of the displacements with respect to time in all three directions is reproduced in the phase error, a linear variation in the displacement manifesting itself as a linear phase error and so on. The normalized reference function H(t) from Eq. (1) is given by:

The first term in the brackets of Eq.117) represents the frequency error resulting from the displacement in flight direction. The second term shows a linear relationship between angular frequency error and velocity component, it being proportional V.t/RO, i.e. the azimuth angle % . This angle is the sum of the so-called squint angle, i.e the angular offset of the boresight of the antenna, and half the azimuth antenna beamwidth. Eq.(19) and E q . 1 2 0 ) also show a linear relationship between frequency error and velocity component. Assuming that the velocities are constant, we can now calculate the phase error op across the transmit pulse length
Qpx =

H ( t ) = w(tj.TE.e

-T/2 s t

tT/l

(11)

6 x . ~,pQpy

6y'%p,

Qpl

= iz.Tp

(21)

where T is the aperture time and W(t) is a weighting function to suppress sidelobes. Often a Hamming weighting function [ 4 1 is used:
W"(t) = a t ( 1 - a ) . c o s [ $ % ]

a =

0.54

(12)

Using equations (11, ( Z ) , and I l l ) , the impulse response of a point target becomes: Z(t)
=

Let us now calculate some practical values for the velocity components and the displacement in flight direction, which would each result in a linear phase error of n/2, a value which would need to be achieved to influence the impulse response. The results are given in Table 2. We will use the parameters for a typical satellite sensor (ERS-1) and airborne Sensor (E-SAR) given in Table 1. ERS-1 E-SAR
75

S(t)

H(t) =

+I --

tr.

S(t).H(t-r) dr

(13) Velocity Closest approach Wavelength Depression angle

Or,

Taking the absolute value of l v obtain:

Zit),

we approximate

Max.azimuth angle , , Q Pulse length

1 ;1 1 1 1
V

7100

I m/s

850

0.0566 20.55

0.0566 m
4;

km de9 rad
PS

0.013

0.14

37.1

Table 1 Sensor parameters This can also be considered as the Fourier transform of the weighting function W(t) modulated by

-3-4

where pI0) is the position error, "(0) the velocity error, a ( 0 ) is the acceleration error and & I O ) is the derivative of the acceleration error at the beginning of the synthetic aperture Table 2 Displacement in flight direction and velocities resulting in a phase e r r o r across the pulse of rI2. The v a l u e s in Table 2 show that the displacement and velocities necessary tc influence the range imp31se response are far higher than achievable with both satellite and aircraft platforms. Hence, motion errors can be considered to only influence the azimuth impulse response, and this will be concentrated on below.
3 . CLASSIFICATION OF PHASE ERRORS

iuruist R t s r o v s t

Let us consider a sinussoical phase error m(t)=

OOCOS 12cfot-~[o)with constant frequency fo and

constant phase shift '[o. The impulse response of a point targec from Eq.115) kcomes: IZ(t1 1
=

0
-20
-15

A.

'

(221
,

T v i ( 7 )

. e jOocos(2nforl

.-jktr

--

dr

1.

4 0

~5

IO

15

20

AiimulhIm]

iquation (221 describes a F3urier transformation of the product of the weighting function W(t) and a frequency modulated signal. This corresponds to the convolution of the Fourier transformed weightinq function W l t ) with a Bcssel function, the latter resulting from Fourier transforming the F I signal. Assuming a small amplitude of the phase error Oo e 1, the frequency modulation can be approximated by an amplitude modulation and the imw l s e response becomes:

Fig. 4a Impulse response wit3 no phase errors and the shift due to a linear e r r o r of n rad.

where Zgltl is the ideal impulse response from Eq. (171. Eq. ( 2 3 ) shows we obtain one main impulse response and so called 'paired echoes', located at t = r2nfo/k, with their maximum being a factor 0 0 / 2 below the main lobe

For phase errors with frequencies above 1/T the paired echoes are spaced away from the main lobe alld result in an increase of the sidelobe level. In general, energy from the mainlobe is transferred to the sidelobes decreasing the Integrated Sidelobe Ratio IISLRI, which corresponds to a loss of Contrast in the image. The ISLR is defined as:
Energy of the Mainlobe ISLR
=

Az#mulh[m]

Fig. 4b Impulse response with a quadracic phase error of n rad. Each term causes different effects on the image quality:

Energy of the Sidelobes If the frequency f o of the sinusoidal phase e r r o r is lower than the reciprocal aperture time lIT, i.e. low frequency phase errors, the paired echoes will merge with the mainlobe and cause its deformation. The low frequency phase error can be expanded into a Taylor series:

* The constant term has no effect on the image quality. In this case, the aircraft can be supposed to fly parallel to the nominal track and a different area is mapped.
The linear term corresponds to a shift of the mainlobe in azimuth direction causj.ng geometric distortions of the image.

3-5

IMPULSE RESPONSE

and

Gm If)

sin2@,

'

( P I :

Gr,,,(f!

(30)

where G , , ( f ) , GDy(f) and G , . ( f ) denote the displacement PSDs of the platform in x, y, z direction. The expected image quality can be predicted, if the power spectral density of the phase error is known. In the following, the effects of the phase errors, classified above, are expressed by means of the PSD. 4.1 Linear Phase Errors At first we consider a deterministic sinusoidal phase error with a constant frequency f , and a constant phase shift y o r a!t)
=

-20

-15

-10

-5

10

15

hzimulh[m]

O~c0~12rf~t-y . o~

(311

Fig. 4c Impulse response with a sinusoidal phase ' e r r o r of 0.1 rad amplitude and 6 cycles across the interval showing paired echoes.

Expanded into a power series, the linear term mLlt) becomes QL(t)
=

mo.211fot

for

yo=z/2

(32)

* The quadratic term causes a broadening of the mainlabe, respectively a loss of geometric and radiometric resolution. * The cubic term Causes an asymetric distortion of the mainlobe and an increase of the sidelobe level, which also leads to a degradation of the radiometric resolution and a loss of contrast.
Fig. 4 shows examples of the influence of linear, quadratic and sinusoidal phase errors on the impulse response for amairborne sensor with a nominal azimuth resolution of 2 . 4 5 m.
4 . MOTION COMPENSATION REQUIREMENTS

The phase history of a point target vlt), corrupted by the linear phase error 0,Iti can be denoted as 9 ! t !
t O,(t! = - kt2 t pl,.2'Tf0t 2

(33)

Thus the impulse response is shifted by the distance 1, where

k.R .Oo.f, . &

(34)

We can see, that the shift of the impulse response is proportional to the slope of the linear phase error, which is determined by 00 and f o . Considering all frequencies contributing to the linear phase error, which are defined by Golf!, the variance of the shift of the impulse response oL2 can be evaluated with

The motion error of a platform is regarded as a stochastic process, which is assumed to be stationary and ergodic. Thus it can be described most completely by a power spectral density IPSD) 131.

The following E q . 1 2 5 ) denotes the relation between the phase error PSD G , ( f ) and the PSD of the displacement of the platform from the nominal track %If). Within E q . 1 2 6 ) and Eq. ( 2 7 ) the factor (1/2nf)2 corresponds to an integration in the time domain, which enables the conversion from the acceleration PSD G , ! f ! to the velocity PSD Gv(f) and to the displacement PSD %If!.

4.2 Quadratic Phase Errors

Concerning the sinussoidal phase error denoted in E q . (311, for yn=O, the quadratic term of the power series becomes

At the end of the synthetic aperture when t = T/2, the quadratic phase error is

The components of the platform motion in x, y and z direction are contributing to the total phase error PSD GO(f). Using Eqs. (10) and 1 2 5 ) , we get for the x-axis

Considering all frequencies contributing to this phase error, which are described by Golf), the variance oQz of the quadratic phase error, measured at the end of the synthetic aperture results in the expression

The components in y- and z-direction are denoted as:

3-6

The relation between the mainlobe broadening and the auadratic Dhase error at the end of the svnthetic aperturk 0, lT/2) was determined empirically, by correlating a simulated backscatter signal of a point target with a Haming weighted reference function ( o = 0.54) and is described by Eq. (391.

where p I is the mean value of the resolution of one look. The looks were evaluated with a Haming weighted reference function and finally root sum squared.
4 . 5 Determination of the Acceptable Motion Error

The ideal resolution is denoted as p i d e a l and the resolution, which is degraded by a phase error is peer. For example, a quadratic phase error of 1x12 at the end of the synthetic aperture results in a mainlobe broadening of 6 . 6 % .
4 . 3 High Frequency Phase Errors

The acceptable residual motion error, which remains uncorrected, repiesents the requir,edperformance of the motion compensation system. The following analysis is based on the E-SAR system parameters given in Table 3 and the image qu.nlity specifications in Table 4 .

It has already been pointed out, that phase errors with periods below the aperture time T will have an impact on the integrated sidelobe ratio o r the image contrast. According to Haslam 141, the required expression is

platform: altitude: * forward velocity (nominal): maximum slant range: * depression angle: * antenna beamwidth: wavelength:

Dornier Do 228 lGO0m . . . 3000m V = 70m/s Ro= 600Om

en= 0'

... 70'

aperture time:
4 . 4 The Effect of Phase Errors

on Multilook

SAR

Images

BB= 10" kL= 0.2:;OBm ic= 0.0566m Ix= 0.0313m TL= 1 5 . 0 s Tc= 3 . 6 8 s
TX= 2 . 0 3 s
=

The multilook technique was developed to reduce the speckle of S A R images. This can be achieved by dividing the synthetic aperture into overlapping subapertures, which are finally summed up incoherently. The impact of motion induced phase errors on each l o o k can be calculated with Eqs. ( 3 5 ) , 1381, 1391, l4O), where the integration time of one look T , has to be taken into consideration.

azimuth resolution (1 look): p

0.66m

Table 3 System parameters of the E-SAR

A linear phase error has the same effect on a multilook image as pointed out in 4.2.1: Each look is shifted in the same direction by the same distance. A

Integrated Sidelobe Ratio: ISLR = -2OdB loss of geometric resolution: < 10% * pixel shift: c 50% o l one resolution cell
~ ~~ ~~

ouadratic ohase error causes a mainlobe broadening and a displacement of the l o o k s from each other.

Table 4 Required image quality.

4 . 5 . 1 Specification

of the PSD for an Acceptable

Motion Error
A power spectral density of the displacement of

The distance dN between the first and last look I 5 1 is given by

the aircraft can be soecified. which ex~rrssesthe

where a, is the nominal Doppler rate,

and bar is the deviation from the Doppler rate due to the quadratic phase error OQ, which is given by 143) *.TN2 From Eqs. (411, ( 4 2 ) , (43) end after replacing the look displacement dM by its standard deviation oQN, we get:
Aa, =

4 . 0 , .

The mean value of the resolution with N looks ob,, can be estimated empirically and becomes apprb-' ximately
~~

2.3

subdivisions, which are characterized by expinential functions, depicted as straight lines, using a double logarithmic scale (Fig. 5 ) . For computation, Eqs. ( 2 5 ) , (351, (38), (391, (40) were used. The coefficients and exponents of these subfunctions were adjusted numerically to achieve the requirements in Tab. 4 for LICIX-band. & , i f ) is limited to a frequency range from 0.001 Hz, which corresponds to a maximum duration of one pass of 1000 seconds or appoximately 16 minutes, t3 an upper boundary of 2.2 Hz. Frequencies above 2.2 HZ would cause a maximum ISLR of approximately -30 dB lx-Band) and are not considered to degrade the image quality markedly. The result is given by Eqs. l46)-(50i and is also shown in Fig. 5 . The exponent of the subfunction (50) was intentionally set to Zero, to reduce the degrees of freedom for further computation. It is remarkable that in this case the requirements'for the linear phase error turned out to be more restrictive than the requirements for the quadratic phase error. Thus only Eq. (351 was used to specify the low frequency section of %,if).

PN

~PI'

'

145)

rors were derived from the PSDs of Fig. 5, and impulse responses were computed according to the ESAR system parameters, using a Hamming weighted reference function ( a = 0.54). Typical results for C-band are depicted in Figs. 6-9. The impulse responses of Fig. 7 and Fig. 9 are normalized to the maximum of an ideal, ""weighted impulse response IZo(t)I from Eq. (16).
The required transfer function S(f) of the motion compensation system is shown in Fig. 10 and is given by

measured PSD o f

the

aircraft -

. "
a
0

-60-80.

s p e c i f i e d PSD f o r LICIX-band

(521

IE-3

0.1 Frequency [Hz]

Fig. 5 The measured PSD of the motion of the DO 2 2 8 aircraft, compared to the specified PSD of the acceptable motion error.
2

Ga,(f)

6.46.10-5

.iTL.f)-3-5 , O.0OlHr i f i l/Tz

(46)

G , , ( f )

3.41.10-6

& .(TC.fl-2'093
2

(471

l/TL i f i l/Tc Goa(f)

= =

4.11.10U8

.(Tx~fl-7'456 , l/Tc i f 5 l/Tx

i

(48)

Tlma

[.I

G , , ( f ) Gas(f)

4.11.10-8

E , l/Tx

2.2

HZ

(49)
(501

Gom(fl

2.2Hz i f <

-.

Fig. 6 Simulated motion error, derived from the measured PSD of the aircraft.

For easier interpretation of the specified PSD G,,(f), the standard deviation uDS of the displacement from the nominal track in line of sight of the antenna can be calculated using:
I

m m

-20

In this case the forward velocity of the aircraft is assumed to be constant. Thus we get a standard deviation of cDs = 0.4 m f o r the low frequency section of GDs(f) for L-, C- and X-band. This means, the RMS deviation from the nominal flight path should not exceed 40 cm within 1000 seconds, to keep the RMS mainlobe shift within 3 3 cm. For the high frequency section of G , , ( f ) L-band: oDs C-band: oDs X-band: oDs

= -

-40

-60

= =

1.84" , 0.45mm , 0.25mm ,

lIT, I f l/Tc 5 f l/Tx s f

2 *

-.

we get
-80
-1.5

-0.5

0.5

1.5

rime [.I

To obtain an I S L R of -20 dB in LICiX-band, the RMS deviation from the flight path should not exeed 0.8 % of the used wavelength within the aperture time .
4.5.2 Simulations

Fig. 7 Simulated C-band impulse response, degraded oy phase errors due to Fig. 5, compared to an undistorted impulse response.

In addition to theoretical examinations, the obtained results were verified by simulations, based on the Monte Carlo method. Therefore, motion er-

5 . MOTION COMPENSATION METHODS

Path Measurement with a Strapdown Inertial Measurement Unit (IMU) Within a strapdown inertial. measurement unit, the sensor elements are directly mounted to the aircraft frame, whereas linear motions are detected by accelerometers and for angular xate rieasurement, gyros are used. The following information is usually supplied by a commercial strapdown system:

i J 4 6 8

* aircraft referenced accelerations, referenced attitude, * inertial velocity, ground speed, * heading, * position.

To obtain this information, the foliowing evaluations have to be made by a strapdown computer:

Fig. 8 Simulated notion err%?, derived from the specified ?SD due to E q S . 1 4 6 ) - ( 5 0 1 .

* compensation of the rotation of the earth and the transport rate, bath beeing included in the gyro signal, * compensation of the gravity, inclcded in the accelerometer signal, * computation of the attitude angles of the aircraft (pitch, roll and yaw), by integmting the anaular rates. * transformation of the accelerations to the earth referenced coordinate system, * computation of the velocities by single integration, * computation of the actual position by double integration.
5.1.1 Systematic IMU Sensor Errors

-50

-60

The inertial sensors may produce an output signal, although the IMU system is actually not moved. Fariell [61 already showed, that an accelerometer bias ab leads to a quadratic phase shift and a cubic phase shift arises from a gyro drift wd. The quadratic phase error m , ( t )
-0..

- 70 - 0.6

is denoted as (53)

-0.2

0.2

0.4

PQ(t)

4n 1 i ; '2 ,ab.tz

and at the end of the syntheiic aperture, where t=Tl2 we get Fig. 9 Simulated C-band impulse response, corre:jponding to Fig. 7 with an acceptable mainlobe shift, loss of resolution and ISLR.
50

m,(T/2)

1 7.7
2 h

.ab.TZ .

(54)

m n
I

50

0
n

10

30

.:

2 0 -

10.

0 IC-3

' I I
~

Taking the E-SAR parameters for L-band, tAe most critical case, we obtain for a phase error of mQ(i"12) = 7112, a total accelerometer bias of ab = 104.6 pg. The cubic phase shj.ft m , ( t ) is qiven by
=

* , ( t )

4 n .1 h 6

' g . 0 .t3

'

I551

where g is the gravity of the earth. With t=T/2 we get I .n .g.ad.T3 . O,(T/2) = (561 12 h

iL

i c i x
1

0.01

0.1

For an acceptable cubic phase error Bc(T/S) = n I 8 , we obtain a total gyro drift of od = 2 . 1 6 " I h for L-band parameters. ra5.1.2 Errors from Digital Data Processing
In addition to t h e phase errors induced by the aircraft motion or by the inertial sensors, further errors will occur due to the imperfect s o l u tion of the motion compensation algorithms and of digital data processing artefacts.

Frequency [Hz]

Fig. 10 Required transfer fmction of a motion compensation system for the E- SAR.

3-9

Using a digital phase shifter, the amplitude of the phase error 9(t) can be adjusted only in discrete steps, due to the number of available bits. The variance cn2, respectively the mean power of the quantisation noise is
(in2 =

flight-path can be performed online in the sensor SAR processor. Thereafter the data can be processed as if the platform moves in a straight line at constant altitude and parallel to the swath

or off-line in the

d s ' 12 '

~rom the flight-path the following parameters are

(571 derived.

where A s is the quantisation step width Quantizing 3609 with N bits, we get
dS =

1581

2% 2N

The true forward velocity is used t o correct the range independent phase errors by ensuring the along track samples are equidistant. This also corrects the image geometrically along track. With on-line correction, equidistant sampling can best be achieved by controlling the PRF of the radar, i.e. the PRF is directly slaved to the forward velocity. Off-line, the same effect is obtained by interpolating the data in the along-track dimension.

and finally, from Eqs. ( 4 0 ) and (511, the ISLR, induced by the quantisation noise, becomes

5.1.3 Delayed Phase Adjusting

Within an online motion compensation System, utilising an inertial strapdown system, the computation time will cause a delayed phase correction. Thus a residual, uncorrected phase error will remain, causing an increase of the ISLR. Considering a phase error m ( t 1 = 9ocos(2nfotl, which will be corrected after a time delay Atd, a residual, uncompensated phase error 9reslt) will remain. B , , , ( t )
= 9~cos~2nf~tl-Bncos[2nf,(t-Atd]]
=

The range delay of the samples is needed to ensure the range cells are correctly aligned. Online this is achieved by adjusting the delay of the digitised echo window. Off-line the range cells are shifted in the computer memory. The phase of the data in each range sample is corrected according to the true slant range to eliminate phase errors. on-line this can be achieved with a digital phase shifter. Off-line the data can be corrected by shifting the phase mathematically

(60)

2~mosin(nf,Atd1.sin[2nfo(t-Atd/211 (611

5.1.6 Example of Motion Correction using an M IU Sensor

The residual signal is sinusoidal too, with the phase error frequency f n . Its amplitude depends on t h e time delay Atd, the frequency io and the amplitude of the original signal 90. Considering all frequencies and amplitudes, defined by: G9(f), the ISLR can be evaluated using Eq. ( 4 0 ) as:

The image in Fig. 1 was taken with the DLR airborne SAR equipped with a LITEF LTR-81 heading and altitude reference system supported by the inertial reference system of the aircraft. The aircraft was deliberately caused to yaw i 2' causing the motion errors already described. From the recorded flight path data, the forward velocity (Fig. 11) and phase errors (Fig. 121 were derived.

ISLR

4.J' sinZ(PfAtd1.G9(f)df
1 IT -.
~

162)

5 . 1 . 4 Sampling and Holding the Detected Motion

Within an on-line motion compensation system, the phase shifter is adjusted in time intervals Ats. This is equivalent to sampling and holding the phase error. Considering a phase error Q(t)= Bocos(2nf,t), the power of the residual, uncompensated phase error, L,(tj within the interval Ats is approximately At, L,(t)
i

1 ~(902nfotlzdt = 4 1 A t : R . O o . f o ) ' At3

0

(63)

Thus the ISLR, caused by sampling and holding the detected motion, becomes Fig, 11 Forward velocity variation for the example SAR image. The flight data were used to correct the SAR data and the resulting image i s shown in Fig. 13. The image is now well focused and the geometric distortion (runway and taxiways) has been corrected.

5.1.5

Flight Path Correction

Once the flight path has been determined, it has to be taken into account during the processing of the SAR data to ensure an ""distorted, well focused image. The correction for the errors in the

ciple of the method is based on the misregistration of the looks if the Doppler rate is incorrect. An error in the Doppler represents a quadratic phase error. In Section 4.1 we have Seen that the shift of the impulse response is proportional to the linear phase error. Assuming a quadratic phase error (See E,q. 2 4 )
oit)
=

2 a(O) t2 . 2 %

(601

Differentiating we obtain:
4n ict) = a(~) t . %
(611

From the earlier analysis it can be shown that the

relative position of the impulse responae becomes:

RO

= V

t a(O) .

(621

For two looks spaced At in time, the misregistration between the l o o k s in meters is given by:
Fig. 12 Phase error profile for the example SAR image.
AS =

v 4t

RO

a(0) .

(63)

. .

Hence, the coefficient of the quadratic .?haSe er101 a(0) can be determined from the misr,sgistration of the l o o k s . Note that R,,, V and A: are all known. The misregistration of the l.ooks can be determined by correlating them to determine the spacing. Knowing the quadratic phase error term, the Doppler rate can be corrected. This method works well where quadratic phase errors dominate. This is the case where the platform flight path i s stable but the velocity is; not known, e.g. Satellite sensors. Higher order terms cannot be derived. The method requires a preliminary processing of the image before correction can be applied.
5.2.1.2

Contrast Optimisation

This method is based on the relationship between the contrast in the image and the Doppler rate used for processing. The image (or parts of the image) is processed and the contrast is msasured by calculating the ratio of the deviationlmean intensity of the image. It can be shown 171 that this ratio is directly related to the spa.:ial resolution. This is a trial and error method for finding the Optimum Doppler rate for azimuth proZessinq and, hence, requires considerable computer effort.
5 . 2 . 2 Reflectivity Displacement Method (FOM)

Fig. 13 The Scene from Fig. 1 corrected for motion errors.

5 . 2 Deriving the Flight Path from the Radar Data 5 . 2 . 1 Autofocus Methods

Autofocu~ is a process whereby the SAR processor attempts to derive the data necessary for accurate focusing of the image from the radar data. The following methods describe typical approaches.
5 . 2 . 1 . 1 Look Misregistration Method

This method 18, 91 of motion compensation using the SAR data is the most comprehensive and in principle is capable of correcting both high and low-frequency errors with impressive results. The technique analyses the frequency spectrum of the SAR data after range compression to derive the flight path of the sensor. Considering the deviation from the optimum flight path in terms of a velocity component in line-of-sight VLos, rather than displacements, we can write:

!uith this method the SAR data are processed to obtain a two-look image. The azimuth processing is performed witn an assuned Doppler rate. The prin--

2 Vit)
fDoppler =

0,

2.VLosItl

7.

(641

3-1 I

Assuming the forward velocity V(t) is constant during the period under consideration, we can derive the following expression for the frequency shift bet,ween two adjacent power spectra spaced At in time:

where r is the range of the selected range samples. we see the frequency shift can be separated into two components, one dependent on the forward velocity and one dependent on the acceleration in line of sight. The spectrum of the SAR data in azimuth is a convolution of the antenna pattern and the ground reflectivity function. If the antenna pattern is broad enough not to influence the ground reflectivity function or, if it can be corrected f o r , the frequency shift can be determined by correlating the two adjacent azimuth spectra. If At is much smaller than the azimuth illumination time, the frequency shift AfDopDler can be determined very accurately.

FREQUENCY SHIFT

(Hz)

Fig. 14 shows two such spectra far the E-SAR airborne sensor operating in L-band with a time offset At Of 1.075 5 .

Fig. 15 Correlation of the azimuth power spectra form Fig. 14.

Fig, 16 Power spectral density of the forward velocity and acceleration in LOS for a DO 228 aircraft.

Fig. 14 Two adjacent azimuth power spectra taken with a time offset bt = 1.075 s. Fig. 15 shows the corresponding correlation function yielding a , , , f A of 12.1 Hz. The next step in the RDM method is to consider the power spectral density of the forward velocity and the acceleration in LOS. Fig. 16 shows the PSD for the E-SAR aircraft, a twin-engined Dornier 228. It can be seen that the velocity variations are mainly low-frequency whereas the accelerations in LOS are mainly high frequency. This is true for most airborne platforms where the forward momentum is much higher than in other directions. This phenomenon enables the forward velocity and acceleration influences to be separated in Eq. (65) by filtering the values of . , f A For the above examples, the filter cut-off frequencies are at 0.05 and 0.1 Hz. This procedure enables the forward velocity and acceleration to be determined. From these parameters, the necessary corrections of the SAR data can be carried out. Fig. 17 shows an example with Fig. 17 Image taken in C-band with the E-SAR Sensor with 12 % velocity variations. The upper image is uncorrected and the lower one corrected using the RDM method.

3-12

the E-SA? sensor operating :n C-band with 12 % YBlocity variations. The upper image is processed with constant velocity and the lower one with the processirg parameters corrected using the RDM method. The defocusing visible in the upper image is eliminated after motion compensation.
6 . CONCLUSIONS

I21

S. BuckreuR: Motion E r r o r s in a n Airborne Synthetic Aperture Radar-System. ETT-Journal Special Issue: "ETT FOCUS on SAR", Lfd. Nr. 2, Bd. 6, 1991, pp.
655-664.

131

The influence of motion errCrS on SAR images has been discussed and the requirements for a compen-. satior Systen derived. Several met:hods of comoecsation have been discus-. . sed, the use of inertial pktiorms and the RDM method showing the best resclts.
~~ ~

(41

as an alternative to using inertial platforms, it

G.E. Haslam and 8. Rei.d: Motion Sensing Requirements for Synthetic Aperture Radar. ProC. IEEE Conf. Toronto, 1983, VOl.1, pp.
126-131.

would be attractive to use the satellite navigation System GPS, being a muck cheaper Solution. A GPS receiver is used in the E-SAR airborne SAR to provide absolute position of the aircraft. HOwever, due to the deliberate errors introduced into the GPS data by the GPS operator (selective availability), the data are cot accurate enough for motion compensation of the SAR data and they are onlv used for annotation U U I D O S ~ S . Note. that.

I51

F.K. Li, D.N. Held, J. Curlander, and c . w,,: .. Doppler Parameter Estimation f 3 r Spaceborne Synthetic Aperture Radars. IEEE Trans. Geosci. Remote Sensing, VoI. GE-23, 1985, pp. 47-56.
~~

I61

provide an attractive method f o r measuring the ilight path and is worth investigation.
An additional method of improving image quality

J.L. Farrell: Strapdown INS Requirements Imposed by SAR. Conf. Dayton, OH, USA. 21-25 May 1984. Proceedinss of the IEEE 1984 National Aerospace and Electronics Conference. Naecon 1984 (IEEE Cat. NO. 84CH2029-7)
D. Blacknell. I.A. Ward and A . Freeman: Motion Compensation and Geometric Cistortion in Airborne SAR Imagery. Progress in Imaging Sensors, ISPRS Symp., Stuttgart, 1 - 5 Sept. 1386.
~ ~~~ ~ ~~~ ~~~ ~~ ~~ ~

I 7 1

was suggested by Chan,the so-called tuned auto :ompersator [lo]. This method analyses strong point targets to identify paired echoes. From these the high-frequency phase f-rmis can be derived. The nethod requires m c h trial and error and can, rherefore, only be regarded a s an acgmentation of other techniques. 'Table 5 summarises the capabilities of the variou!; methods.

I81

J.A. Moreira: A New Method of Aircraft Motion E r r o r E X traction from Radar Raw Data for Real Time SAR Motion compensation. Proc., IGARSS Symposium, Vancouver, Canada, 1989. J. Moreira: Motion COmOenSation SAX-Processino Vacilit y at DLR. EARSeL '90, Touiouse, France, 5-8 June 1990.
Y. Chan:

Method

1 I

Forward velocity

Lo$-frequ. errors in LOS yes

1 I

I91 High-frequ. errors in LOS yes

1101

Inexial

neasurenent
LOOk

yes

misregistratior. Contrast optimilsacion

either or

no

A Tuned Auto-compensator for Residual Antenna Motion in Synthetic Aperture Radar Systems. IEEE Trans. Geosci. Renote Sensing, Vol. GE-24, lo". 1986, pp. 1025-1027. [I11 J.C. Kirk: Motion Compensation F o r Synthetic Aperture Radar. IEEE Transactions on Aerospace and E:lectronic Systems, Vo 1. AES-11, No.3 (May 1975) pp. 338-348. T.A. Kennedy: Strapdown Inertial Measurement Units f o r Motion Compensation for Synthetic AFSertiire Radars Presented at the IEEE 1988 National Radar Conference 0885-8985/88/1000-0032,iEEE AES Magazine, Oct. 1988.
~~

either or

no

I121

? a b l e 5 Comparison o f the capabilities of the va-

rious motion compensation methods.

7 . REFERENCES

[I31

I11

R. Horn: C-Band SAR Results Obcained by an Experimental Airborne SAR S c n s o r . IGARSS Symposium, Vancouver, Canada, 1989 July 10-44.

JOGS D.K. : Inertialnaviqation in der Strapdown Tech nik. Special Issue. Ortung und Navigation.
Journal 2/1983.

4-1

THE REAL APERTURE ANTENNA OF SAR, A KEY ELEMENT FOR PERFORMANCE

H. Ottl Deutsche Forschungsanstalt fiir Luft- und Raumfahrt e.V Institut fiir Hochfrequenztechnik 8031 Oberpfaffenhofen Germany
by

SUMMARY
F O ~ a SAR system flying on an airborne or spaceborne platform, the real antenna must be designed in such a way so as to avoid ambiguities and achieve the envisaged resolution.

Although a SAR is, with respect to geometric azimuth resolution, independent of its distance from a target, the ground range resolution depends on the incidence angle and, of course, on the bandwidth dependent slant range resolution. The antenna size and its half power beam width (HPBW) in azimuth and elevation define its azimuth resolution and, for a given off-nadir angle and chosen altitude, the swath width. The minimum antenna size, measured in wavelengths, depends on the altitude, velocity of platform and chosen off-nadir angle. In real antenna design, the aperture size will be somewhat larger in order to allow far amplitude taper (at least in elevation), for electronic beam steering and possibly for beam shaping. This paper explains the interdependence of antenna parameters with SAR system performance.
1. INTRODUCTION

cell
Fig. 1 Coordinate system and illumination geometry.

Spaceborne imaging radars and, in many cases, airborne imaging radars require enormous antenna di-

mensions (measured in wavelengths) if real aperture systems are envisaged for ground resolutions of several metres or less. Such large apertures can neither be launched into space nor installed on high flying aircraft. Coherent RF-technology all o w s the composition of a synthetic aperture length in flight direction, which defines the azimuth resolution, while in the orthogonal direction (in range) resolution is achieved by the short radar pulse length or, if chirp modulated, by the compressed pulse length. A schematic view of SAR'S illumination geometry is illustrated in Fig. 1. Each target within the footprint of the real aperture antenna will be illuminated during the time needed to fly along one synthetic aperture length, which corresponds approximately to the length of the footprint. The data takes are characterized by time, Doppler frequency and distance between the flying ( o r orbiting) real aperture antenna and the target (Fig. 2). These data are the input for a SAR processor which generates the SAR images. TWO steps are generally performed by a SAR processor. Firstly, the data are range compressed, that means correlated with the range reference function. Accordingly, one could imagine that this step produces a s many fan beams as range bins are foreseen. The HPBW of one such fan beam in range is dependent on the slant range resolution and the incidence angle; the HPBW in azimuth is still approximately equal to the HPBW of the real aperture antenna. Secondly, azimuth compression is performed, which means that the data are correlated with

in relationship to the synthetic aperture length, resulting in a ground resolution of approximately half the real antenna lensth for a sinqle l o o k image
It should be not.ed that. the illustration of res~. ~. olution c e l l s created by multiple beams across range, which are sweeping along the azimuth direction, is only one possibility of describing SAR. It does not include illumination geometry, that means changing incidence angle direction and value during a flight along one synthetic aperture length.
~~~~~ ~~ ~~~ ~~

2. GEOMETRIC RESOLUTION, ANTENNA APERTURE AND

SWATH WIDTH It is important to understand the improved along track resolution of a synthetic aperture antenna in comparison to a real aperture antenna of the same length 1 and same azimuth aperture taper factor ahr. To explain this, we assume an antenna carrying platform passing a target. In the case of a real antenna, the target will he fully iiluminated if the Doppler frequency is zero. The antenna beam. oriainatina from the Dhase center of the ant e n n a , is.orthog&al to the' flight direction (Fig. 3). Resolution is given by:

Hatched Area: Antenna Footprint

Constant Doppler Hyperbolae

Fig. 2 Position of resolution cell within a SAR image is determined by slant range, Doppler frequency and time.

t
m
c
C

flight direction

a, * C m

Doppler impact due t o antenna velocity

2 2
4L

a,

m
2
a,

onstant frequency circle

>

> , a,
L

Tiq. 3 Very long real aperture antenna illuminates a target The Doppler frequency shift of the backscattered signal corresponds to the one way case.

4-3

flight direction

:onstant frequency

Doppler impact due t o t w o way phenomena of synthetic aperture antenna

'.real

aperture antenna flying along the synthetic aperture

Fig. 4 Synthetic aperture antenna illuminates a target. The signal from the real aperture antenna propagating to the target undergoes already a Doppler shift; the backscattered signal is "Doppler shifted" again resulting in twice the value of the one way case. H
=

19 = off-nadir angle

altitude of antenna above ground, incidence angle),

( 5

the HPBW of the synthetic aperture

Phs

ahs

A
I

(4)

azimuth HPBW R h r
h = wavelength

ahT

7 ,

(21

aha

azimuth aperture taper factor,

The signal reflected from the target reaches the different parts of the real aperture antenna with a phase distribution corresponding to the one way distance. The Doppler fregllency occurring at the different parts of the moving "long, real aperture" antenna represent also the one way case,

L = length of synthetic aperture. The HPBW Of the real aperture antenna which flies along the synthetic aperture (the antenna is much smaller than i n the above mentioned comparison) is used for the definition Of the antenna footprint. The length of the synthetic aperture is given by the length of the footprint and therefore,
L =

In the case of a synthetic aperture antenna, a

small signal source (phase centre) flies along the synthetic aperture. We have, therefore, a two way case for the phase difference along the synthetic aperture (Fig. 4 i , which means that phase differences (or Doppler frequency) are twice the values achieved in the real aperture case of the same length. This phenomenon causes a phase change of 180' twice as fast as in the case of the real aperture antenna, meaning that the nulls of the synthetic aperture antenna main beam have a spacing which is half of the real aperture antenna. Therefore, a synthetic aperture antenna has an along track resolution (HPBW of SARI which is approximately twice as good as for the case of a theoretical, real aperture antenna of the same length. The along track (azimuth1 resolution for a single l o o k image is given by

rar

A H ahr 1 COS19

'

(5)

Substituting L from ( 4 1 and leads to

Bh9

from ( 3 ) in ( 5 )

which is the well known approximation for finest h . = ah=, rap corresalong track resolution. For a ponds to half the length 1 of the real aperture antenna. However, it should be noted that in many cases the amplitude distribution in azimuth Of the real aperture antenna is not tapered, resulting in ahr = 0 . 8 8 , while the sidelobe suppression achieved by the reference function of the synthetic aperture azimuth pattern (more than 40 dB below mainlobe) leads to ahs = 1.3. Therefore, it is a realistic assumption to expect an rap 0.8 1.

4-4

The swath width is a parameter which is dependent on along track resolution rap, platform velocity u and off-nadir angle 8. a s will be shown below. Flying along a synthetic aperture length cover!;, in the case of SAR, twice the normally encountered Doppler frequency shift f , . For zero offset processing (using in-phase and quadrature channels for positive and negative Doppler identification! only fo per channel is used. From geometric considerations (see also Fig. 1 and 21, f, is defined by a ,
=

, H h sin8 2 PRF a h >

c

cos28

elevation taper factor.

U Since PRF 2 f , and f D = 2 ahr i, the phpical dimensions of the r e a l aperture antenna are given by

. 2 FOI high flying platforms (satellites!, small sild

f ,
=

2" fD -sin- h

Phr

(7)

Bhr

is

In many cases, ahr 5 0.88 (constant amplj.tude) while a , will be approximately 1.3 to a l l o w suppression of the sidelobes in the elevation plane. Furthermore, range ambiguities should a h a not occur within the main lobe at all (main lobe width is approximately 2 to 3 times larger thar pv!.
Therefore, empirical estimations of the real antenna dimensions could be based on

2u i ;

Bhr

2ahr U i .

(81

The Nyquist sampling theorem requires a pulse 1,spetition frequency (PRF! equal to 01 higher than the f, which is used in SAR processing per channel. Generally, a PRF of 1 to 1.3 times f , will be chosen. Two radar pulse returns must not be received simultaneously from the illuminated footprint iinstantaneous field of view. IFOVI of the r e a l aperture antenna. Looking at Fig. 5, we notice that the Slant range difference between the edges of the illuminated swath width (corresponding to B , I is given by the slant swath width
~ ~ ~~~~

f = SAR frequency.

It can be concluded that the estimated minimum dimensions of the real antenna enforce a trade-off between resolution and swath width. High resolution can only be achieved with a small swatn width; increasing the swath width reduces resolurion (Fig. 6 and Fig. 7 ) .
3. D I R E C T I V I T Y , ? O L A R I Z A -

TION The remarks given below refer to spaceborrie antennae, which are more difficult to realize due to their size and the associated, complicated, feeding network. These antenna, of several square metres area, are usually designed as flat (foldable) arrays. The directivity of such antennae is proportional to their area but depends of course on the chosen taper. Compared to constant amplitude distribution with a sidelobe Level of about -13 dB, a good sidelobe suppression (in elevation! reduces directivity easily by 1 to 1.5 dB, which must be compensated by an increased antenna area. For a rectangular, flat array antenna of length 1 and height h, the directivity D is given by

Fig, 5 Distance between two pulses must be > 2R, in order to avoid range ambiguities occuring within the IFOV. R, = slant range swath width.
H

B, sind
C O S ' 8
'

R ,

if an equal phase and amplitude distributimm is assumed. In this case, we cal: 1 h the effective (I! antenna area
Aeff
=

The distance between two radar pulses must be > 2 8 . . to avoid the reflected Dulse from the far raige edge overlapping with the reflected pulse from the near range. Therefore,
j
~ ~ ~~~

1h

(15)

For any deviation from constant phase and amplitude distribution, the effective antenna area will be smaller than the real aperture. The relationship between the two areas is called aperture effici.ency q a , It reduces the directivity proportionally and therefore,

c = speed of light. Substituting a , for p, gives the antenna height h

4-5

Fig. 6 High resolution corresponds to long synthetic aperture and small swath width (left side of fiourel while a wide swath width enforces a short synthetic aperture and l o w resolut i o n ?right side of figure)

The antenna gain G is smaller than the pattern directivity D due to internal antenna losses. This thermal efficiency, 01 radiation efficiency, is defined by

chirp bandwidth

The ground resolution across track is accordingly given by

ry

-C

28

sin6

'

(19)

For a passive microstrip antenna of Several square metres or a waveguide array of several metres length (spaceborne S R R antenna), '11 might be 0.5 or even lower (depending on the frequency).
Considering these high one way losses associated with such arrays, one solution would be to use an active array with distributed transmitireceive modules ( T / R modules). The one way loss of such systems is much lower and leads to q l = 0.8 - 0 . 9 . For lower microwave frequencies, such as L-band and S-band, light weight and volume considerations lead to microstrip arrays while for higher frequencies, such as C-band and X-band, a waveguide array might be favoured for its lower losses. The bandwidth of the antenna should be broad enough to easily achieve the desired range resolution (chirp bandwidth). Frequencies (signals) outside the useful bandwidth should be rejected or, at least, strongly attenuated.

Dual polarization capability has been increasingly requested in recent years in order to allow for full polarimetric SAR operation. For instance, a horizontally polarized signal pulse will be transmitted and the back scattered signal will be received in the same polarization and simultaneously in vertical polarization; then a vertically polarized signal pulse will be transmitted and the back scattered signal will be received in the same polarization and simultaneously in horizontal polarization. This Sequence will be continuously repeated. Assuming that the time between 2 transmitted pulses of different polarizations is negligible, we achieve the complete polarimetric matrix of backscattered signals (scattering matrix). It contains much more information about radar illuminated target classes than the copolarized back scattered signal of one polarization.

However, it should be noted that polarimetric SAR operation ( a l s o named "quad-pol" operation) requires doubling the PRF for a SAR with 2 receiver Typically, microstrip and wave guide arrays attain channels. This results in reducing the swath width a relative bandwidth of 3% to 5%. That means, for to half of the single polarisation case and, accora high slant range resolution rR it is advisable dingly, doubling the antenna height h for proper to use higher frequency bands for the SAR. For exrange ambiguity suppression (exception: airborne , ' ample, a slant range resolution of 0,5 m correspond: SAR is usually not affected by range ambiguity approximately to a chirp bandwidth of 300 MHz (whict problems, due to the low altitude and Smaller is 3% of 10 GHz) . swath width).

'

Slant range resolution rR is defined by

The polarization ratio between copolarized and cross-polarized antenna pattern should exceed 30 da within the HPBWs in azimuth and elevation. It

Ic ._

3 200
E m

t i 0
0

i7
f

k.100
0 c

a , 3 U

r,
I
I

E
._ 20.

2
X

10 single look along track resolution

! 00 -

[ml

Fig. 7 Azimuth resolution rap influences proportionally the swath with; the parameter is incidence angle is very difficult to reach this performance for the large arrays needed for spacecraft. Screening the feeding network and accurate mechanical alignment of single radiating elements are a "must" for a good polarization ratio. The request f o r 30 dB is based on the facts that cross-polarization occ u r s in many Cases in the range of 5 dB to 20 dB below the copolarized back Scattered signal and that the performance of a spiceborne SAR sensor (noise floor1 seldom allows a signal to noise raLIO S , of more than 20 d B .
& .

ties and therefore reduce the revisit time intervals, - to collect radar data under differen: imidence angles. The off-nadir angle range used in various ::tudies extends from 15' ( 2 0 " ) to 50" ( 6 0 " ) covering a data take opportunity width Of about 200 km to 600 kni on the ground (depending on the orbital altitude, e.g. shuttle o r satellite). For a n array antenna with discributed TIR nadules, only phase shifters need to be added in order to produce an active phased array antenna. The number of TIR modules and the spacing of the radiators lor groups of radiators1 fed by a single T I R module depends on the chosen phase controlled angle range.

4. ACTIVE PHASED ARRAY For spaceborne remote sensing, a fast off-nadir angle change capability is needed

to allow scan-SAR operation, - to increase the number of data take opportuni-

look steering mode increases the overall synthetic aperture (for multi look images resolution rap). Only spot-wise data take.? are possible in this mode.

5000

d = distance between phase centres of single radiators (or groups of radiators) fed by neighhouring TIR modules,
~ u 9 =

half the desired phase controlled angle range.

2000

S-Band C-Band X-Band Ku-Band

1000

m
4

- 500 E .i

i"

zoo

In case of At9 = + Z O O , the spacing must be less than or equal to 0.75 h . Spotlight or azimuth look steering modes require a beam which is electronically controlled by phase shifters within a small (squint) angle range. These modes allow high resolution observations of small areas (spots) by sacrificing the observational possibilities before and after each spot (Fig. 8 1 . The required squint angle A q far Such operations is seldom larger than +I' to 12'. The spot size is frequency dependent because of the IFOV, which increases proportionally with the wavelength (Fig. 91.
Equation (20) will not be applied for squint angles of 1' to 2" in order to keep the number of T I R modules and its phase shifters small. The radiator groups in azimuth fed by a single TIR module have a high directivity, which is used to suppress the grating lobes. There are two causes of gain reduction due to electronic beam steering. Any angle deviating from the mechanical boresight angle reduces the antenna area proportionally to cosAt9 or cosdq. Furthermore, each single radiator has a pattern which usually has its maximum at the (mechanical) boresight angle. The reduced gain at other angles must be multiplied with the array factor (which represents the far field of point sources arranged in the geometric layout of our antennal. For instance, the gain loss of a well designed active phased array at Au9 = 20" will be approximately 0.5 dB.
5 , CONCLUSIONS

100

50

10

20

30

40

50

60

70

80

target incidence angle

Fig. 9 Illuminated area for spot lightmode andlor azimuth look steering mode increases approximately proportionally with the wavelength. The generation of several '"main beams", which are called grating lobes, OCCULS if the spacing is too large. I n such a case, several angles exist with e q u a l phase distribution; it is at these angles that the grating lobes occur. Suppression of grating lobes is accomplished by satisfying

The real aperture antenna is a key element performance. It defines andlor influences

for

SAR

1-3

the along track resolution rap by its length 1, the swath width by its height h, the across track reso1utj.m rR by its bandwidth B, the range ambiguity suppression by its amplitJde distribution in elevation (taper factor a w l , - the required onbaard power by antenna losses [thermal efficiency qi), - the polarinetric performance by its polarization decoupling, - the data take opportunity, scan SAR and spotlight made SAR by its electronic beam steering.
It is, therefore, worthwhile investing efforts j n a good antenna design and accompanying technologies.
6 . REFERENCES

This paper is a lecture based on principles, which have been published several times by other authors and by recent studies, in which the author was involved. The references belcw were used as sources, b u t due to the general aspect of this lecture, no special reference is given in the text.
1. Skolnik. M. I., '"Radar Handbook", McGraw-Hill Book Company Inc., 1970.
2. 3.

Barton, D . K . and ward, H . R . , "Handbook of Radar Measurement', Artech Home, Inc., 1984. MeinkeIGundlach, "Taschenbuch der Hochfrequenztechnik", Dritte Auflage, Springer Verlag, 1 9 6 8 , Kapitel H. Antennen.
Jasik, H., "Antenna Engineering Handbook", McGrawHill Book Company Inc., 1 9 6 1 .

4. 5.

tilaby, F.T., Moore, R.K. and Fung, A.K., Microwave Remote Sensing", V c l . I and 11, AddisonWesley Publishing Company, 1 9 8 1 and 1 9 8 2 . tilaby, F.T. and Dobson, M.C., "Handbook of Radar Scattering Statisticj for Terrain", Artech House, Inc., 1 9 8 9 .

6.

7.

Colwell, R.N., Simonett, D.S. and Ulaby, F.T., "Manual of Remote Sensing", Second Edition, Vo1. I, The Sheridan Press, 1983, Chapter 9 and lo.
Jordan, R.L., Huneycutt, B.L. and Werner, M . , "The SIR-CIX-SAR Synthetj.c Aperture Radar System", ?roc. IEEE, V o 1 . 79,No. 6, June 1991, pp. 827-838. Jatsch, W . , Langer, E., Ottl, H. and Zeller, K.H., "Concept of an X-Band Sycthetic Aperture Radar for Earth Observing Satellites", JEWA, V o l . 4, No. 4, 1990, pp. 325-340. Synthetic Aperture Radar for EOS SAR", under 1'"blication in ?roc. of I S Y Conference, Munich, March 29 - April 4 , 1 9 9 2 .

8.

9.

1 0 . Otil, H. and Wahl, M., "X-EOS, a Multi-Mode X-Band

5- 1

Polarization Effects and Multipolarization SAR

A. Freeman

Jet Propulsion Laboratory Pasadena, CA 9 I109 USA

Introduction Imaging radar polarimetry has excited much attention in t h e literature over the past f e w years ([11-[61). since NASA/JPL f i r s t d e m o n s t r a t e d a successful SAR polarimeter system in 1985 [I]. That system was known as t h e CV-990 L-Band radar; NASAlJPL now have a fully operational. t h r e e - f r e q u e n c y polarimetric SAR flying on a DC-8 platform. which h a s taken part in m a n y science d a t a acquisition campaigns since 1 9 8 8 [71. S e v e r a l o t h e r institutions h a v e r a d a r p o l a r i m e t e r systems which are. operational or under development. including t h e University of Michigan. MIT/Lincoln laboratories, t h e Environmental Research Institute of Michigan IERIM) 181. t h e Canada Center for Remote Sensing (CCRS), FEL-TNO in t h e Netherlands and t h e Technical University of D e n m a r k . NASA/JPL a r e c u r r e n t l y completing t h e construction of a t w o f r e q u e n c y , fully polarimetric SAR system which will fly on the Space Shuttle in 1 9 9 3 / 1 9 9 4 IO]. These systems generally t r a n s m i t and receive horizontally I H ) and vertically ( V ) linearly polarized electromagnetic fields. Imaging r a d a r polarimeters a r e usually implemented using a Synthetic Aperture Radar ISAR) approach t o give a high resolution image in two dimensions: range and azimuth. For each pixel in the image a polarimetric SAR gives sufficient information t o characterize t h e polarimetric scattering properties of t h e imaged area (or target) as seen b y t h e radar. Using a polarimetric SAR system ils opposed to a single-polarization SAR system provides significantly more information about t h e t a r g e t scattering mechanisms and allows b e t t e r discrimination between different types of surfaces. I n these notes a brief overview of SAR polarimetry i s offered. The notes are intended as a text t o accompany a lecture on SAR polarimetry as part of a n AGARD-NATO course. For a more in-depth treatment, t h e interested reader is referred to the recenl review paper by Zebker and v a n Zyl 1101 and the testbook on 'Radar Polarimetry for Geoscience APPlications'. edited by Ulaby and Elachi [ I I ] . For a discussion of inverse methods t o determine scattering phenomena f o r polarimetric radar data t h e interested reader 1s referred to the testbook on 'Inverse Methods in Electromagnetic Scattering', e d i t e d b y Boerner 1121. For b r e v i t y , a discussion of t h e development history of polarimetric radar has not been included in t h e s e notes. nor h a s a comprehensive b i b l i o g r a p h y . T h e r e a d e r i n t e r e s t e d in m o r e information o n these topics would d o well to begin by examining references [ I O ] - [ 121. Covered in t h e notes are: the polarization properties of electromagnetic waves; the concepts of radar scattering a n d m e a s u r i n g r a d a r b a c k s c a t t e r w i t h a SAR: polarization synthesis: scattering matrix. Stokes matrix and covariance matrix representations of Polarimetric SAR data; polarization Signature plots: design and calibration of Polarimetric SAR systems; polarization filtering for target detection: fitting a simple model t o polarimetric SAR measurements of naturally occurring features; and supervised classification of polarimetric SAR data.

Polarization of Electromagnetic Waves The polarization of a n y electromagnetic w a v e can be characterized b y t h e ellipticity angle. X , and t h e of the polarization ellipse, shown orientation angle, in Figure I . The intensity of the wave is represented b y the parameter I,, where

w,

I, = a i + a?
Some commonly occurring polarization states are: Vertical (linear) - (w = 0, % = Oo) Horizontal (linear) - (w = 90. = 0') Right-hand circular - ( x - 4 5 O ) Left-hand circular 1% = 4S0)
~ ~

11)

For t h e t w o circular polarization States a b o v e , t h e orientation angle. is unspecified. Another useful w a y of c h a r a c t e r i z i n g t h e polarization S t a t e of a n electromagnetic wave is the Stokes vector.

w,

b
I ,cos 2~ cos 2%
I, sin 2~ cos 2%
(21

w h e r e I,. 0,U and V a r e t h e four Stokes parameters. w h i c h all h a v e t h e s a m e dimension. T h e Stokes

Polarization Ellipse
Fig. 1 Polarization ellipse

5-2 parameters are related via:

1 : = Qz+ Uz + Vz

(3)

Measuring Radar Backscatter w l t h a SAR system

A SAR System is simply a high-resolution active

so that only three of them are actually independent.

The polarization state of a n electromagnetic w a v e can be represented as a (unique) mapping t o a point on a sphere of radius I,. called the Poincare' sphere. w h e r e 0. U. and V are the Cartesian coordinates of the point. The angles 2% and 2Iy define the latitude and longitude of the point in the spherical coordinate system. The four common polarization states listed above are shown as points on the Poincare' Sphere in Figure 2 . L-circular

microwave sensor, capable of measuring the (real or complex) radar reflectivity of a surface. SAR's usually operate as monostatic radars. mounted on a moving platform, w i t h a n a n t e n n a looking out a n d d o w n towards one (or both) sides relative t o i.he platform motion vector. Most SAR systems measure t h e radar b a c k s c a t t e r using l i n e a r l y polarized a_ntennas, typically w i t h polarizations given b y the h and ? coordinates in Figure 3. I n its ideal realization, t h e r e s u l t i n g SAR i m a g e should just r e p r e s e n t t h e s c a t t e r i n g m a t r i x e l e m e n t . S p q , in c o m p l e x representation. or t h e r a d a r cross-section, a p q . in i n t e n s i t y . f o r t h e receive ( 4 ) and t r a n s m i t ( p ) polarizations of t h e r a d a r . The scattering matrix determines the relationship between t h e w a v e incident on the scatterer and the scattered wave. After I111. the scattering matrix is defined via:

where

( )

is t h e electric field vector 01

the w a v e

incident on t h e scatterer,

is the electric field

Vertical

R-circular

(X = -450)
Fig. 2 Poincare' sphere

vector of the scattered wave. k o i s the w a v e number of t h e illuminating w a v e . and R t h e ( r a d i a l ) distance between the scatterer and the radar antenna. A fully Polarimetric SAR System wouid typically measure all four of the scattering matrix elements simultaneously (or near-simultaneously) as complex numbers far each Pixel within the area being imaged by the SAR. Polarization Synthesis Knowledge of t h e s c a t t e r i n g matrix allows t h e calculation of t h e backscatter mtensily, or radar crosssection. for a n y possible combination of transmit and receive a n t e n n a polarizations (e.g. left-hand circular t r a n s m i t a n d r i g h t - h a n d circular r e c e i v e ) . This procedure is called polarization synthesis. For a n y given r a d a r receive a n d t r a n s m i t poiariz.ation, t h e radar cross-section (RCS) can be calculated I 1 11 via:

Radar Scattering Measurements In Figure 3 t h e general geometry for measurements made b y a bistatic radar i s illustrated. The transmitting antenna transmits an electric field whose components are expressed in terms of a local Cartesian coordinate system (<, ?, 6 ) w i t h origin a t t h e t r a n s m i t t i n g antenna. We ;an define another Cartesian coordinate system ( x , y. z ) with origin a t the scatterer. The two coordinate systems are related v i a :

-,.

o N = 4 n ) @ SP f

(6)

, .

= sin

4; ii - cos $j y
8i Z

(4a)
$j

Y = -cos $; cos

- sin

, . cos 6 i y + sin 6i
ej

i '

(4b)

ii = -cos $j

sin 0; ii - sin $i sin

7 + cos ei Z
1s

t 'IC)
.

where s is the scattering matrix defined in ( I ) , and 4' , p t a r e polarization field vectors for t h e radar receive and t r a n s m i t polarizations. respectively. For linear polarizations (horizontal. h or vertical. v ) no1.e that t h e RCS is given by:

defined with origin a t the antenna which receives the electromagnetic w a v e which i s scattered b y t h e Scatterer. This coordinate system i s related to t h e scatterer coordinate system in a similar fashion t o the above, Substituting subscripts s for subscripts i in ( 4 ) . This choice of coordinate system; ensures t h a t t h e primed (%, ?, 6 ) and unprimed h', v', n ' ) coordinate systems coincide for a monostatic system. i.e. w h e n the receiving and transmitting antennas a r e a t t h e Same location. The radar scattering measurements made b y a monostatic radar s y s t e m a r e c e i e r r e d t o as r a d a r backscatter measurements.

A third coordinate system

(: h , -,). Y ' , n

In w h a t follows w e will c o n c e n t r a t e ,on linear polarizations, since most SAR systems just. measure linear polarizations. Expressions ( 5 ) and ( 6 ) above represent the quantities which are directly measurable by a SAR. The units for a p q a r e in meters squared. Both a p q and s p q are lunctions of spatial position I ( 2, ) in t h e Scatterer coordinate frame1 , time (1). viewing geometry ( e i , $I), a n d r a d a r w a v e l e n g t h ( h ) in addition to t h e Dolarizations of t h e t r a n s m i t t e d 'and r e c e i v e d

5-3

Transmit Antenna

, I
I

g;
\ \ \ \
I I
\

Receive Rt Antenna A:

. .

I .

I I

, ,
/

.&
n
-

-J

I I

L - -

# - -

Fig. 3 Coordinate systems for radar scattering.

electromae n e t i c w a v e s . If t h e r a d a r b a c k s c a t t e r measurements made b y t h e SAR are calibrated, each image pixel value should correspond t o a n accurate and precme measurement of one of these quantities. and b e r e p e a t a b l e , under t h e Same conditions, b y t h a t or another sensor. Stokes matrix a n d Covariance matrix representations I n the monostatic scattering case, it can be shown t h a l the reciprocity principle 1101, which dictates that

M 14 = 0.5Im

(sh', s h v )

+ 0.51~1 (Sh: s v v

M 22 = 0.25( Shh S h i

- 2 s h v s h : f sw s v : )

M 23 0.5Re (SA; s h v ) - 0.5Re M 24 = 0.5 Im

(sh; shv

(si: s v v
sw

)
)

) - 0.51m (Sh:

M 33 = 0.5 ( s h v SA: ) + 0.5Re (SA: M 34 = 0.5Im (SA;Sw )

sw

shv = s v h

(8)

is generally applicable. This simplifies some of t h e analysis of polarimetric SAR data and quantities derived f r o m i t . For e x a m p l e , a n o t h e r u s e f u l w a y of representing t h e polarimetric scattering information contained in t h e scattering matrix is t o construct a covariance matrix b e t w e e n t h e Scattering matrix elements. Without reciprocity the covariance matrix would have t o be a 4x4 matrix containing all possible cross- products b e t w e e n t h e f o u r scattering matrix elements. With reciprocity a 3x3 covariance matrix is sufficient. i.e..
Shh S h i shv
sw

M 44 = 0.5(Shv s :, ) - 0.5Re (Sh:

s ,

(10)

The f i r s t element. Mi I , in the Stokes matrix is often referred to as the total power. I t is related to two of the olher Stokes matrix elements via:
M11=M22 +M33

+Mu

(11)

Shh s h z s h v sh:

Shh Svv

sh;l
Shi

shv svv
s v v sv:

s m sh:

:I
s ,
SPY s; )
sw

(9)

Note that it is possible to recover the covariance matrix e l e m e n t s from t h e Stokes matrix elements. I t is not possibie. h o w e v e r , t o recover t h e original scattering matrix from either, since the overall or absoluze phase of t h e scattering matrix (far example the phase of t h e Shh term) h a s been lost in forming the cross-products. The r e l a t i v e phase b e t w e e n t h e scattering matrix e l e m e n t s is preserved in t h e covariance matrix and Stokes matrix formats, in t e r m s such as t h e shhSv: cross-product, whose argument is t h e relalive phase terms. between the Shh and

The covariance matrix is also Hermitian, so t h a l t h r e e 01 its elements ( S h y ~h;l , S h i and sh: contain r e d u n d a n t information, which is already contained in three of the other elements.

s , ,

s ,

A n o t h e r w a y of r e p r e s e n t i n g t h e c r a s s - p r o d u c t s derived f r o m t h e scattering matrix elements is in the Stokes matrix format 1131. For reciprocal scatterers t h e Stokes matrix M is a 4x4 symmetric matrix. with t h e following elements:

11 = 0 . 2 5 ( s l c h s h i
Shh S h i

2 s h v sh: f

M 12 = M

- svv

sv:

) )

13 = 0 S R e ( S h i s h v

) + 0.5Re (Sh:

i t is still possible t o carry out polarization synthesis on t h e covariance matrix a n d Stokes matrix f o r m a t polarimetric SAR data t o synthesize l h e radar cross section measured by a radar of arbitrary transmit and r e c e i v e polarization [ I O ] . A c o m m o n t e c h n i q u e employed on SAR measurements of radar cross section t o reduce t h e variance d u e t o speckle is to perform i n c o h e r e n t averaging or multi-looking. Using t h e scattering matrix data. t h e synthesized r a d a r cross section would be calculated for each pixel using ( 6 ) . The calculated radar cross section values would t h e n b e averaged t o reduce the speckle variance. Multi-looking can also be carried o u t by averaging t h e covariance matrix or Stokes matrix elements for several pixels t o produce a single composite result, then performing

5-4

polarization synthesis on the composite result (141. The two approaches can be shown t o be mathematically equivalent and give the Same answers, but there is a considerable Saving in computation time using t h e covariance matrix and Stokes matrix approach. Example Scattering Matrices Table 1 gives examples of some simple scattering matrices for selected man-made targets. The targets a r e the trihedral corner reflector and the sphere (which have the same scattering matrix). the dihedral corner reflector, the dipole and the polarimetric active radar calibrator (PARC) 1151. Only the scattering matrices for the trihedral and the sphere show no depend_ence a n the rotation of the scatterer about the vector n which defines the line of sight between target and radar. By rotating the dihedral. the dipole and the ?ARC about the line of sight it is possible to change the scattering matrix for these targets as seen by the r a d a r . The scattering matrices corresponding to more complicated man-made targets. such as vehicles or planes. may be very different f r o m the Simple ones given in Table 1.

Fig. 4 L-band total power image of San Francisco from the NASA/JPL system.

I n Figure 4 , we see a total power image of San Francisco. California which was derived f r o m L-band ( 2 4 c m ) p o l a r i m e t r i c SAR d a t a collected b y t h e NASAIJPL DC-8 SAR system 171. The image is interesting because it contains three distinct types of scatterer: the man-made buildings and other Structures in the urban w e a s ; t h e largely vegetation covered areas in t h e Golden Gate Park and t h e Presidio (just south of the Golden Gate Bridge); and large stretches of open water in the bay and out in the Ocean.
CO-POLARIZE0 5IGNATURE CROSS-PCLARIZED SIGNATIIFF

In Figure 5 . a n example of a useful tool for visualizing t h e polarimetric scattering properties of a t a r g e t . known as the polarization signature plot, is shown. I n this case. the polarization signature plot was calculated from a composite Stakes matrix formed from pixels in one of the urban areas in Fisure 4. After I l 3 l . t h e polarization signature plots represent the isynthesized) response of the target to all possible like- and crosspolarized r a d a r transmit/receive combinations. The

Fig. 5 Polarization signature for a predominantly urban area in the San Francisco image.

Table 1: Scattering matrices for man-made targets, normalized so that the largest matrix element is one

Target type

Scattering matrix

Trihedral , Sphere

Dihedral

cos2a sin2a
sin2a -cos2a

Dipole

cos *a sin a cos a

sin *a

sina cas a

PARC

sin a cos a -sin *a

cos2a
-sin a cos a

[ais a rotation angle for each target type. about the line of sight to the radar.
a-values of Oo,
45O

and 90 are commonly used. in prac1iSe.l

5-5 polarized signature plots have no zero points for any Orientation angle/ ellipticity angle combination. The like-polarized signature plot is similar in farm t o t h a t of t h e w a t e r . except that t h e polarization signature of t h e vegetated area appears t o sit on a 'pedestal'. The presence of such a pedestal indicates that the individual scattering. covariance or Stokes matrix measurements used t o calculate the polarization signature plot for the area in question w e r e not identical. In fact, the height of the pedestal can be related to the variance in t h e measurements. Note t h a t t h e polarization signature plots for t h e w a t e r shown in Figure 6 do not have a s i g n i f i c a n t p e d e s t a l , w h i c h Suggests t h a t t h e measurements averaged to produce those plots w e r e fairly similar. Design of Polarimetric SAR systems

CO-POLARIZED SIGNATURE

CROSS-00LARlZEC SIGNATURE

Fig. 6 Polarization signature for a n m e a n area in the San Francisco image.

45

45

el1,"tlCitY

CROSS-POLARIZED SIGNATURE

An example of a polarimetric SAR system is NASA/JPL's DC-8 SAR system. first tested in J a n u a r y 1 9 8 8 . which C - and P-Band) and operates a t t h r e e frequencies (L-, four polarizations, HH, H V . VH and V V ( i n this notation H V means Horizontal polarization on transmit, Vertical on receive). This system is t h e latest in a series of NASA/JPL synthetic a p e r t u r e radar systems that have been designed. built and tested primarily b y the Radar Science and Engineering Section a t t h e Laboratory. The radar will serve as a test-bed for SIR-C with which it h a s similar characteristics. as well as acting as a useful science tool for the remote sensing community in its own right.
A detailed description of the NASA/JPL's DC- 8 S A R can b e f o u n d in 171. A block diagram of t h e basic transmit/receive circuitry required for just one of t h e frequencies is shown in Figure 8. This basic design contains just o n e transmitter, t w o receivers and t w o a n t e n n a s (one H. o n e VI. In the DC-8 SAR case, t h e radar system g e n e r a t e s a pulse a t L-Band which is frequency shifted using a common reference oscillator t o g e n e r a t e t h e P- a n d C-Band pulses. A f t e r amplification a t each frequency, t h e transmitted chirp is alternately polarized b y the operation of a switch 10 either t h e H or the V antennas. The antenna patterns for each of t h e frequencies cover approximately t h e same range of incidence angles. between 2 0 ' and 70'. On receive. t h e radar can collect both N and V channel data simultaneously a t all t h r e e frequencies, t h e six receiver channels being entirely separate. The receivers h a v e no Sensitivity Time Control (STC) or Automatic Gain Control (AGC). but t h e receiver gains can be adjusted over a 2 6 4 8 range of settings. A wide instantaneous dynamic range of -4SdB is achieved b y using &bit analog-to-digital converters (ADC'sI, which a r e clocked t o g e t h e r t o e n s u r e c r a s s - c h a n n e l coherence. The ADC's each o p e r a t e a t 4SMHz and produce real data samples (not I and Q) a t a sustained d a t a r a t e of b e t w e e n 2 0 a n d 6 0 MBytes/second. depending on the data collection mode.

Fig. 7 Polarization signature for a vegetated area in the San Francisco image. polarization signature plots a r e given a s functions of orientation a n d ellipticity angle, and a r e normalized w i t h respect to t h e total power. Note that t h e likepolarized polarization signature in Figure 5 shows a maximum response w h e n t h e orientation angle is 90 or 00 ( 1 8 0 0 ) . This behavior. and the double-null structure visible for (yI = 4 5 O and 1 4 5 0 x = 0 " ) is v e r y similar to t h a t expected [or a dihedral corner reflector w i t h rotation angle, a On. Thus t h e polarimetric r a d a r backscatter from the urban area is seen t o b e similar t o that produced by two flat plates: one flat on t h e ground. the other vertically upright.

The scattering matrix model for Bragg scattering from a n idealized rough surface. such as wind-blown w a t e r . is:

=(: El

with a.b real, b > a >O

(12)

and <ab') = ab

i.e.. a scattering matrix w i t h zero cross-polarized r e t u r n . HH a n d V V r e t u r n s which a r e completely correlated and zero phase difference b e t w e e n t h e H H and V V returns. The ratio a / b and t h e absolute backscatter level a r e dependent on the surface roughness (i.e., t h e sea-state in this case) but need not b e k n o w n a priori. Although w e specify a rough surface for ( 1 2 ) t o hold, it should not be too rough, since Second o r d e r Bragg scattering can give rise 10 significant cross-polarized backscatter. The polarization signature corresponding to a patch of water from t h e San Francisco image is shown in Figure 6. We see t h a t t h e like-polarized signature shows t h a t the V V r e t u r n is greater t h a n the HH (i.e. b .a) and t h e linear ( X Oo) Polarization response in t h e crosspolarized signature is zero for all orientation angles.
~

I n Figure 7 w e Show t h e polarization signature plot corresponding t o a largely vegetation-covered area in t h e Golden Gate Park from t h e San Francisco image. In this case w e see t h a t t h e cross-polarized and like-

The operation of the DC-8 SAR is such that alternate H and V polarized pulses a r e transmitted. The r e t u r n echoes a r e received b y both t h e H and V receivers simultaneously. Thus for o n e H-polarized transmitted pulse t h e H - a n d V-polarized r e t u r n echoes a r e recorded ( H H and V V r e t u r n s ) : t h e next transmitted pulse will b e V-polarized and t h e V H and V V r e t u r n e c h o e s a r e r e c o r d e d . This process is r e p e a t e d throughout a d a t a - l a k e . Then t h e recorded r e t u r n echoes undergo SAR processing ( r a n g e and azimuth compression) to produce SAR images corresponding t o each of the set of HH. H V . VH and V V return echoes. The SAR processing operations applied t o each polarization a r e identical. After processing. the H H and H V images must be resampled in t h e azimuth dimension so t h a t they a r e registered with the VH and V V images: prior to t h i s o p e r a t i o n t h e t w o s e t s of i m a g e s will b e misregistered by o n e pulse repetition interval. After registration t h e Scattering matrix measurements can b e

Circulator
1% Receiver

H Antenna

Polarization Switch

V Antenna

Circulator
Fig. 8 Simple block diagram for a oolarimetric SAR system c o n s i d e r e d t o b e Spatially coincident. i.e. t h e y correspond to the same patch of ground. Polarimetric Calibration Calibration of the data produced by polarimetric SAR's (e.g. 111. 1161. [I711 is necessary if the data is to be used in a q u a n t i t a t i v e f a s h i o n Goals for calibrating polarimetric SAR data are given in Table 2. The basic problem in calibrating polariinetric radar data can b e seen from the following expression. after 1181. for the measurements made by a polarimetric radar for which the 2x2 R and T matrix system model is valid:

(provided R = R and T = T . w i t h denoting an e s t i m a t e ) . Note t h a t t h e r e s u l t still h a s n o i s e superimposed on it. At t h e complex SAR image stage (which is what polarimetric SAR scattering matrix data is). although the mean noise contribution i s zero, t h e r m s Uncertainties d u e t o noise can not be removed. After square-law detection, however. the average noise power can be s u b t r a c t e d off t h e r e s u l t i n g RCS measurements (though this still does not remove t h e r m s uncertainties d u e to noise). Polarization synthesis. in which t h e target RCS, O p 4 , in response t o a radar with transmit polarization p and receive polarization q is synthesized from t h e scattering matrix, involves a square-law detection step. so the noise bias l.erm can be
~

subtracted off the Synthesized image pixel powers provided it can b e estimated. We characterize the (complex) noise t e r m s a s having two-dimensional, zero-mean. Gaussian distributions. with the following properties: where

6, and 6 , a r e the H V and VH cross-talk

or
injk)
=

polarization impurity terms o n receive, 6, and 6, a r e the H V and VH cross-talk terms on transmit, f l is the channel imbalance (amplitude and phase) between t h e H and V channels on receive, and f 2 is the channel imbalance (amplitude and phase) between the H and V channels on transmit. We shall ignore the radiometric calibration problem for the moment. setting Ks = 1. and absorbing the a f a c t o r into the individual noise matrix elements, n p q . We shall also ignore the absolute phase calibration problem for now, setting r$$ = 0. The polarimetric radar data calibration problem can then be stated as follows: to solve for the scattering matrix s from the measured matrix M w e need estimates for the m a t r i c e s R and T , w h i c h r e p r e s e n t t h e r a d a r polarimetric characteristics on receive and transmit. Ideally. R and T are identity matrices for a perfect radar polarimeter. Once we have obtained estimates for R and T they can then easily be inverted and applied to 113) to recover an estimate for i.e..

(njkni) =

qk
o
,

(njkn;,)= 0 , f o r j # l o r k # m
(njks,;)
=

for any j , k , 1, m.

(15)

where is t h e noise power lor noise-equivalent sigma-zero) in the Polarization channel jk. We assume that the noise terms are uncorrelated with each other and with the scattering matrix (signal) terms. An important Point to note is that, in equations ( 1 3 ) and (x. y) dep,endence in any of the terms. I n practice. the elements cf R.T and N m a y v a r y significantly w i t h i n a SAR image. especially in t h e across-track dimension. Depending on how quickly the elements of these matrices v x y , it may be necessary to calculate them at many diffel-en1 points across the image. This is a Strong factor in deciding what type of approach to adopt in calibrating data from a polarimetric radar (see below)
( 1 4 ) . w e have not included a n y

s.

-t

-1

(R) M (?)-I

5-7

The v a l i d i t y of t h e majority of t h e poiarimetric calibration algorithms in the literature depends on the v a l i d i t y of t h e s y s t e m model a b o v e for r a d a r polarimeters, which w a s first put forward in 111. This system model contains just six relative p a r a m e t e r s . including four cross-talk (or leakage1 t e r m s and two channel imbalance terms. one for the H,V transmitters and one for the H.V receivers. Determination of these six p a r a m e t e r s , followed b y correction for a n y deviations from the ideal is then sufficient to calibrate t h e radar data polarimetrically. so that the HH. H V . V H a n d V V scattering matrix m e a s u r e m e n t s can b e meaningfully compared. The generally accepted polarimetric s y s t e m model relies on t h e constancy of the system, particularly the t r a n s m i t t e r s , t h e receivers and t h e a n t e n n a s . For instance. it assumes t h a t t h e ratio of t h e power t r a n s m i t t e d b e t w e e n H and V polarizations n e v e r varies. and that the receiver gains for H and V are kept at a constant level relative t o each other. This is not always t h e case for r a d a r polarimeters 1191. For example, in operating the NASA/JPL system the gain of the H (or V) receiver can be Switched by up to 6dB. depending on whether a like-polarized (HH or V V ) or cross-polarized ( V H or H V ) r e t u r n is expected. I A s described above. t h e NASA/JPL s y s t e m h a s t w o receivers. one f a r H and o n e for V. and alternately transmits H and V pulsesl. The reason for operating t h e radar this w a y is that the cross-polarized returns from n a t u r a l t a r g e t s a r e usually lower t h a n t h e likepolarized r e t u r n s . so t h e gain i s increased w h e n expecting a cross-polarized r e t u r n to k e e p t h e signal level w i t h i n the optimum range f o r t h e analog-todigital converters. This gain differential should b e removed during processing. The E R I M P-3 polarimetric S A R solves the Same problem With a different design, having only o n e t r a n s m i t t e r . one receiver and a n automatic gain control ( A G C I , which varies the gain of the radar receiver in some pre-set fashion 181. The AGC setting in amplitude and phase is usually different for H H , H V . V H and V V r e t u r n s . which are collected in sequence. The design for t h e CCRS polarimetric SAR has a n elaborate switching sequence for it's two receivers, so t h a t H H a n d V V r e t u r n s go t h r o u g h t h e s a m e receiver. and V H and H V returns go through t h e other receiver. I n none of these cases is t h e 2x2 R and T matrix system model necessarily valid. unless t h e gain differentials between receivers a r e properly removed from the data.

the problem. w e need a t least another six equations in the various terms involved. There a r e three approaches to obtaining the required six equations: t h e first is t o use man-made targets with known scattering matrices: t h e second is t o make assumptions regarding t h e general properties of t h e backscatter being measured: and the third is to make assumptions about t h e radar System p a r a m e t e r s . Of course, a l l t h r e e of t h e s e approaches may be combined to find a n appropriate solution. The first attempts (1201 - 1221) t o calibrate polarimetric r a d a r s y s t e m s used only the man-made target approach. There a r e only a limited set of such targets available. Table I lists t h e normalized Scattering matrices for Some of t h e more common ones. Barnes 1211 and Woods I 2 2 1 both used combinations of t h r e e passive devices t o come u p w i t h t h e six additional e q u a t i o n s r e q u i r e d t o solve t h e problem. I n his approach, Woods I221 used a trihedral and two dihedrals, one a t no rotation, the other at 45O. Barnes I211 also used this approach, and presented another solution using two dipoles ( a t O0 and 90 rotation angles) and one 4 5 O dihedral. The dipole signatures w e r e achieve'd using screened trihedral corner reflectors (1231, 1241). Three targets w e r e found to be necessary t o solve the problem because it was v e r y difficult to obtain more than two of t h e r e q u i r e d equations f r o m each t a r g e t signature. Yueh e t al 1251 presented a more general solution using passive targets, in which a t least two must either have singular or non-singular scattering matrices, and t h e targets must not be the same. Whit1 and Ulaby I261 showed how the problem may be solved provided just one of t h e t h r e e t a r g e t scattering matrices is nonsingular. Freeman. Werner and Shen 1151 showed how t h r e e polarimetric active r a d a r calibrator (PARC) signatures could b e used t o solve t h e problem. The sensitivity of all 01 these approaches to small errors in the rotation angles of the devices used was pointed o u t in 1191. Other authors have used assumptions about t h e system and t h e backscatter to a r r i v e a t a solution t o t h e problem. v a n Zyl I271 made the following assumptions about the polarimetric radar system:

8 3 s2 = 84
81 =

fl =

fi

(17)

e,

Let us now return to the problem of estimating R and 'I, in o r d e r t o calibrate t h e scattering matrix d a t a . Expanding (131, w e obtain f o u r equations relating the scattering matrix m e a s u r e m e n t s w i t h t h e i r actual values: Mhh
=

S h h 6 Zshv + 6 4 s v h 6264sVV ' n h h

+
+

(16a)

i . e R = T. These a r e t h r e e of t h e required equations. v a n Zyl also a s s u m e d t h a t t h e b a c k s c a t t e r w a s reciprocal. i.e. Shv = Svh, which would a t first seem t o provide a f o u r t h equation, b u t on f u r t h e r examination of equations ( 1 6 b ) and ( 1 6 ~ this ) makes one of them r e d u n d a n t . so t h e net number of equations does not increase. Next. v a n Zyl assumed t h a t for most natural backscatter types, azimuthal s y m m e t r y holds 131. for which t h e like- and cross-polarized backscatter t e r m s are uncorrelated, i.e.,

These f o u r equations contain t e n unknown quantities (ignoring t h e noise). which a r e t h e four (complex) scattering matrix e l e m e n t s a n d six (complex) r a d a r system unknowns ( t h e 6's and the f ' s ) . Clearly. to Solve

This g a v e t w o more e q u a t i o n s , which allow t h e calculation of the two cross-talk terms, 6 1 and 61. These can v a r y significantly with incidence angle as can b e seen from the plots shown in Figure 9. The solution was completed b y estimating f l f 2 from the ratio of the HH and V V measurements from a trihedral. From ( 1 6 a ) and ( 1 6 d ) it is e a s y to see t h a t for a t r i h e d r a l , w i t h scattering matrix as given in Table 1,

5-8
Table 2. Calibration Goals

Long and Short-Term Relative Calibration (Between

passes and within an image frame) Absolute Calibration (any channel) Cross-frequency calibration Polarization Amplitude Imbalance (between po1;lrization channels) Pol.arization Relative Phase Calibration (between polarization channels) Pol'arization Cross-Talk error (isolation)

f l dB

*3 dB

+ l S dB
*0.4 dB (2-way)

*I00
(2-way)

-30 dB

a) Cross- talk amplitudes

b ) C r o s s - t a l k phasies

I40
IO0

60

20
-20

-60
I

- 60

-180

IO

i
I
delta2

10 15 20 25 30 3 5 40 45 50 55

15 20 25 30 35 40 45 50 55

Local incidence a n g l e ( d e g r e e s )

Local incidence angle ( d e g r e e s )

Fig. 9 Plots of cross-talk amplitude and phase for the NASA/JPL L-band polarimetric S A R System. Showing variations as a function of elevation (or incidence) angle. radar system terms to three. Klein and Freeman I301 showed that, if backscatter reciprocity w e r e assumed, t h e n only two man- made targets w e r e necessary t o solve the polarimetric calibration problem. Sarabandi, Ulaby and Tassoudji [311 simPlified t h e problem b y assuming that ail four cross-talk t e r m s w e r e zero, using b a c k s c a t t e r reciprocity a n d a n estimate of f t f 2 from a trihedral oc Sphere t o complete the solution. Whit1 and Ulaby [321 proposed that t h e r e exists a rotation of the H - V coordinate system for which 6q = 0. Then backscatter reciprocity can be used to give t h r e e more equations and a single trihedral signature can be used to solve for f l f 2 and 61, completing t h e solution. This approach gives a solution for which t h e r e is a n a r b i t r a r y ( s m a l l ) rotation 01 t h e H - V coordinate s y s t e m . although t h e y a r e c8rthogonal. S a r a b a n d i a n d Ulaby [331 used t h e simplifying

= f l 2

(19)

Mhh

provided second order terms in the 6's and noise terms are ignored. In an alternative approach. Klein [ Z 8 l showed how just the assumption of backscatter reciprocity could be used to give three equations relating the system parameters, which could replace those in ( 3 0 ) . The remainder of the equations in Klein's approach were similar to van Zyl's. This w a s followed by an analysis of the problem in Freeman et a1 [291. in which it was shown that t h e a s s u m p t i o n s u n d e r l y i n g v a n Zyl's a n d Klein's approaches w e r e similar and t h a t t h e backscatter r e c i p r o c i t y a s s u m p t i o n could b e u s e d i n a transformation t o reduce t h e n u m b e r of unknown

5-9

assumptions that the different pairs of cross-talk terms were equal (but not that f i = f 2 ) . This was followed by a rotation of the H - V coordinate System such that 61 82, with a trihedral signature then yielding the remaining three equations.
~

C a l i b r a t i o n of t h e p h a s e d i f f e r e n c e b e t w e e n polarization channels ( i . e determination of arg(fi 1 and a r g ( f 2 ) alone) using corner reflector signatures and signatures of moderately rough surfaces is addressed in 1341. An examination of the assumptions on the system and backscatter behavior required for this approach to be successful was presented in 1351. Polarization Filtering We h a v e seen t h a t i t is possible t o synthesize t h e response of any t a r g e t t o a r b i t r a r y combinations of transmit and receive polarizations from measurements of the scattering matrix, the covariance matrix and the Stokes matrix. We h a v e also seen from t h e example polarization s i g n a t u r e s t h a t t r a n s m i t / p o i a r i z a t i o n combinations exist for which t h e synthesized r a d a r cross-section is maximized or minimized and t h a t t h e p a r t i c u l a r c o m b i n a t i o n s of t r a n s m i t a n d r e c e i v e polarizations for which t h e maximum and minimum RCS values a r e reached will depend on t h e t y p e of scattering (e.g. Bragg. dihedral mechanism). Consider the problem of trying t o detect a n isolated target ( e . 8 a ship) against a clutter background (e.8. t h e ocean) using a polarimetric SAR. In this case w e would choose t h e t r a n s m i t a n d receive polarization combination which maximized the ratio of the synthesized target RCS over t h e average of the synthesized clutter RCS, i.e. t h e signal-to-clutter ratio:

is short (less than 1 0 meters). densely packed. and has s p a r s e foliage. Yet another wetland forest t y p e is s w a m p f o r e s t . consisting of s h o r t trees or s h r u b s growing in standing w a t e r . Besides forest, t h e r e a r e reed and Sedge marshes. These are areas of herbaceous v e g e t a t i o n i n Standing w a t e r . T h e sedges a r e considerably taller than reeds ( Z O O cm versus 40 cm). I n addition t o t h e s e naturally occurring vegetation types there a r e agricultural areas consisting of cleared areas, bare soil and various crops. Table 3 contains typical radar backscatter measurements for the different vegetation types identified in the area imaged by the DC-8 SAR. From the table it can be seen that the v e r y low radar cross sections correspond to open water and bare ground a t ail t h r e e frequencies. The upland forest spans a fairly n a r r o w range of cross-sections. S h e range of r a d a r cross sections is given in t h e table b y the two values shown i n t h e table. Radar cross section t e n d s t o i n c r e a s e as t h e angle of incidence d e c r e a s e s . If incidence angle effects are considered. t h e radar cross Section for the upland forest is essentially constant. For other t y p e s of vegetation, a t P-Band w e see a steady climb in t h e H H cross section a s the biomass of t h e canopy increases until w e reach a Plateau w h e r e the upland forest begins. The same feature is visible in the L-Band and C-Band data. but a t L-Band the plateau is reached for r e - g r o w t h vegetation ( 2 - 3 m in height), while a t C-Band t h e plateau is reached for farmland vegetation f c l m in height). This t y p e of saturation of r a d a r backscatter w i t h biomass h a s b e e n observed Previously. At all three frequencies. especially C-band. w e t h e n see a Secondary increase region in t h e H H radar cross section data. 'The sites corresponding t o these data points may have much less biomass than t h e upland forest. b u t t h e y have a brighter radar r e t u r n . These areas generally have Some feature which adds a secondary scattering mechanism i n addition t o t h e volume Scatter which dominates the upland forest. One Site was a rain forest area where the trees had been cut b u t not cleared (clear-cut in Table 3). S h e site had a n enhanced HH polarization r e t u r n d u e to the many t r e e t r u n k s laying o n t h e g r o u n d . Other sites w i t h e n h a n c e d b a c k s c a t t e r w e r e sites w i t h v e g e t a t i o n growing in standing water (reeds. sedge. high marsh forest, s w a m p forest. flooded Bajo). I n this case t h e backscatter probably consists of volume scatter f r o m the vegetation canopy plus double bounce scattering d u e t o interaction between the vegetation and standing w a t e r . I n o t h e r Sites. t h e double-bounce mechanism may be enhanced because the vegetation understory is relatively clear (in t h e areas designated coffee and palm forest. for example).
A simple, three- component scattering model

SCR is also known as the contrast ratio. Several authors h a v e offered mathematical solutions t o this problem (1361-1381). The f i r s t s t e p is t o characterize t h e scattering properties of the target and the clutter, i.e. t o d e t e r m i n e Sr and sc. Then t h e Polarizations on transmit and receive which maximize the SCR have t o b e d e t e r m i n e d . Application of those t r a n s m i t a n d receive polarizations t o a n entire image is known a s polarization filtering. Radar mapping of vegetation t y p e s Due to t h e concern in t h e scientific community Over t h e global warming problem, mapping of vegetation cover (and changes in vegetation cover). especially in tropical rain forest areas, using remote sensing d a t a has received considerable attention recentiy. Synthetic Aperture Radar. with it's ability to collect data during day or night, or in cloudy conditions, is ideally suited for monitoring t h e e a r t h ' s remaining tropical r a i n forests. I n w h a t follows. w e present multi-frequency, pOIarimetric SAR results from a tropical rain forest site in Belize, Central America, as a n example of t h e ability of polarimetric SAR t o map different vegetation types. An image of a lOa10km area is Shown in Figure 10. The area under study is dominated by old growth upland tropical rain forest. There are also small areas which have been cleared of upland forest v i t h i n the last five y e a r s and allowed to re-grow. I n addition, t h e r e are areas of almost pure palm forest. consisting of either cohune or botan palm trees. There are several types of wetland forests. High marsh forest has vegetation resembling the upland rain forest, b u t standing Water is present during a large portion of the y e a r . Bajo is another t y p e of wetland forest which grows in areas having v e r y poorly drained clay soils. The vegetation

We can begin t o understand the phenomena which give rise t o t h e r a d a r backscatter characteristics seen in Table 3 and in other polarimetric SAR images of natural surfaces b y using a v e r y simple conceptual model of scattering. This model is more fully developed in 1391; h e r e a n outline is presented. We assume that scattering f r o m a veeetation laver is a combination of volume scattering f r o m vegetation, double bounce Scattering from the ground/trunk interaction and scattering from a rough surface. For volume scattering. w e assume t h a t t h e r e t u r n is f r o m r a n d o m l y o r i e n t e d , v e r y t h i n cylinder-like scatterers. The double bounce scattering component c a n b e modeled by scattering f r o m a n upright frotation angle = Oo) dihedral corner reflector . I n the model, both surfaces in the reflector a r e made of dielectric material, and t h e two surfaces m a y h a v e different dielectric constants. corresponding t o t r u n k and ground. T h e t r u n k h a s Fresnel reflection coefficients Rth and R t v for horizontal and vertical polarization, respectively. Likewise, t h e gr6und h a s Fresnel reflection coefficients Rgh and Rav. For t h e

5-10
vegetation Class Open Water
Bare Soil

IYIHI
-32.6 -25.1
I-230

- -IH/Y\
VlHH

-7.4 -9.5 ~10.7 -11.5 ~5.8 -5.8 -8.4 ~5.4 -6.4 -5.3 -7.2 -7.0 ~7.0 -8.8 -8.5 -7.5 ~ 2 8 . 7 -3.1 0.63 -16.5 -23.7 0.75 -9.0 98.1 0.54 ~13.3 -18.6 0.75 -10.1 -22.7 0.31 ~9.7 15.1 0.17 -6.9 165.4 0.06 -9.2 7.9 0.25 -9.2 0.25 7.9 -8.4 7.8 0.16 -8.6 48.5 0.2 -8.8 -5.7 0.26 -8.6 31.4 0.22 -8.0 52.1 0.12 -7.6 6.9 0.4 -4.0 -2.5 0.32

- -7.7 -1.8 ~3.3 -1.5 -0.1 0. I 0.4 0.2 0.2 0.2 0.6 ~0.2 0.6 1.7 2.0 2.1 -6.6 ~10.4 -11.6 ~11.7 -5.5 -5.6 ~7.6 -5.1 ~5.1 -5.2 -6.4 -5.2 ~5.6 -7.7 -8.0 ~7 8

-23.7
-9.0

'JHH

1
HHYV

-__

Reeds

Swamp Forest Upland Forest linland Forest .~..~.. High M. F. 2. Palm Forest Sedge Flooded Bajo corree clear-cut
High M. F. 1

14.5 -13.8 ~12.8

~

- 1 1.5 - 1 1.0 - 1 1.3

-10.3 -9.7 -9.2 -9.0 -8.2

.
*

cross section result~a n d ratios

liigh

ace in dB: phase dirrerences a r e in degrees

M. F. -High Marsh Forest

-m Even Odd -0.76 0.65 0.16 0.69 0.08 0.02 0.09 0.02
0.00 0.01

~0.3 -7.6 -5.7 -6.2 -2.1 -7.4 -5.6 -5.3 -5.0 -6.1 -6.4 ~6.6 -6.0 -3.2

~18.7 ~10.0 .35.8 -1.2 3.2 ~2.1

0.29 0.76 0.44 0.47 0.44 0.42

Table 3: Radar backscatter measurements from typical examples O f t h e different vegetation classes. vegetation Class open water
Bare Soil

- band Odd :"en - 0.67 0.77 0.02 0.65 0.14

0.10

VOl. 0.24 0.20 0.0s 0.19 0.76 0.64 0.47 0.84 0.77 0.73 0.59 0.68 0.73 0.54 0.63 0.51

- - LbJX vo1. Odd EW" vo1. - -- 0.23 0.27 0.17 0.23 0.87 0.87 0.62 0.95 0.95 0.97 0.78 0.91 0.94 0.69 0.65 0.69 0.24 0.56 0.1 I 0.20 0.19 0.14 0.36 0.16 0.23 0.12
0.29

Reeds Farmland Regrowth

Bajo

Swamp Fores Upland Forest Upland Forest

High M. F. 2

0.16 0.04 0.0s 0.04

0.02

Palm Eores1
Sedge

Flooded B q o cmee Clear-cut

High M. F. I

0.07 0.05 0.00 0.30 0.30

0.09 0.03 0.93 0.16 0.09 0.26 0.37 0.12 0.18 0.23 0.39 0.25 0.21
0.46 0.07 0.19

0.05
0.01

0.07 0.67 0.08 0.05

0.1 1

0.59 0.01 0.02

0.01

0.00 0.08
0.02 0.00

0.29 0.03 0.05 0.03 0.14 0.06

0.00

0.12
001

- - -

0.07 0.29 0.24

0.25 0.06 0.07

- -

__

0.24 0.17 0.23 0.37 0.19

0.0 I 0.00 0.0 I 0.01 0.02 0.04 0.04 0.05

0.70 0.44 0.30 0.79 0.78 0.85 0.52 0.83 0.76 0.88 0.70 0.75
0.81

0.74 0.59 0.76

Table 4: Backscatter mechanisms f r o m the model fit for typical examples of the different vegetation cimses.

surface scatter. w e assume a first-order Bragg model ( 1 2 ) is adequate to describe the backscatter. For all of these three backscatter components. w e assume that like- and cross-polarized returns are uncorrelated. and that the backscatter is reciprocal i H V = VH). Now, if the volume. double-bounce and surface scatter components are uncorrelated, the total second order Statistics a r e the s u m of the statistics for the individual mechanisms: i.e., the backscattered powers rather than voltages are added. Thus our model for the total backscatter is:

where f s . id and fv are the surface. double-bounce and volume scatter contributions to the V V cross Section, p
i s a real number and

a is given by:

(251

This model gives us four equations in five unknowns. I n general, a solution can be found if one of t h e unknowns is fixed. Since neither the surface or doublebounce mechanisms contribute t o the H V t e r m in t h e model, w e can use this to estimate the volume scatter contribution directly The v o l u m e contribution can then be subtracted off the lShh 12, ISvv l2 and ShhS,,' terms. leaving t h r e e equations in f o u r unknowns. After [401. w e then decide whether double-bounce or surface Scatter i s t h e dominant contribution in the residual based on the sign of the real part of S h h S v v * . If Re( ShhSvv. 1 is positive. w e decide that surface scatter is dominant. and fix a = 1. If Re( ShhS,,' 1 is negative. w e

5-1 I

Fig. 10 P-band (68cm) NASA/JPL DC-8 SAR image of a tropical rain forest area in Belize, Central America. decide that double-bounce ~ c a t f e ri s dominant in t h e remainder and fix p = 1. Then w e calculate f s . fd and p or a from the backscatter measurements. In Table 4 w e show t h e fraction of t h e total power contributed b y each of the three mechanisms (with odd = surface. d b l = double- bounce and v o l = volume Scatter) f o r t h e vegetation classes given in Table 3. From Table 4, w e see that the model predicts that t h e dominant backscatter mechanism for the upland forest i s volume scatter a t all t h r e e frequencies. There is a small ( < 2 0 % ) contribution f r o m t h e double- bounce mechanism at P-Band. practically none a t L-Band, and none a t all a t C-Band. The surface Scatter is v e r y low at P- and L-Band. b u t contributes ' 2 0 % of t h e C-Band r e t u r n s . This m a y be caused b y r e l a t i v e l y thick branches in t h e upper canopy which may look like surface backscatter a t that wavelength. For t h e lower biomass sites and the enhanced backscatter sites. w e see t h e v o l u m e scatter percentage d r o p for all t h r e e frequencies, while surface and/or double-bounce components become significant. The b a r e soil results. f a r e x a m p l e , indicate t h a t s u r f a c e s c a t t e r i n g is dominant a t ail t h r e e frequencies, while for reeds, t h e double-bounce term is dominant. Supervised Classification Polarimetric SAR data can be a v e r y effective tool for m a p p i n g d i f f e r e n t s u r f a c e cover t y p e s . S e v e r a l t e c h n i q u e s a r e a v a i l a b l e in t h e l i t e r a t u r e f o r performing supervised classification o n polarimetric r a d a r d a t a . First training sets a r e selected, which represent different surface types. Then the elements Of t h e covariance matrix for each surface t y p e can b e averaged t o find t h e average polarimetric scattering properties for each class. Kong e t a1 [411 developed a supervised classifier in which t h e following vector i s formed for each pixel in the image:

(27)

t h e n a distance measure class i. where

dj

(F) is calculated

for each

C i is t h e covariance matrix calculated for t h e ith class f r o m t h e t r a i n i n g s e t and P,(i) is t h e a p r i o r i Drobability that t h e pixel belongs in class i. The pixel under classification is classified as a member of class i if
dj

N) < d; . .(x) . , for all i

, ,

ti

(29)

This classification scheme is a maximum likelihood scheme for polarimetric radar data.

5-12

Another type of classifier is known as the minimum distance classifier. For polarimetric SAR data. a minimum distance classifier w a s implemented by van Zyl and Burnette. For this type of classifier the distance measure used is:
(30)

I81

Sullivan. R . e t ai, Polarimetric X/L/C-band SAR Proc. IEEE National Radar Conf.. Ann Arbor. MI. PP. 9-14, 1988 Carver. K.. e t al, Shuttle Imaging Radar-C Science Plan. JPL Publication '86-29, 1986. Zebker. H. A . and v a n Zyl, J. J., Imaging Radar Polarimetry: A Review. Proc. IEEE, vol. 7 9 , No. 1 1 . November 1991, pp. 1583-1606, Ulaby. F.T. and Elachi. C. (ed.). Radar :Polaiimetry for Geoscience Applications, Artech Hau:;e, 1990. B0erner.W. M., e t a1 (eds.). Inverse Methods in Electromagnetic Scattering, Hingham, M A , Reidel, 1985. van Zyl, J. J., Zebker, H. A . and Elachi, C., Imaging r a d a r polarization s i g n a t u r e s : t h e o r y a n d observation, Radio Sci., vol. 2 2 , pp. 529-:543, 1987. Dubois, P.C. and Norikane, L.. Data v o l u m e reduction for imaging radar polarimetry, Proc. IGARSS '87, Ann Arbor. MI, pp.691- 696, 1987. F r e e m a n . A., S h e n . Y. a n d W e r n e r , C.L., Polarimetric SAR Calibration Experiment Using Active Radar Calibrators, IEEE Trans. o n Geoscience and Remote Sensing, Vol. GB-28, No. 2 , March 1990. Freeman, A , . Werner, C. and Shen, Y., Calibration of Multipolarization I m a g i n g R a d a r , P r o c . IGARSS '88, pp. 335-337. 1988. Freeman, A , Shen. Y.. v a n Zyl. J.J. and Klein, J.D., Calibration of NASA/JPL DC-8 SAR data. Proc. IGARSS '91, Espoo, Finland. June 1971 F r e e m a n , A , . SAR Calibration: An Overview, Submitted to IEEE 'Trans. on Geoscience and Remote Sensing. October 199 1 Freeman, A , , A New System Model for Radar Polarimeters, IEEE Trans. on Geoscience and Remote Sensing, Vol. 2 9 . No. 5 . P P . 7 6 1 - 7 6 7 , September 1991. L i v i n g s t o n , P.S. a n d Kaplan. D..., R a d a r Calibration P r o c e d u r e , Technical Note 1 4 3 2 , Naval Ocean Systems Center, Sa" Diego, C A . November 1985. Barnes, R.M., Antenna Polarization Calibration Using In-Scene Reflectors, Lincoln Laboratory Technical Report TT-65, Sept. 1986. Woods. M . A . . A . Calibration Procedure For A Coherent Scattering Matrix Radar Royal Signals and Radar Establishment, Memorandum 3 8 8 9 May 1786 K e n n a u g h , E . M . . Polarization P r o p e r t i e s of Corner Reflectors With Modified Walls, IOhio State University. Report 612-6, January 1957. Kennaugh, E . M . a n d Chang, C., Design a n d T h e o r e t i c a l P e r f o r m a n c e of Reac1.ive-Wall Corner R e f l e c t o r s . Ohio S t a t e U n i v e r s i t y Technical Report NOW 64-0215-D, January 1965. Yueh. H . A . , Kong. J.A., Barnes, R.M.. rand Shin, R.T.. Calibration of Polarimetric Radars Using InScene Reflectors. J. of Electromagnetic Waves & Applications..

I91
[IO1

where
11 11

1121

(31)

I131

I141
i . e a subset of the covariance matrix elements and the decision r u l e that the scatterers which are

1s
Y

represented by

are a member of the class i if

#

I151

dj

(Y)< d, (Y) for all j

(32)

Acknowledgment Part Of t h e work described in these notes was carried out by t h e Jet Propulsion Laboratory. California Institute of Technology. under a contcact w i t h t h e National Aeronautics and Space Administration.

(161

I171

Bibliography
[I!

1181

Zebker. H . A . , v a n Zyl, J.J. and Held, D.N., Imaging Radar Polarimetry From Wave S y n t h e s i s , J . Geophys. Research, Vol. 92. pp. 683-701. 1987. v a n Zyl. J.J., Papas. C.H. and Elachi. C.. On t h e optimum polarizations of incoherently reflected waves. IEEE Trans. on Antennas and Propagation. voi. AP-35, pp. 818-825, 1987. Borgeaud. hil., S h i n , R . T . a n d Kong, J.A., Theoretical models for polarimetric radar clutter, J. Electomagnetic Waves and Applications. vol. 1, pp. 67-86. 1987. Boerner. W . M . . FOO, BLP. and Eom. H . J.. I n t e r p r e t a t i o n of polarimetric Copolarization phase term in the radar images obtained by the JPL airborne L-band SAR system, IEEE Trans. on Geoscience and Remote Sensing. vol. GE-25. P P . 77-82, 1987. Uiaby, F.T.. Held, D.N.. Dobson. M.C., McDonald, K . C. and Senior. T. E. A , , Relating Polarization phase Difference of SAR Signais to scene properties. IEEE Trans. on Geoscience and Remote Sensing. vol.GE-25. pp. 83-92. 1787. Evans. D. L.. F a r r . T. G., van ZYI, 1. 1. and Zebker, H . A , . Imaging radar polarimetry: analysis tools and applications. IEEE Trans. On Geoscience and Remote Sensing. vol. GE-26, pp, 774-789, 1988. Held. D.N.. e t a l . The NASA/JPL multifrequency. m ~ l t i p o l a r i z a t i o n airborne SAR s y s t e m , Proc. IGARSS '88. Edinburgh. Scotland, P P . 3 4 5 - 3 4 9 , 1988.

1191

121

1201

131

I211

141

1221

151

I231

1241

161

I251

1 7 1

5-13

1261

Whitt. M.W.. Ulaby. F.T., Polatin. P. a n d Liepa. V.V.. A General Polarimetric Radar Calibration Technique. lEEE Trans. o n Antennas and Propagation. Vol. AP-39. No.1. pp. 62-67. January 1991. v a n Zyl. J.J., Calibration of Polarimetric Radar I m a g e s Using Only I m a g e P a r a m e t e r s a n d T r i h e d r a l Corner Reflector Responses. I E E E Trans. o n Geoscience a n d Remote Sensing. Vol. GE-28, No.3, pp. 337-348. May 1990. Klein. J.D.. Calibration of Complex Quadpolarization SAR Images Using Backscatter Correlations, s u b m i t t e d t o IEEE T r a n s . o n Aerospace and Electronic Systems, 1989. Freeman. A , , v a n Z y l . J.J.. Klein. J.D.. Zebker. H . A . and Shen. Y.. Calibration of Stokes a n d Scattering matrix f o r m a t polarimetric SAR d a t a , accepted for publication i n IEEE Trans. o n Geoscience and Remote Sensing. 199 I Klein. J.D. and Freeman. A,. Quadpolarization SAR Calibration Using Target Reciprocity. Journal of Eleclromagnetic Waves & Applications. Vol. 5 , No. 7 . p 735-751. ~ 1991. Sarabandi. K. and Ulaby. F.T., and Tassaudji. M.A., Calibration of Polarimetric Radar Systems With Good Polarization Isolation. IEEE T r a n s . o n Geoscience and Remote Sensing. Vol. 28. PP. 3 3 7 348. May 1990. W h i t t . M . a n d U l a b y , F.T., A P o l a r i m e t r i c calibration technique w i t h insensivity to target orientation, Proc. lGARSS '90, Washington. D.C.. pp. 1089-1092. M a y 1990. S a r a b a n d i , K.and Ulaby, F.T., A Convenient T e c h n i q u e For P o l a r i m e t r i c Calibration of Single- Antenna Radar Systems, IEEE Trans. on Geoscience and Remote Sensing, Val. GE-28. No. 6 , pp. 1022-1033,November 1990.

1341

Zebker. H . A . . a n d Lou, Y.. Phase Calibration of Imaging Radar Polarimetric Stokes Matrices. IEEE Trans. o n Geoscience and Remote Sensing, Vol. GE-28, No. 2. pp. 246-252, March 1990. S h e e n , D.R.. F r e e m a n , A . and Kasischke, E.S.. Phase Calibration of Polarimetric Radar Images. I E E E Trans. On Geoscience and Remote Sensing, Vol.GE-27. pp.719-731,Nav. 1989. Ioannidids, G. A . and Hammers, D. E., Optimum P o l a r i z a t i o n s f o r T a r g e t Discrimination i n Clutter. IEEE Trans. o n Antennas and Propagation. Vol. AP-27, pp. 357-363, 1979. Dubois, P. C. a n d v a n Zyl, J. J., Polarization filtering of SAR data. Proc. IGARSS ' 8 9 , Vol. 3, pp. 18l6-l819, 1989. Swartz, A . A , , Yueh. A . H.. Kong. J. A.. Novak. L M. a n d S h i n , R. T.. Optimal polarizations f o r achieving maximum contrast i n r a d a r images. Journal of Geophysical Research, Vol. 93, No. 8 1 2 , PP. 15252-15260. 1988, Durden. S . . e t al. Multi-frequency polarimetric r a d a r o b s e r v a t i o n s of a tropical r a i n f o r e s t , s u b m i t t e d to I E E E Trans. On Geoscience a n d Remote Sensing. April 1992. v a n Zyi, J . J . , Unsupervised classification of scattering b e h a v i o r using r a d a r p o l a r i m e t r y d a t a , IEEE Trans. On Geoscience a n d Remote Sensing, vo1. 27. P P 36-45. 1989. Kong, J. A.. Swartz, A . A,. Yueh. H. A , . Novak, L. M . and Shin, R. T.,Identification of Terrain Cover using t h e O p t i m u m P o l a r i m e t r i c Classifier. J o u r n a l of E l e c t r o m a g n e t i c W a v e s a n d Applications. Vol. 2. pp. 171-294, 1988.

1351

(271

!361

1281

1371

I291

I381

1301

1391

I311

1401

1321

1411

1331

6-1

RADIOMETRIC CALIBRATION OF SAR SYSTEMS

1 3 Y H. Ottl neutsche Forschungsanstalt fur Luft- und Raumfahrt e . V Institut fiir Hochfrequenztechnik 8031 Oberpfaffenhafen Germany
1. SUMMARY

Most SA8 image interpretation performed in recent years was based on data which were often insufficiently calibrated. Ground truth data were used for comparison and interpretation. The importance of calibration was recognized by the need for reproducible data, by the introduction of multifrequency and multipolarizatian systems (interchannel calibration) and the long term scope of remote sensing. Hydrologists, especially, requested an absolute calibration with tolerances of less than 1 dB. Internal calibration schemes, as well as the use of external passive and active calibrators, were introduced to achieve this goal over the wide dynamic range. The in-flight measurement of the antenna pattern by means of ground based receivers became increasingly important due to pattern changes caused by electronic beam steering and the necessity of beam alignment in case of multifrequency and/or polarimetric operation modes. The use of radiometric corrections to compensate for near rangelfar range differences caused by antenna pattern and geometry of illumination will be explained in the lecture. The impact of geocoding on radiometric levels will
a l s o be mentioned, including the phenomenon of

the dynamic range of radar cross-sections IRCS). Flight passes were a l s o performed with DLR'S aicborne SAR. Results concerning the transfer function between image amplitude and radar backscattering coefficient of nLR's airborne C-band SAR were published in [7] and, with NASAIJPL'S airborne SAR performance, in 1 8 1 . The latter reference describes the cross-calibration between the 2 systems as well as the polarimetric calibration of the JPL system. The main emphasis of the lecture will be on the spaceborne sensors, because an angle dependence of oo has a strong impact on image evaluation due to the large incidence angle range coverage of airborne systems.
3. RADIOMETRIC CORRECTION OF NEAR RANGEIFAR RANGE DIFFERENCES AND OF ANTENNA PATTERN INFLUENCE

Spaceborne SAR sensors usually have a half power beamwidth (HPBW) of a few degrees in elevation. Therefore, the influence of incidence angle changes on can, in many cases, be neglected if the instantaneous field of view (IFOV) covers flat terrain. The range dependence cannot be neglected, because it affects the signal-to-noise ratio S , by a power of 3 .
A

over-/underexposing hilly regions, caused by incidence angle changes.

2. INTRODUCTION

similar statement can be made for the antenna gain directional pattern G(19), since the signalto-noise ratio increases with the square of G ( r P ) , as can be seen in the simplified SAR-equation below:

For the Radarlab missions, SIR-C/X-SAR Science Plans (1, 21 were published which described experiment goals within the various disciplines (such as geology, hydrology, glaciology, vegetation, technology). At that time, it was not possible to m a r a n t e e SAR Svst.em oerformance as well as was
K is assumed to be constant, because the average power Pa", wavelength h , azimuth taper factor ahs (of the synthetic aperture), slant range resolution rR, Boltzmann's constant k, temperature ,T , noise figure F, platform velocity u and pulse taper factor aB do not change with respect to the off-nadir angle 8 .
In order to keep S,(ril seeming constant acLoss the swath, assuming at the same time that so(*) is constant across the swath, the following correction function CF is required:

lity and relative and absolute radiometric calibration. Consequently, a number of selected experiments in the field of technology are dedicated to radiometric calibration. The engineers designing and manufacturing the SAR sensors were requested to provide internal calibration loops. Means for X-SAR internal calibration are described in [ 3 ] . Similar loops were integrated in the ERS-1
141.

An overview of the efforts and plans of five years ago was published in ( S I , and one year ago a workshop was dedicated to SAR calibration 1 6 1 .

In 1989, the flights of NASA/JPL's airborne SAX over various European test sites were used in a calibration campaign. For this purpose, 42 trihedral reflectors, 4 dihedral reflectors, as well as a receiver prototype and an ARC prototype, were positioned on grassland and concrete surfaces within DLR's Oberpfaffenhofen test site. The corner reflectors were located in special configurations, taking into account viewing angles, spacing and

(2) G2 (8) This correction causes a relatively accurate amplitude distribution within a SAR image (across the swath) but, of course, it cannot change the real signal-to-noise ratio.
CF = As an example for deriving CF, the X-SAR elevation pattern is used to illuminate a "flat Earth" under 40" off-nadi; (Fig. 1).

R3(r11 sin8

6-2

Fig. 1 Illumination geometry; swath width corresponds to HPBW.

swath width

X-SAR RADIOMETRIC CORRECTION

*t

off-boresight angle [deg]

Fig. 2 Correction function CF across the swath. Near range is on the left, far range on the right; boresight angle of antenna is 0 " . The computed correction function CF across the swath within the HPBW is shown in Fig. 2. The asymmetric shape is caused by the illumination geometry.
4. INFLUENCE OF BORESIGHT ERRORS DUE TO UNKNOWN ATTITUDE ERROR IN ROLL DIRECTION

For known antenna pointing and given (or measuredl pattern shape in elevation, it is possible to ap-

6-3

-40

-1 0

6
-5
~

0 5 off-boresight angle [des]

10

Fig. 3 Predetermined X-SAR elevation pattern.

.'

. ,

,,

'.

'.

-5

-4

-3

-2 -1 0 1 2 off-boresight angle [deg]

Fig. 4 Gain error function for various boresight angle errors; solid line i 0.3O, and dasheddotted line t 0.4O. ply equation (2) to correct the image amplitude.
5. ANTENNA PATTERN MEASUREMENT

In case the pointing error is unknown, a radiometric. error will be produced which, in most cases, cannot be corrected.
F o r instance, the predetermined X-SAR elevation pattern (Fig. 3 ) is used to compute a gain error function (Fig. 4) for various boresight uncertainties ( 0 . 1 " to 0.4' angle deviations). An HPBW of about f 3" (in Fig. 41 is relevant for the swath width of X-SAR.

The gain e r r o r function has to be squared to show its impact on image amplitudes (grey levels of SAR image).

ment.ioned in Dart 3 of this _ DaDer. . . the antenna pattern must be known for precise radiometric correction. In view of strong mechanical stress during launch, possible deployment errors and thermal effects, the actual inflight antenna pattern should be used to achieve the highest quality results. In our institute, we use a large number of high precision calibration receivers, aligned in the cross-track direction (see Fig. 5) to measure the azimuth cuts of the actual antenna pattern. Every receiver digitizes and stores each received radar pulse as 16 samples. In parallel, UTC is co-registered with high precision ( l o - ' ) , so that the measured azimuth-cuts can later be time-correlated and the required elevation main cut of the antenna pattern can be reconstructed. Of course, the precise location of the SAR-sensor v s . time must be known to convert the time of pulse receipt
AS
~ ~ ~~~~ ~~

6-4

, , ,
Fig. 5 : Schematic experiment set-up.

Receiver

1201
c

\
H
H

3
c

Q
3

0 Q

80-

0 0

15 SAMPLE NUMBER

0
been received from, to about
?

Fig. 6: Registered pulseshapes from NASA/JPL's C-band HH DC-8 SAR. into antenna angles. It is a l s o mandatory to precisely know the attitude data of the platform as well as the location of each receiver.
I n t h e 1991 NASAiJPL airborne SAR-campaign over o u r test si.te, we had a first opportunity to test OUT measuring System [ 9 1 .

The relatively large deviations of the neasured pohts from the fitted curve in Fig. 7 can be explained by 3 facts:
1) precise flight and attitude data haw! not yet
2 ) the positions of the receivers were cmnly known

Fig. 6 s h o w registered radar pulses from the C-band SAR of NASAIJPL'S DC-8 (horizontal polarization). Slight distortions, i.e. decreasing power from begin to end of the pulse can be seen, probably due to the wide chirp used and the relatively small antenna bandwidth.

loom,

3) at the time of the experiment not a l l of our equipment was fully calibrated.

An azimuth cut can be obtained, by integrating the pulse energies and plotting them against time, The time-correlation of all azimuth cuts gives the main cut of the antenna elevation pattern (see Fig. 7 ) .

These drawbacks will be eliminated in our future experiments with spaceborne SAR-sensors (all receivers are fully calibrated now, positions are determined using differential GPS and precise orbit and attitude data are available from the satellite owners). The co-registration of the time of pulse receipt

6-5

k7

o
-5
-10

C .-

m 0)
C

C
(3

W +

m
W

N ._

-15

-20
15
30

45 60 off-nadir angle [deg]

75

90

Fig. 7 Reconstructed main cut of the antenna elevation pattern.

I
Fig. 8 X-SAR calibration block diagram can yield two sets of further information: Precise measurement of cross-band beam alignment in azimuth and elevation direction (especially suitable in the X-SARISIR-C mission) and possible antenna squint angles. Our system will be used in all present and nearfuture SAR missions (ERS-I, JERS-I, X-SARISIR-C, PRIRODA) ,
6 . INTERNAL CALIBRATION

fore and after data takes. It can be used to deduce an optimum reference function for processing. The test signal can either be taken from the low power part of the radar transmitter (HPA is switched off) or from the directional coupler between the HPA and antenna. In the latter case, the low noise amplifier (LNA) will be bypassed. These loops are indicated in the block diagram in Fig.8. Internal calibration of active array antennae is more complicated due to the large number of transmitireceive modules (TIR modules), which can easily reach several hundred to several thousand. These TIR modules of the next generation SAR antennae will be designed and manufactured in monolithic microwave integrated circuitry (MMICI based on GaAs technology. Besides control and monitor units (power, phase shifter, polarizationl, a time sequential check-out of the transmitter and receiver stages is necessary. Different concepts using separate feeding networks to distribute or collect calibration signals within the active a r rays are presently under study at various centres. For internal calibration, the balance and orthogonality of in-phase and quadrature channel (after demodulation) are important. E r r o r analyses for X-SAR have shown that 5" deviation from orthogonality causes about 0.2 d ! 6 amplitude error and 0.2

Spaceborne SAR systems should have life times between 3 and 6 years without too much degradation in performance. Internal calibration loops are advisable to monitor system stability. For the high power output, monitoring can be achieved by measuring a small portion; the measuring point could be a directional coupler between the high power amplifier (HPA) and the antenna. Receiver gain stability and linearity can be measured by feeding a small part of the radar signal via attenuators through the whole receiver chain. For power leakage problems, the HPA is usually switched off during this test phase and the signal is taken from a low p w e r section. Furthermore, the actual chirp can be measured be-

40 I , I I I

30
20

m
U
b
Y
0

10

0.0
- 10

TRACK 2
C - Band

+ grass

/ VV

- 20 + concrete

cat.-factor: (18.6i0.6) dB
1
I
I

- 30

0.0

100

200

300

400

500

600

image amplitude
Fig. 9 E- SAR system transfer function.

40

30
20

m
U b
u
0

10
0.0
-

10

/-+1

A A

tronsfer function corner reflectors

TRACK 2
C - Band

+ gross

/ VV
I

- 20

- concrete
I

cal.-foctor: (6.1 i0.8) dB

I I

- 30
0.0

25

50

75

100

125

150

image arnplitude
Fig. 10 DC-8 SAR system transfer function. dB imbalance an additional error
Of

up to 0.2 dB.

7 . EXTERNAL CALIBRATION

Anothei Socrce of SAR signal Contamination i s the total system noise. This noise is mainly composed of contributions from the sensor itself (ADC, LNA, ohmic losses) and the earth surface. In the case of the X-S?.R system, this quantity is estimated before mission and measured in a receive-only mode during mission. On the rail data stage a noise subtraction takes place by using the cesulting noise data.

External calibration can only be usefully performed after having accomplished the previously described internal calibration and radiometric corrections. The main objective in the scope of absolute radiometric calibration is the determination of the overall SAR system transfer function relating image amplitudes Pi to oo-values, as expressed in the following equation:

40

30 I6

x
A

pr

- 4

m v
b

20

A A E-SAR
X

(28.9f1.1)dEm'
(29.0+0.4) dEm'
-

10

DC- 8

0.0
I
I

lheoreticol

29.3
I

dBm'
I

6-8

Antenna

Sf =

projected slope area in front of mountain

S , =

projected slope a r e a behind the mountain

L
hountain
Fig. 12 1:llumination geometry for a mountain

Fig. 13 Geocodfd Matterhorn scene

purpose, about a dozen control points from maps (Scale 1:50.000) will be used (for instance for ERS-1 images1 .

No additional radiometric correction will be applied ( s e e Fig. 13). A very rough correction would be the inverse application of equation ( 4 ) , assuming an isotropic scattering behaviour of the slopes.
This assumption is not valid, especially not in the presence of vegetational layers. A tool to include radiameric correction according to the incidence angle could be the incidence angle dependent oo. But again this implies some a priori knowledge about the ground consistency.
REFERENCES
(11 NASA

Shuttle Imaging Radar-C Science Plan. JPL Publication 86-29, Sept. I , 1986. The X-SAR Science Plan DFVLR-Mitt. 85-17. preparatory Investigations Concerning the Calibration of Spaceborne SAR-Systems. PTOC. IGARSS '91, Helsinki, 3-6 June, 1991. Design and Performance Assessment of an Ultra Stable Calibration Subsystem for a SAR and a Scatterometer. proc. IGARSS '87, Ann Arbor, 18-21 May 1987. Radar Calibration Techniques Including Propagation Effects. Ad". Space Res. Vol. 7, N O . 11, pp. ( 1 1 ) 2 5 9 - ( 1 1 ) 2 6 8 , 1 9 8 7 . proc. of the SAR Calibration. Workshop, DLR, Oct. 9-11, 1991. Radiometric Calibration of an Airborne C-band Synthetic Aperture Radar. Proc. of The 17th Internat. Symp. on Space Technology and Science, Tokyo, 1 9 9 0 . The Oberpfaffenhofen SAR Calibration Experiment of 1989. Journal of Electromagn. Waves and Appl., Vol. 5, No. 9, 1991, pp. 935-951. Ground-based Measurements of Inflight Antenna Patterns for Imaging Radar Systems. To be published in IEEE T r a n s . on Geoscience and Remote Sensing, 1992.

[ 2 1 Ottl, H.

Valdoni. F. et al.
[ 3 ] Seifert, P.

Blotscher, H

[ 4 1 Menard, Y.

Oudart. P.

151 Hartl, Ph. Heel, F . Keydel, W. Kietzmann, H

161 Heel, F.

Freeman, A. (Organizers)
I71 Heel, F. Ottl, H.

( 8 1 Zink, M. Heel, F.

Kietzmann, H

191 Seifert, P. Lentz, H. Zink, M. Heel, F.

Comparative Investigations of 1101 Zink, M Heel; F. Polarimetric Calibration MeKietzmann, H. thods. Progress in Electromagnetics Research Symp., PIERS '91, Cambridge (MA., USA), July 1991.

7- 1

SAR SIMULATION

by D. Hounam Deutsche Forschungsanstalt fOr Luft- und Raumfahrt e.V Institut f"r Hochfrequenztechnik 8031 OberDfaffenhofen Germany
1. SUMMARY
f i r t~ ~ challenoe SAR enaineer is the The ~ far . t.he ~ -~~~ system design, which requires juggling with the system parameters until the required performance is met. This can be described as a 'bottom up' type of analysis, i.e. starting with a set of performance requirements and external constraints, the engineer attempts to find a set of design parameters to meet his goal. This procedure is often called parametric analysis and culminates in a set of performance specifications.
~~~~ ~~ ~~ ~~ ~ ~ ~~ ~~

The use of software tools as an investigative method is particularly important in the case of synthetic aperture radar sensors, as the geometry cannot be reproduced in the laboratory. Also, the complete SAR system, from the target via the propagation path, sensor and image processor to the final image, represents a highly complex data chain, which cannot be treated in part. The lecture discusses different approaches from parametric analysis tools to full-blown simulators capable of analysing all elements of the SAR system. The latter will be illustrated with the aid of the SARSIM simulator, which was used for confirming parameters of the ERS-1 Active Microwave Instrument ( M I ) . Particular emphasis will be placed on tion of target scattering mechanisms, standing of which is essential if the SAR systems is to be fully exploited, modelling of sensor characteristics. the simulathe underpotential of and on the

The second task arises when the system design has firmed up and the engineer needs to monitor whether the performance is being met. This is the top down approach, deriving the performance characteristics from the estimated or measured sensor parameters. The performance of a SAR sensor is described in terms of imaging parameters (11. As a SAR image only materializes after considerable data processing, the SAR processor characteristics have also to be considered in the analysis. Clearly, somewhere the engineer has got to put stakes in the ground or his design will never converge. It is probably fortunate that some constraints exist such as physical dimensions, platform flight path lorbit) available power etc., limiting the degrees of freedom. Nevertheless, it will generally be necessary to freeze those parameters which are not part of the design task. For example, the SAR processor characteristics will be fixed when designing the SAR sensor and vice versa.
3.1 Parametric Analysis

The author would like to thank T. Pike and S. Pot ter for the use of material and K.-H.Zeller for useful discussions.
2.

INTRODUCTION

The term SAR simulation is used for a wide range of Software tools to help design and evaluate SAR systems. Clearly, such an approach can be much more economical than performing experiments, bearing in mind the difficulty, and in the case of satellite sensors, impossibility of achieving a realistic geometry. In this lecture, we will differentiate between analytic tools and simulators. The former term encompasses all purely algorithm based programs, whereas the simulators attempt to mimic as many parts of the SAR system as possible. In general, an analytic approach to solving a problem is preferable to simulation, because it leads to a more complete understanding of the problem. However, sometimes problems arise which are not accessible to analysis. Experiment or simulation are then the only alternatives. Simulation, like experimentation can also usefully augment analysis and provide valuable confirmation that the understanding of the mechanisms within the system is correct. It is clearly out of scope of one lecture to discuss the algorithms of such tools in detail and so the emphasis will be placed on architecture and performance.
3 . ANALYTIC TOOLS

choices open to the SAR engineer, frequent interaction with the user is necessary. There is no hard and fast rule as to where one starts in designing the system, but the timing scheme, i.e. the choice of pulse repetition frequencies lPRFs1 and the position and duration of the received echo need to be defined at an early stage,
3.1.1 Timing

The minimum PRF is determined by the width of the received Doppler spectrum, it needing to be higher to satisfy the Nyquist sampling theory. The steps to arrive at the PRF are as follows:

* The width D lazimuth dimension) , of the antenna is~derliedfrom the required azimuth resolution Remember the rule of thumb that the single look azimuth resolution is half the antenna width.
~ ~~~~ ~ ~~ ~ ~ ~~~~~~~

* The antenna beamwidth can be calculated from the physical width:

When confronted with analysing a SAR system, one is faced with a multitude of parameters many of which are strongly interdependent and all of which impact on performance and sensor design

k is the broadening factor due to weighting of

the antenna aperture and h is the wavelength. k is 0 . 8 8 8 for an unweighted aperture, which can be shown to be optimum, if the PRF is chosen to just meet t h e sampling requirement (see below).

7-2

'.Rort:

9
1100.0 Orbit

m.PRf.
L"!ry:

C d r q I
L 8

$ 6

:I,

:,

.
Ili.3
,I2 i 857

285 5

(27.i
' O " 9 '

* , , , *
19, S l l l l i S B S

G r ; " n d

[ i n ]

F i g . 1 Timing f o r t h e ERS-1 s a t e l l i t e derived by t h e Cross-track Geometry Program. The wanted echo

window i s i d e n t i f k d by t h e v e r t i c a l I n e s .

* N O W t h e width of t h e received Doppler spectrum

BD can be c a l c u l a t e d :

2V OB
B , =
~

'

(2)

where V i s t h e v e l o c i t y of t h e sensor platform. To s a t i s f y Nyquist, t h e PRF needs t o exceed t h e bandwidth:

P2F > 1.2 BD

(31

The bands a r e p l o t t e d on a graph of PRF versus o f f - n a d i r angle [angle between t h e t a r g e t and t h e v e r t i c a l ) and ground d i s t a n c e ( d i s t a n c e between t h e t a r g e t and t h e ground t r a c k ] . The axes can a l s o be swapped around. For t h e wanted swath, t h e u s e r can s e l e c t t h e PRF, rank (number of transmit p u l s e s a f t e r which t h e echo i s received) and t h e maximum and minimum o f f - n a d i r a n g l e s , i . e t h e l i m i t s of t h e antenna e l e v a t i o n p a t t e r n . The program c a l c u l a t e s t h e p o s i t i o n of t h e unwanted anbiguous echoes enabling t h e requirements on t h e e l e v a t i o n p a t t e r n t o be d e r i v e d .

The above s t e p s a r e simple approaches which can r e q u i r e more s o p h i s t i c a t i o n , depending on t h e case i n hand. IC e f f e c t , a software u t i l i t y needs t o be develoued t o handle a l l a s o e c t s . For examnle. . . the l a s t s t e p , where adequate sampling i s considered, ensures t h a t ambiguous responses i n azimuth, due t o a l i a s i n g unwanted energy i n t o t h e 'wanted Doppl e r s p e c t r a l band, can a l s o be influenced by t h e c h a r a c t e r i s t i c s of t h e azixuth compression r o u t i n e i n t h e S A 8 p r o c e s s o r . However, t h e design of many SARs has s t a r t e d down t h i s p a t h . The timing of t h e received echoes r e q u i r e s consid e r a t i o n of t h e cross-tn.ck oeometrv and t h e m l s e c h a r a c t e r i s t i c s of t h e r a d a r . The a r t i s t o f i n d space between suczessive t r a n s m i t p u l s e s f o r t h e echo w i t h enough margin f o r t h e r e c e i v e r t o s e t t l e back t o f u l l performance. Another c o n s t r a i n t i s t h a t t h e echo from immediately beneath t h e sensor ( n a d i r ) should a l s o not f a l l c l o s e t o t h e echo window. Having an incidence angle of 9 0 " i t can be s t r o n g enough t o overload t h e r e c e i v e r . P a r t i c u l a r l y , f o r s a t e l l i t e geometries, many p u l s e s may be underway b e f o r e t h e echo i s received ( f o r ERS-1 9 ) and j u s t a s many n a d i r echoes e x i s t . These ambiguous echoes i n t e r f e r e w i t h one another and have t o be suppressed by t h e antenna e l e v a t i o n p a t t e r n .
~ ~ ~

3.1.2 Link Budget

Once t h e timing has been defined, t h e geometry i s known and t h e l i n k budget can be computed according t o t h e radar equation. This i s a s t r a i g h t forward t a s k and doesn't r e q u i r e much s c f t w a r e sophistication.
The followina form of t h e radar e a u a t i o n c a n easil y be r e a l i s i d with a s p r e a d s h e e t . Table 1 g i v e s an example with t h e values f o r t h e ERS-1 s a t e l l i te.
~1
~~ ~~ ~~~~

~.~
~~

The radar equation f o r a s i n g l e look i n t h e f i n a l image c a n be written:

where, Ga
P N ,
= =

An example of a software u t i l i t y f o r analysing t h e

rp
T~~

= =
=

timing is t h e Cross- track Geometry Program developed by t h e DLR. Fig. 1 shows a t y p i c a l example f o r t h e ERS-1 s a t e l l i t e . The program computes t h e val i d bands where t h e echo i s corrupted by n e i t h e r t h e preceeding or succeeding t r a n s m i t pulses, nor t h e n a d i r echo. Forbidden a r e a s a r e shown hatched.

ri = PRF = h = r = co =

one way antenna g a i n , t r a n s m i t power, receiver noise density, t r a n s m i t p u l s e length, compressed p u l s e l e n g t h , azimuth i n t e g r a t i o n time, p u l s e r e p e t i t i o n frequency, wavelength, s l a n t range, v e l o c i t y of l i g h t ,

7-3
IMPULSE RESPONSE
A Z I M
O f

0
-5
-10 -15
- 20

Number

S t e p s ' lOO.000

N , . o ~ n i r p . R ~ a . c ~ l I r :2 2 . 0 0 0
Look B o n d r i d i h Hz' 10.000 15.000

~ o r g t iv e ! o c i l y

mls.

Hommioq i o c l o r AHPLITUOE E P R O R S iinaor

0 5k 0.00 0.00
1.58
0.00

de.

Ouadialic de.

- 25

Rippie cycler:

-50

R i p p l e x'mplhluda d B :
Ripple Phose dag:
PHASE

0.00

- 35
- 40

ERRORS
0.00
'

L i n t o r deg. Oucdralic daq

0 00
6.00

-45

R i p p l e cycler:

78pple Amp], d r g . :

5.72

- 50
-55
- 60

R i p p l e Phone d e q . .

0 00

-->lSLR

l o o sploll<--

-20

-15

- 10

-5

IO

15

Az~mu;i[m]

S p a t i a l R e s . (m) 2 . & 5 ! 4 20 P S L R (dE):5.3562 I S .I R (d8)Zl 0206

Fig. 2 Azimuth impulse response computed with IMPRES. incidence angle, satellite velocity, o0 = backscatter coefficient, LnTw = atmospheric loss, .._.. L? = processing loss.
=
=

v,

P I %m!hQ--

The terms have been grouped to separate the var i m s influences.

[Ai
[Bj
= =

m g U n
iskplpr lia. lrpe

3.2 18
4yoW

461
5.518
-1s8.486yYw

76.4 3.8 4

K$.sfyne cr*+amib,,

m.5 m.7e

[Cl = [Dl = [El =

[Fl
= =

[GI [HI

RF Sensor parameters, called 'nomin a l gain' for. ERS-1, range compression gain, noise bandwidth, free space loss, range resolution, azimuth resolution, target backscatter coefficient, losses.

The radiometric resolution y is computed from the expression: 1 I t

(5)

WI F a T

mrp. @=

22%-

e m
1 . io'sbg

-n.9

d:;

14.418

a m " 0 . " E g r i n e I " swielcdty Rmds

s i a t w

59.2 -12.5 9.6

0.5

-3.6 3 . 0

14.861

where L is the effective number of looks

3.1.3 Impulse Response

The shape of the impulse response of a SAR sensor is dependent on the used weighting functions and the phase and amplitude errors. The latter can be due to sensor imperfections and, in the case of the azimuth response, also due to errors in the knowledge of the flight path. The DLR program IMPRES [ Z j derives the range impulse response from the chirp characteristic, the range weighting function and amplitude and phase errors. With the azimuth response, instead of the chirp, the Doppler characteristics determine the resolution. The weighting function and errors can either be entered a s formulae or from look-up tables. Fig. 2 shows an example of the azimuth response showing panedechoes. Fig. 3 shows an impressive two-dimensional representation, which however, can be difficult to interpret.
IMPRES also evaluates the impulse response and computes resolution, peak sidelobe ratio and inte-

~ ~

'

Nrterdh%

6
~ ~

z.sm

Table 1 Spreadsheet of the link budget using the example of the ERS-1 satellite. grated sidelobe ratio
3.1.4 Performance Estimation

The above tools are designed to be quick and easy to use and run fast enough to be able to iterate the design. They can, therefore, be called upon to perform both the top-down and bottom-up design

7-4

IMPULSt
R *

KLSPONSL

.:

Fig. 3 Two-dimensional impulse response tasks mentioned above. Separate utilities like these have the advantage over comprehensive parametric analysis packages that they are easier to master. The DLR Performance Estimator I31 soft.ware is such a package which uses more sophisticated versions of the above utilities. The Performance Estimator was originally developed for the X-SAR sensor to be flown on space shuttle together with the SIR-C sensor in 1993. It grew out of the need to provide the operators duricg the mission with a tool to ensure that the radar is optimally adjusted before each data take. Experience with the preceeding SIR-A and SIR-B missions had shown that the predicted shuttle position and actitude was so inaccurate that relying an I)reset Darmeters meant that taraet sites could be missed. This led to an architecture for the Performance Estimator which has not only fulfilled the requirements of shuttle missions but also of conventional SAR satellites like ERS-1 and airborne SARS like the DLR E-SAR. Fig. 4 shows a schematic of the Performance Estimator. Its basic architecture follows the approach described above. In the case of satellite platforms. the ceometr" . h e orbit ~ , ~ is calculated ~ i from t parameters and the position of the target site. FOZ airborne sensors or cases where the orbit parameters are not known, a fixed geometry can be entered. The radar parameters are stored in tw3 ways, differentiating between those which will eventually be fixed and those which can be changed by the operator, for instznce, by means of telecomands from the ground. Examples of the latter are timino Darameters. .~oains antenna oointino. etc. The 'fixed' parameters are stored in the socalled systems file, whereas the parameters under operator control are calculated and ranked by the Estimator, providing the user with a choice. Clearly, the 'fixed' parareters c a n be changed at any time. Typically the specifications of the r a dar parameters would be used a s fixed parameters in the early phase of sensor design to be replaced by true measured values later.
~~ ~~~ ~~

Fig. 4 Schematic diagram of the Performance Estimator. where different philosophies can be used, depending on the sensor. The Estimator first calculates the timing parameters by running through all the available PRFs to see which PRFs meet the requirements. Where image quality is affected simple dlgorithms, as given in Section 3.1.1 are used for the bottom up analysis. The PRFs for which adequate solutions arm? found are ranked according to criteria, which can be selected by the operator. The solutions 8112 disolaved on the timino ~ a o elsee Table 2 1 . The ti,~~~ ~, ming page also lists the performance parameters which are dependent on geometry and timing like swath width and position, and spatial re:jolution.
j ~j~~
~~~

The algorithms used in the Estimator are coded in the program but in some cases a choice of algorithms is provided. An example of this is the algorithm for choosing the optimum gain setting,

~.~~~

~~~~~

~~~~

~~

7-5

Azimuth ambiguities occur, due to aliasing of unwanted Doppler frequency bands into the wanted region, The latter a r e suppressed by the antenna azimuth pattern. The azimuth ambiguity ratio is calculated by integrating the energy in the Ooppler band selected by the SAR processor and the energy in the ambiguous bands and dividing. Apart from the azimuth antenna pattern, the processed Doppler bandwidth and the weighting function used for azimuth compression have to be taken into account. A large number of weighting functions can be selected. Again the azimuth antenna pattern can be entered as an analytic function or as a look-up table. The distributed target ambiguity ratio is calculated by combining range and azimuth ambiguity ratios. Table 4 shows typical values for the ERS-1 satellite.
SlLBClED m ? - L 7 0 a . o " z
Y e M

ran

IETTrBC-

.-.vas
X "

ma.4

smmi-3. WrE
" * I

24I.9

218 4

270.9

m3.4
48.5

as5.3

llO.8
I S ,

,,.,
8>.4

rm
, . I B

la.>

416

(3.0

12.1

35.9

6>.J

mu

a
-

Table 2 Example Performance Estimator timing page

(ERS-II
,

The next phase of the Performance Estimator is to calculate the imaging performance for the various PRF solutions. Normally, the three highest ranked PRF solutions %,ill be analysed but the operator can change the ranking or the number of solutions, if he desires. Imaging performance is displayed on three screens: signal to noise, ambiguity ratios and target detection. Using built in algorithms for the radar equation digitisation noise, bit error noise and a variety of scenarios for the normalised backscatter coefficient oo, the Estimator computes the signal-tonoise ratios for each selected PRF solution and nine positions across the swath. The SIN at the imput to the analogue-to-digital converter, i.e. the video SIN, the SIN for each processed look and the multilook SIN are ail calculated. From the latter the radiometric resolution can be computed. The dynamic range is the range of signal powers over which a particular requirement, e.g. radiometric resolution is met. The Estimator searches over a range of co values until the radiometric resolution goes out of specification. Table 3 shows typical results for the ERS-1 satellite.
IEU.C?ln m r - L i o * . o " T

Table 4 Example of the Performance Estimator a m b guity performance. The target detection characteristics describe the ability of a SAR sensor to detect point targets above a distributed target background. Man-made targets generally consist of single or combinations of discrete scattering centres and can be regarded as having deterministic properties. Such scatterers, often called point targets, are usually much smaller in extent than the resolution cell size. Although the radiometric resolution can be calculated for such targets, the user is more interested in the ability to detect targets above the system noise or distributed target background (clutter).

To detect a scattering centre or point target successfully it must be brighter than the brightest point in the speckled background.
In principle, the radiometric resolution can be calculated for point targets and this can be a useful parameter if the radar cross section of the target needs to be measured with known precision. The application considered here, to image man-made objects consisting of one or several dominant scattering centres, requires that the weakest scatterer can still be resolved above the background. If sufficient scatterers from the target can be detected, the size and shape of the object can be reconstructed and, hence, the object described.

GlIN

szmnrs. LI

on
rm

" D i l

"mTxxc

P%so_WCE

maw : Q
WI.IDL"IION

99 89ul.

ion

m*s.

LOO 001 d I:'mIuIITI) m"I"&'wT xrmo

2 5 9 dE

Table 3 Example of the Performance Estimator signal-to-noise performance (ERS-1). The ambiguity performance in range is calculated from the geometry, timing and antenna elevation pattern. Range ambiguities occur when the slant range differs by half the pulse repetition interval (PRI). The ambiguity ratio is computed by integrating the energy in the wanted region and ambiguous regions and dividing. The antenna elevation pattern can be entered as a look-up table or as a weighted pattern with up to seven weighting coefficients.

In the approach described here, two parameters

from radar detection theory are used as a measure of the ability to detect a scatterer. The false alarm rate (FAR) is the probability that the intensity of the background in an image pixel is such that it is falsly interpreted as a point target. The detection probability is the probability that a point target can be detected above the background.
A scatterer is considered to be detected in an

image pixel if the intensity of that pixel exceeds the intensity of the surrounding pixels by an

amount large enough to meet the required false alarm rate and detection probability.
CaIB 1

To analyse the problem quantitatively, assumption: have to be made about the object to be detected and its surroundings. In the following, three scenarios for the analysis have been chosen. 1) The first scenario consists of a single stabl e ooint tamet in a backaround consistino of a distributei target. The2assumption is &at, although the point tzrget, by definition, is smaller than the resolution cell, only the point target contribrtes to the energy in the corresponding pixel. This would apply to ;he case where the point target is surrounded by a surface of low scattering cross section large enough to fill the resolution cell ;and where the system noise is negligible compared with the bac:kground 0". This scenario, is typical for large metallic structures, such as aircraft, where the scatterers are surrounded by large metallic surfaces reflecting the energy aivay from the sensor.
~ ~~~

threshold lor WIR

2)

The second scenario considers a single pojnt target in a distributed target background where the background, the system noise as well as the point target contribute to the energy in the point target pixel. This is a more general case than the first scenario and would be typLcal of buildings and small isolated objects.

3)

In both the above cases it is assumed that the target cross section remains constant. In reality, it will vary according to the aspect anale falintl. ~. For such a variation. ~~.~ a Rice distribution can be used. The third scenario therefore assumes tha; the energy in the target c e l l is a combinacion of a varying target cross section as well as background energy and system noise.
1 .1
~~

Table 5 Example of the Performance Estimiitor detection performance.

The principle of the model is that a threshold is set according to the background and the required false alarm rate. Fig. 5 shows the position of the threshold for the three detection scenarios with the probability density of the image intensities. The FAR is obtained 3y integrating the probability density of the background from the threshold to infinity. The detection probability is the area under the probability density curve of the target from the threshold :o infinity. For the first scenario it is always one With this model the Estimator can calculate FAR and detection orobabilitv. ~.TvDical results for the ERS-1 satellite are shown i.n Table. 5 . Fig. 6 shows the detection probability plotted against swath position with the false alarm rate as a parameter
~ ~~~~~~~

~~

. . ~ ~
~~ ~~

~~

The model elements of the Estimator are sumarised once more below:

a GEM6 Earth model which includes Earth rotation, a circular orbit (to calculate the platform velocity), uncertainties in the height of the platform, nadir and target, antenna and platform pointing errors, theoretical and measured azimuth and e1evat.ion antenna patterns, a variety of ground clntter models which represent the expected radar backscatter coef-

Fig. 6 Detection probability for the ERS-1 satellite as a function of FAR. The object is a large vehicle. ficient versus incidence angle,

*
a

a variety of target models (RCS char,ncteristics embedded in noise and clutter1 for False Alarm Rate (FAR) detection mod<?lling, the prediction of signal. attenuation versus rain-rate, a variety of ways of selecting the PRFs, as well as user selected,

* * *

the selection of an optimal set of PllFs far PRF hopping,

a variety of receiver gain setting algorithms which use either centre swath signal, mean swath signal and mean or the minimumlmaximum backscatter coefficient, as well as user selected,
a choice of either fine or coarse range resolution, various system hardware gains and losses, the possibility of overlapping azimuth looks, characterisation of the thermal and Analogue to Digital Converter WJCI noise sources,
a variety of amplitude weighting functions, as well a s user selected,

the system analysis performed to date,

a listing of all system files available and

the switching to other radar systems using function keys.

The Performance Estimator is typical of an algorithm based Simulation tool. The software uses DEC windows and allows the user to play with different parameter settings and check the performance. Clearly, the algorithms that are used are tried and tested, so that few major surprises are to be expected in the results. Full-blown simulators, as described below, mimic all the steps in the SAR system and can, hence, be used for scientific investigations.
~~ ~~~ ~ ~ ~ ~~ ~

4. SIMULATION

the ability to degrade the range ambiguity ratio due to an expected range of radar backscatter coefficients. The software characterising these functions has the following features: partitions which can be displayed on the screen with descriptions, DEC windows designed to allow the user to go forwards and backwards through all the options to make and assess system changes quickly, the ability to generate antenna patterns from a Fortran file, a facility to examine graphically the manner in which, for example, the antenna patterns and the ADC noise are modelled, fast evaluation of a large number of swath points which a r e split up into their azimuth look components, all the timing information that can be paged for all the PRF and includes for example the predicted interference cases, tables and graphs of ambiguity, signal to noise and imaging data for all the selected PRFs that are easily copied to a specified printer, tables and graphs of required target size for detection or detection probability versus false alarm rate for verification modelling, graphs of the components of the video signal to noise ratio and the factors that degrade this, the analysis of the effect of changing any system parameter and plotting the change produced in any of the performance measures,

A simulator attempts to model a s many parts of the SAR system without recourse to analytic functions. The advantage of this approach is that genuine data are generated, which can be examined at each stage. Fig. 7 shows the elements of a SAR system which many be simulated and the evaluations which can be performed. Although all elements contribute to the imaging performance, some may be able to be simplified depending on the application. For instance, if the application is concerned with the investigation of SAR data processors, a very primitive raw data generator with simple targets and ideal sensor may be adeauate for t . h e task. For this reason simulators tend to grow, starting off being tailored to a particular application and later being augmented to become a more comprehensive tool.
~~ ~~

~~

~~~~~~~~

The first step in simulating a SAR system is to model the target. The target can most easily be simulated, if it is described by a two-dimensional matrix of scatterers each modelled as a complex vector voltage reflection coefficient. Fig. 8 shows such a target model consisting of xy cells. As will be seen, t h e simulator only has to form the sum of a l l vectors within the instantaneous

the antenna. The raw data generation can be the most complex part of the simulation chain, even exceeding the complexity of the SAR processor. Generating such a two-dimensional array of vectors assumes that the scattering process lends itself to this reoresentation. Distributed taraets with differing normalised reflector coefficients oo can be synthesised with such a model, as can isolated point targets. Mixing distributed and point targets is also possible. The speckle associated with distributed targets can be generated with an appropriate statistical model. Most Simulators working in this way include utilities to help the user generate target patterns

~~~

run information and a series of warning messages when either algorithms are not functioning or when the user h a s made a mistake,
graphics software that allows the user to easily select graphs, automatically scaies and grids the data, plots requirement C U I Y ~ Sand, if applicable, fully labels the output and provides a rescale facility for comparisons with other graphs. In addition radar sensitivity data are given under each modelled swath point, book-keeping functions for the performance plots which allow the user to keep track of

The following description of the DLR simulator SARSIM will be used to illustrate such a simulator.
4.1 The System Simulator, SARSIM

SARSIM was originally developed by Ferranti in the UK in 1977 taking about 15 man-years of effort. The original software was extensively modified by Ferranti in 1978/79. bv Loaica iUK1 in 1981/82 and currently by the DLR since-1978. TWO versions of the program exist, one on the DLR CRAY and more recently on a VAX4000 workstation. SARSIM is described in [ 4 and 51

SIMULATED PARAMETERS

Sensor
Transmitter Antenna

PIatform
Attitude Flight path

Data handling

ProDaaation Path
Bit-errors

Attenuation

Tarqet
Geometry
Composition Scattering mechanism

SAR Processinq Facility

e
I

Raw data analysis Ranae comDression Corner turn Range migration correction
inrerpoiation

Muiti-iooking

EVALUATION

lmase Processinq
Verificationo f system design Influenceo f system oarameters Investigation of SAR imaging properties Develooment of mterDretation aiaorithms Training o f image interpretors "

Fig, 1 Simulated elements and evaluation tasks of a comprehensive SAR simulator

Fig. 9 Schematic of the SARSIM structure FOE each physical element of the system, there is a corresponding software module. These software modules simulate the physical element characteristics with the important exception of the, linear gain of the element. This means that the overall model should be seen a s a relative, rather than absolute, assessment tool. A more detailed schematic of SARSIM is given in Fig. 10. Here one can see that the simulator has t,een divided into two segments. The first segment, signal generation, deals with simulation of the SAR video signal, that is the SA8 system up until the point where the analogue signal (in baseband) comes out of the receiver system mounted on the platform. The second segment, data handling and SAR processing, then simulates the remaining on-board electronics, data-link, and SAR processing to produce the SAR target image. Auxiliary programs can then be used to create a hard copy of this image, apply post processing algorithms or to make image quality measurements. SARSIM has in-built options to generate several

Constant Range

Fig, 8 Representation of the simulator geometry with the target represented by two-dimensional array of complex reflection coefficients. A schematic of the SARSIM model is given in Fig, 9. SARSIM is an '"end-to-end" Simulator producing as an output the SAR image of the input target.

7-9

processing, where the Doppler history of the target is split into N segments, each segment is separately processed and then added incoherently after detection. This technique improves the radiometric resolution at the expense of spatial resolution. Finally the data can be corrected (calibrated) if desired. As an alternative to the built-in processor, SARSIM data can also be processed with the DLR motion compensation processor.
4.1.1 The Target

The target is represented by a matrix of complex voltage reflection coefficients assumed to lie directly on the earth's surface. Distortions and effects due to local height variations cannot be simulated. Similarly, variations of radar reflectivity with incidence angle and time cannot be simulated. The complex voltage reflection coefficient v, t iVi is related to the radar cross-section a of the corresponding ground element by the following equation,
V :
t : V
= ko

where k is a constant, The target size is restricted to 192,000 elements. The target generation segment of SARSIM consists of a two stage algorithm. In the first stage, a matrix of cells is generated according to some pattern corresponding to the ideal radar crosssection of the target. In the second stage, each cell is divided into a number (at least one1 of subcells and the value of the reflection coefficient for each sub-cell is derived from the value for the whole cell. This second stage permits the introduction of speckle into the model, by assuming that the amplitude of each sub-cell value is drawn from a Rayleigh distribution. The mean of the Rayleiyh distribution is specified by the cell value generated in the first stage algorithm. The phase of each sub-cell value can either be fixed or drawn from a uniform distribution ( - x , % ) Each sub-cell value can be thought of as representing the reflectivity from a point target located at the centre of the cell. Therefore, the target, even though representing an extended target, consists of a matrix of point targets. This is a standard method for modelling extended targets [ 6 1 . Superimposed on this, the user can specify up to 16 deterministic point scatterers. Fig. 11 illustrates a possible target example. Here, each cell has been split into four sub-cells. The amplitude, Ai of each sub-cell can be determined in one of three ways:

@-

..
Fig. 1 0 SARSIM configuration standard types of targets. Slope effects, such as layover and shadowing, are not simulated. The basic principle of simulation is a "pulse by pulse" simulation, that is the relative positions of platform and target, together with platform attitude are calculated for each radar pulse and then this is used to calculate the returned signal for each transmitted pulse. The returned signal is represented by a sequence of floating-point (complex) numbers obtained by convolving the transmitted pulse form with the target, weighted by the antenna pattern and taking into account any special receiver chain features. The sample spacing of the returned signal is chosen to match with the Analogue to Digital Converter (ADCI sampling frequency. A s the simulation is performed directly in baseband, RF system element effects must be carefully accounted for. After the ADC, the signal may be prefiltered or resampled if such an option is present in the real radar system. In Fig. 10, alternative options of on-board analogue range processing or on-ground digital range processing are shown, these simply reflect alternative positions of performing the range processing in the SAR system. The range processing (and azimuth processing) can be performed in the frequency or time domain. After the range processing the data is reordered (corner turn fcllowed by range migration correction) before the azimuth processing. Illustrated in Fig. 10 is multi-look

Fig. 11 Target examples. 1) Ai is fixed and proportional to 2)

One value per cell is drawn from a Rayleigh

7-10

distribution (mean and all Ai withj.n the cell take this value.
~

6)

In options B - E, values not explicitly specified in the pattern are set to a background value.
4.1.2 Geometry

Ai value for each sub-cell in turn drawn from a Rayleigh distribution. The same distribution is used for all sub-cells within the cell. Similarly the phase, i of each Sub-cell can be determined in one of three ways:
3)
An

1)

mi takes a Constant value for every sub-cell

in the matrix. One value per cell is drawn from a uniform distribution and a l l mi within the cell take this value. A m i value for each sub-cell in turn is drawn from a uniform distribution.

2)

3)

In the case where the sub-cell and cell are identical, then options 2 and 3 in both of the above are similarly identical. A cut-off point can a.lso be specified where option 1 is used for all azimuth bins greater than the cut-off point, and options 213 are used for the remaining azimuth bins. This is referred L O as a semi-diffuse target.

T WO models exist for specifying the geometry of the SAR system. The first model uses a flat earth model together with a platform model assuming either constant linear velocity o r stepwisfi constant acceleration components. This is suitab1.e for simulating an airborne SAR. The second model uses an ellipsoidal earth model, together with z:n ellipsoidal orbit model. The second model is suitable for spaceborne SAR simulations. In both models the platform coordinate origin ( x , y) is defined as the platform position when the beam centre and target centre are in alignment. The platform position is expressed in Cartesian coordinates. In the case of the ellipsoidal earth model, a local spherical approximation is made to the earth's surface, and the target is assumed to lie on this spheroid. This assumption is valid so long as only a few seconds of flight time are simul.ated. Earth rotation is taken into account by combining the earth's surface velocity with the platform velocity to produce a relative velocity vector, and modifying the platform yaw angle.

4.1.3 Platform Position and Attitude

The following options are provided far generation of the (first stage1 matrix of cells:
A)

User specified pattern: a two dimensional array of amplitude data is read in from a fi.le provided by the user. Cyclic pattern: a set of cycles is generated in either the azimuth or range direction. Each cycles consists of an alternating sequence of high refleczivity and low reflectivity amplitudes. The length of the sequences within each cycles is fixed, but increases by one cell between successive cycle-series. An example is given in Fig. 12.
1ST
ZND

B)

The platform position and attitude are calculated for every pulse in the time interval l-T.i2, T / 2 1 , that is the simulation is constructed time symmetrically about time Zero, t h e zero being the time when beam and target are in alignment. In the case of the elliptical orbit simulation, the orbit and geometry parameters are sufficient to dekrmine the theoretical position of the platform for every pulse. However, an additional perturbation may be introduced in the form of a constant linear acceleration term. The flat earth simulaiion, requires instead input data relating to the linear- velocity components at time zero and linear acceleration components before the platform position Fber pulse can be calculated. The platform attitude per pulse is expressed in yaw, pitch and roll angles. These can either be considered as constant or as conforming to some specified angular motion.
4.1.4 Transmitter In SARSIM, the transmitter i s ideally modelled, with the only feature characterized being the ideal form of the transmitted pulse. It is not possible to simulate any f o m of noise or distortion in the transmitter system. Mismatch ibetween the up- and down-converters is assumed to result in a residual carrier frequency and this is modelled in the receiver simulation.

3n0
C"CLE-SE"#ES

CIcLli-sIIII~s
AVPLlTUDL

C"CLL-SE",IS

AlMPLlTUDE

AMPLITUOE

Fig. 12 Cyclic target pattern.

C)

The following forms of transmitted pulse (can be simulated:

1)
2)
A

Linea: pattern: a set of point targets of lin e a r l y increasing amplitudes in one l o r more)

pulse with constant frequency

range bin or azimuth bin.

D)

Two point targets: two point targets of specified amplitudes are placed in either one range bi.n or one azimuth bin. Single point target: a point target of specified amplitude at a specified position. Checkerboard pattern: a pattern of alternating sqwresl rectangles of high amplitude and low amplitude. Square well PatLern: a uniform target with a square of zeio reflectivity inset.

A linear FM pulse; the frequency of t.he transmitted pulse increasesldecreases linearly during the pulse trahsmission. A quadratic FM pulse; the frequency variation of the pulse during transmission is a quadratic function of time. a binary phase encoded pulse; the frequency of the pulse is constant, but the phase can take two values: 0 and n .

3)

2)

F)

4)

C-)

The phase structure of the pulse is specified by a binary string.

7-11

In all four forms cf the transmitted pulse, the amplitude of the pulse is constant. 4.1.5 Antenna The antenna is defined by a two-way antenna gain pattern, specified in azimuth and elevation. The antenna gain function is assumed separable, that is the antenna gain in any given direction is obtained by multiplying the appropriate azimuth and elevation gains together. No provision is made for gain variations during the simulation period. The following gain patterns can be used in the simulation:

Fig. 13 Target element contributions to the mth range bin return using signal generation primary option (time domain). range bins, It should be noted here that the target cell range dimensions should be chosen equal to, or smaller than, the range bin width. This is to minimise problems occuring when few target cells happen to fall in a given range bin, a result cf the fact that the target is in fact modelled by a number of discrete point scatterers, rather than by a true extended target. The primary method of signal generation inherently includes all range migration effects due to earth rotation, earth curvature and radar wave front curvature. The simulation of the range modulation can either be performed in the time domain (by convolving the returned signal directly with the transmitted waveform), or in the frequency domain (by multiplying the appropriate Fourier transforms together and then taking the inverse transform). TWO secondary options for received signal generation exist, both of which attempt to reduce the amount of computer time required. The first of these options generates the expected return (phase history) from a point target located at the centre of the target matrix assuming an ellipsoidal, locally spherical, earth's surface and an unperturbed elliptical orbit. Appropriate account is taken of the azimuth antenna pattern. The azimuth modulation is performed in the frequency domain and so the Fourier transform of this phase history is used to calculate the return range gate values. The same phase history is used for all range gate values. Range migratidn effects are not accounted for in this model. Note that the target element spacing is assumed to be defined by the range bin width in range and the pulse repetition interval in azimuth. The second option extends the above idea to partly include range migration effects by taking into account that the return from a point target appears in several range gates, or alternatively expressed, that neighbouring target gates contribute to the return from one range gate (Fig. 14). The transition points between neighbouring rangeltarget gates are calculated at the centre of the target (mid-swath) and assumed to apply to all rangeltarget gates. The phase history segments are convolved with the target in the time domain. In both of the secondary options the range modulation can either be performed in the frequency or time domain, as in the primary option signal generation.
4.1.8 Receiver

1)

Uniform pattern

= 1

2)

Modified cosine pattern. cos (XI , where

G =
($2
~

= -

nd

x2

sine

where:

3)

D is the effective antenna aperture (in azimuth or elevation), h is the radar wavelength, I 3 is the azimuthlelevation offset angle. Sinc pattern.
=

G
4)

where x is defined above

User specified pattern. The user must specify a table of gain against azimuthlelevation offset angle. The beamwidths of the antenna pattern (in azimuth and elevation) must be specified by the user. These values define the extent of the generated antenna pattern.
4.1.6 Propagation

The radar equation is not included in the SARSIM model. Most of the terms in the radar equation are constants (for most system simulator applications), apart from the range gain dependence (r41. For satellite simulations the lack of range dependent attenuation is thought to be of negligible significance (though of more significance for airborne applications). Atmospheric distortions are only included in as much as the optional addition of a phase noise term. The noise follows either a uniform, Rayleigh, exponential, or normal distribution: in the case of the normal distribution an auto-correlation function can also be specified. The autocorrelation function can either be linear, quadratic, exponential, Gaussian, or User supplied.
4.1.7 Signal Generation

The primary option for received signal generation generates the returned signal for each pulse in turn in the time domain. The distance to each target element is calculated and then the returns (after weighting by the appropriate antenna gain1 from all target elements within a range gate are summed, taking into account the appropriate phas e s , to give the final return range gate value. This operation is performed for all range gates within the sampled return pulse and for all pulses simulated and is, hence, not an insignificant task. Fig. 13 illustrates the target returns which contribute to the return range gate value.
The . h e curved radar rancle bin de. . . . exact form of t
~~~~~~ ~~~~ ~ ~~~~ ~ ~~

pends on the orbit and earth geometry Simulated. The width of the bin is determined by the sampling frequency of the ADC, which, together with the receiver gate times, also determines the number of

The receiver system typically consists cf a low noise amplifier a down-converter, an IF amplifier, and an IQ detector. The low noise amplifier is assumed to be perfect apart from the addition of thermal noise. The down-converter is similarly assumed perfect apart from the possibility of a residual cariier frequency resulting from a mismatch between the up- and down-converters. This residual frequency produces an additional term in the un-

7-12

compressed radar pulse. The IF amplifier is chzracterised by a filter res?onse and its non-linearities. The filter characteristics can either be specified by the user in the form of a gain against input signal ampli.tude table or by an analytical expression. After the IF amplifier comes the IQ detector. This is modelled with the following error sources:

Fig. 15 Optimum equally spaced level quantiser quantisation threshold.

21

Quantise the data. Rescale the data to cornpensate ?or scaling introduced in I ) , if required.

3)

4.1.10 Prefilter/Presumer

* * * *

phase and amplitude errors, orthogonality error, DC offset, gain imbalance.

Receiver system phase noise can be introduced immediately prior to the IF emplifier. The receiver noise form (both phase noise and thermal noise from the low noise amplifier) follows either a uniform, Rayleigh, exponential, o r normal distribution: in the case o f the normal distribution an auto-correlation function can be specified. The auto-correlation fmction can either be linear, quadratic, enponenrial, Gaussian, or user SUPplied.
4.1.9 Analogue to Digital Conversion

The purpose of the digital prefilter is to reduce the quantity of data to be transmitted or stored. The filter can operate either in the range or azimuth directions, or on the transmitted chirp replica. The filter can either operate on both channels (full quadrature1 or just on one (real) channel Inon-quadrature) by setting the remaining (imaginary) channel to zero. The prefilter is assumed perfect apart from errors introduced by the finite word length within the digital filter. Word-length effects (both before and during the filter1 may be simulated by specifying the number of bits available in the digical filter registers.
4.1.11

Datalink

The Analogue to Digital Converter (ADCI can be thought of as consisting o f sampling and quanti-. sing operations. The sampling operation is inhe-. rent in the SARSIM model, as the analogue signal is represented by a set of iiscrete samples, each sample consisting of a (complex) pair o? floating point numbers. The quantisation process then converts these floating-point numbers to a finite set of quantisation levels consisting of both posirivr and negative values. Values exceeding the largest qunatisation level (ignoring sign) are set to this l e v e l . The nlimber of overflows is counted. Quancisation of both the radar signal and transmitted chirp replica can be simulated. The quantiser itself can be specified in two forms:
1.

The datalink between the an-board system ,ind ground based processor can either take thf? form of a telemetry link (satellite SAR) or a Higll Density Digital Tape (aircraft SARI. The model restricts the datalink simulation to allowing the user to specify a link transmission u.ord length and then simulating bit errors over the link. The bit errors can either be Stochastic, where the spacing between bit errors is a random variable di~awnfrom an exponential distribution, or deterministic, where a user-specified cyclic table (of up to 50 values) is used to determine the distance between one bit error and the next. The setting o f the bit in e r r o r is switched. Tie datalink is simxlated by the following algorithm: 1. Scale and quantise the data to the relevant link transmission word length, if necessary. The scaling factor must be defined by the user. Insert bit errors. Rescale the data to compensate for scaling introduced in (11, if required.

The user specifies up It0 14 threshold values and corresponding output levels.
The user specifies the nurrher of bits n, thereby defining the number of quantisation levels 1 2 n l . The optimum equally spaced symmetric quantiser is used as shown in Fig. 15.

2.

2.
3.

The qdantisation process consists of the following steps:

11

4.1.12 Processing

Scale fhe data and add a DC offset, if required. The scaling can either be user-specified or automatic. For autonatic scaling, the ma.ximum, mean or mean squared, of the samples in the first p i l s e , is adjusted (scaled) to be a user-specified fracticn of the maximum

For a frequency modulated or "chirped" pulse, the first step in the SAP. processing is to rem,svethe frequency modulation. This is knoivn as range processing and can either be performed an-board in the analogue system using a direct replica of the

transmitted pulse, or, on-ground using either a replica O L theoretical values for the reference function generaion. As far as the simulation model is concerned, the difference in these two approaches to range processing is apparent only in the fact that the range processing simulation has two possible positions in the radar data chain (Fig. 10). Range processing can either be performed immediately prior to the ADC (on-board analogue range compression), o r immediately after the datalink (on-ground digital range compression). It would also be possible to simulate on-board digital range compression if required. The first step in the range processing is to generate a range reference function or copy of the transmitted pulse form. There are three ways of doing this: 1. Ideal replica. A copy of the transmitted pulse form used in the transmitter segment is generated. Standard reference function. The possible forms of the reference function correspond to those possible for the transmitted pulse, i.e. linear FM, quadratic FM and binary phase encoded. The user must specify the appropriate reference function parameters. User-specified reference function

range naint response :*,get

i-/

range

2.

,--i

7-13

Fig. 16 Range Migration Correction migration effects in the time domain.

For each range bin in turn, the range walk for a

target cell is calculated and applied. The range walk is calculated by considering the range history used in the azimuth reference function generation, and so the range walk function is updated whenever the azimuth reference is updated. The feature that distinguishes SAR from other imaging radar systems is its ability to achieve high azimuth resolution through coherent integration of the returned radar signal, AS a target cell passes through the beam the phase of the returned signal changes, and this can be picked out by convolving the (azimuth) returned signal with the appropriate reference function. The convolution can be performed in one, or split into segments (subapertures or looks), each segment being convolved separately. The resulting looks are detected (amplitude taken) and summed to give, what is commonly called, a multi-look image. The simulator can perform all these operations, starting with the azimuth reference function generation. The following reference functions can be generated, either as one reference function (single look processing) or as several subaperture reference functions (multi-look processing) :
1.

2.

3.

Once the range reference function has been generated it is conjugated, and then used in the compression. It is also possible to specify a reference function weighting. Compression can either be performed in the frequency or time domain. F o r frequency domain compression, the reference function and signal must first both be transformed to the frequency domain, multiplied together, and then inverse transformed back to the time domain. For time domain complession, it is necessary to directly correlate the reference function with the signal. TWO methods of time domain correlation are possible in the simulation model: the overlap add technique and overlap Save technique. These techniques are specially designed for the case where one function (reference function) is considerably shorter than the other (range line), and are fully described in ( 7 1 . After range processing, the radar data on disk must normally be re-ordered before the range migration correction and azimuth processing can be performed. Up until this point the radar data file has consisted of successive range lines or azimuth bins. The azimuth processing requires the data are ordered so that the data file consists of successive azimuth lines or range bins. This process is called corner turning. Range migration is caused by the fact that the distance (hence time delay) to a target cell changes as the target cell moves through the beam. This is due to earth rotation, curvature of the earth, and the ellipticity of the orbit. The end result is that the response from one target cell is not in one range bin, but split into bands over several range bins. Therefore, before the radar can be processed in the azimuth direction, it is necessary to re-order the data so that the response from one target cell is observed only in one range bin (Fig, 16). This must be repeated for each pulse and across th swath, and can involve some form of interpolation. Range migration itself can be split into two components: a linear component (range walk) representing the difference in range to a target cell between when it enters the beam and when it leaves the beam, and a higher-oder terms component (range curvature). The simulation model has an algorithm to correct for only the linear

Curved Earth Time Invariant: the same phase history is assumed to apply to all target cells in a given range bin. The user must specify the bin number before the reference function can be calculated. Flat Earth Time Varying: the phase history for an individual target cell is calculated. The user must specify the coordinates of the target cell, This option allows several different azimuth reference functions to be convolved with each range bin, by generating reference functions for target cells with different azimuth coordinates within the bin. Curved Earth Time Varying: as for option 2, but for curved earth geometry.

2.

3.

In all three of the above options the azimuth reference functions are generated for a particular range bin. The reference functions generated though, may be applied to all range bins, or alternatively updated (new reference function generated) every n range bins, as required, where n is specified by the user.
The platform geometry used in the azimuth reference function generation is, in the basic case, the same as that used in the returned radar signal generation, i.e. it is assumed that the platform position and attitude are exactly known. Errors can be simulated by generating a new platform data file, specifying different geometry andlor error values, this is then used for the reference function generation, or, more simply, by specifying geometrical or Doppler errors. In the case of geometric errors, the user specifies slant range and yaw angle errors, which are directly used in the reference function generation. For Doppler errors, it is necessary to convert the Doppler errors to equivalent range and yaw errors. Doppler errors

7-11

a r e s p e c i f i e d i n terms of a Doppler s l o p e e r r o r ( s l a n t range e r r o r ) and a mean Doppler e i r o r (yaw angle e r r o r ) . The azimuth r e f e r e c c e f u n c t i o n can be weighted. Once t h e a p p r o p r i a t e azimoth r e f e r e n c e f u n c t i o n s have been generated, it i s necessa:y t o convolve them rvith t h e range migration c o r r e c t e d s i g n a l dat a . it i s poSSi?hle t o perform t h e azimuth process i n g i n e i t h e r t h e frequency o r time domain, using e x a c t l y t h e same techniques and algorithms a s j.n t h e ranae o r o c e s s i n a . After azimuth orocessina. t h e s i g n a l ' d a t a a r e - d e t e c t e d , t h a t is t h e m o d k i (or moduli squared) of t h e complex values a r e t a ken, and, i n t h e case of multi- look processing, t h e i n d i v i d u a l looks a r e summed. Word length e f f e c t s i n t h e summation can be simulated, and t h e summation can be weighted ( f o r example t o accommod a t e antenna gain p a t t e r n e f f e c t s ) . 4 . 1 . 1 3 SARSIM Application 3xample
I n t h i s s e c t i o n an example of a s a t e l l i t e SAR s i mulation i s given. The c o n f i g u r a t i o n of t h e system is based on t h a t f o r :he ESA Remote-Sensina s a t e l l i t e (ERS-1). E r r o r s and n o i s e Sources w i t h i n t h e system a r e not simulated i n t h i s example.

F i g . 1 9 Signal a f t e r azimuth compression seri.es of small p l a t e s or f a c e t s e n a b l i n g t h e t a r g e t t o be e n t e r e d a s a complex shape. The model is capable of handling hidden s u r f a c e s ( r a d a r shadaws) and can generate p o l a r i m e t r i c s i g n a t u r e s , a f e a t u r e of i n c r e a s i n g i n t e r e s t f o r t h e next gener a t i o n of SAR s e n s o r s . 5 . CONCLUSION

Fig. 1 7 shows t h e radar t a r g e t used i n t h e example s i m u l a t i o n : a s i n g l e p o i n t t a r g e t on a non-reflect i n g background. Fig. 1 8 i s a p l o t of t h e range compressed SAR d a t a . F i g . 1 9 shows t h e f i n a l image p l o t of the single point t a r g e t .

Parametric a n a l y s i s and simulation s o f t w i r e packages f o r i n v e s t i g a t i n g SAR systems have been desc r i b e d and t h e i r m e r i t s d i s c u s s e d . The u s e f u l n e s s of s i m u l a t o r s i s o f t e n c a l l e d i n t o q u e s t i o n a s t h e c o s t and manpower i n v e s t e d can be l a r g e . I t i s , t h e r e f o r e , worth considering two a p p l i c a t i o n s of t h e SARSIM simulator.

j . , , :

, .

f o r t h e a p p l i c a t i o n i t was designed fo1. The wave mode images small a r e a s of t h e ocean s u r f a c e , t h e d a t a being s t o r e d an-board t h e s a t e l l . i t e t o achieve g l o b a l coverage. From t h e images, t h e ocean wave spectrum was t o be derived by g e n e r a t i n g t h e two-dimensional Fourier power spectrum. The l i m i t ed s t o r a g e c a p a c i t y of t h e recorder meant t h a t t h e imaged a r e a was l i m i t e d t o approx 2 Y 5 km. T h i s a r e a was found t o be t o o small t o d e t e c t long waves. The only p r a c t i c a l s o l u t i o n was t o reduce t h e amount of recorded d a t a by halving t h e nu.nber of b i t s used f o r q u a n t i s i n g each SAR d a t a sa.np1e from 4 b i t s 1 1 4 b i t s Q t o 2 b i t s I12 b i t s Q . Analysis showed t h a t q u a n t i s a t i o n n o i s e would be a ' x e p t a b l e , but would s u f f i c i e n t information be r e t a i n e d t o d e r i v e t h e wave spectrum?

. .
1

., . ... .. .
.. .

The problem was analysed using SARSIM [ 9 1 . A n ocean scene was simulated (Fj.9. 20) and t h e image (Fig. 21) used t o generate t h e power spectrum (Fig. 22). Analysis of t h e l a t t e r showed lthe wave Spectrum was s t i l l reproducible. The new q u a n t i s a t i o n scheme was introduced and today t h e ERS-1 s a t e l l i t e has a coverage of 5 x 5 km i n t h e wave mode.

F i g . 1 8 Signal a f t e r range compression.

4 . 2 Simulator with Facet Backscattering Model

The above example shows how simulation can remove f i n a l doubts even if a n a l y s i s of t h e prob:.em i s successfull.
A n o t h e r example a l s o concerns q u a n t i s a t i o n n o i s e . When p l o t t e d a g a i n s t i n p u t s i g n a l power, t.he quant i s a t i o n noise shows a d i s t i n c t minimum i n c r e a s i n g a t l o w Dowers, due t o t h e d i s c r e t e a u a n t i s a t i o n s t e p s and a t high powers, due t o s a t u r a t i m ( s e e Fig. 23 and F i g . 2 4 1 . These curves were d e r i v e d t h e o r e t i c a l l y [lo] ( d o t t e d ) and SARSIM was; used t o confirm t h e r e s u l t s . The simulation r e s u l t s a r e joined by dashed lines
~ ~~~

The two-dimensional r e f l e c t i o n c o e f f i c i e n t t a r g e t model used i n SARSIM i s adequate f o r many a p p l i c a t i o n s , f o r examole sensor i n v e s t i a a t i o n s b u t i t i s very l i m i t e d i f s c a t t e r i n g mechanisms of complex t a r g e t s a r e t o be examined. The f a c e t backscatter i n g model used i n t h e SARAS s i m u l a t o r , developed by Naples University ( 8 1 models t h e t a r g e t a s a

7-15

P,~nrl,N~nlnrl

O a l n

e Iinollfion

iCIYItI

Fig. 24 5-bit quantisation noise curve. matched filters in the SAR processor. This is an example of simulation throwing up surprises and leading to further investigations.
6 . REFERENCES

111 Hounam, D. Pierschel, D.

Derivation of the Technical Specification of the ERS-1 Active Microwave Instrument to Meet the SAR Image Quality Requirements, Proc. of IGARSS '87 Symp., Ann Arbor, 18-21 May 1987.
IMPRES, A Program to Generate the Impulse Response of a Synthetic Aperture Radar. DLR-IB 551-3192.

[2] Chorherr, G. Hcunam, D. Fig. 21 S A R image of the scene in Fig. 23. [31 Hounam, D. Potter, S. Schmid, R.

Algorithms and Software User Guide for the Performance Estimation of Synthetic Aperture Radars. DLR-IB 551-4192. SARSIM: A Synthetic Aperture Radar System Simulation Model. DFVLR-Mit. 85-11. Analysis of ERS-1 SAR Performance through Simulation. PTOC. IEEE ' 8 6 National Radar Conf., Los Angeles, March 12-13, 1986, PP. 12-18. Principles of Radar Measurement. Proc. EARSeL Radar Calibration Workshop, Alpbach, Austria, December 1982, ESA SP-193, pp. 11 20. Theory and Application of Digital Signal Processing. Englewood Cliffs, NJ: Prentice Hall 1975. SARAS: A Svnthetic Aperture Radar (SAR) Raw Signal Simulator. IEEE Transaction on Geoscience and Remote Sensing Vol. 30, No. 1, Jan. 1992.

[41 Pike, T.K

[51 Pike, T . K .

161 Krul, L.

171 Rabiner, L.R. Gold, 8 .

It31

s i . " . l i N . . l " . ,

Cain

Franceschetti. G. Migliaccio, M. Riccio, D. Schirinzi, G.

~~~ ~~~~~

Fig. 23 2-bit quantisation noise curve

For both the 2 bit case (Fig. 23) and the 5 bit case (Fig. 24), the simulated values are much les.2 than the theoretical results. An at least partial explanation of this is the non-white nature of quantisation noise [ll] and the influence of the

7-16

19: !.lolframTt, A.P Pike, T . K .

Q u a n t i z a t i o n Study i o 1 EXS-I Wave Node. DFVLR Research Report. DFVLR-FB 84-39, October 1984. A n Optimal Q u a n t i s a t i o n cf Coherent Radar Echoes from T e r r a i n o r Sea Surface. DPVLR, I n s t i t u t fiir Hochfrequenztechnik, Report Sept. 1 9 8 4 . Simulation and S t u d i e s of Spaceborne S y n t h e t i c Apert u r e Radar Image Q u a l i t v with Reduced Bit Rate 15th I n t e r n a t i o n a l Sympos i u m on Remote Sensing o f t h e 3nvironment, Ann AIb o r , May 1 9 8 1
~ ~~

I101 Sappl, E .

[11] Li, F.

Held, D. Hofieycutt, Zebker, R .

8.

%I'O 5
%1'0 7

Iex!d 811 5

w 001 7

ool

BP D'E T 8PD'LT BP S I F BP 0 1 T

un 06 01 wn

SI

EP 0 3 7 BP L1- 5 EP PI- 5
%OZ 5

I -R

x-?
sion Laboratory (JI'L) under contract with NASA [Curlander, 91bI. Therc are two European space agencies working on t h e X-SAR processor development: the German Aercspace Research Estahlislrmcnt (DLR) and the Italian Space Agcncy (ASI) [Runge, 901. The major challenge to the SIR-C processor design is to cope with a large number of radar modes. Nominally the SIR-C science team has selccted Seventeen data acquisition modes from all the possible combinations with eight radar chanriels (Scc Tablc 3); two pulse bandividtlis and three data

multi-band, polarimetric SAR image [Freeman, 691, [Klein, 921. For radiometric calibration, the built-in-tc:t-equipment (BITE) d a t a are designed for probing the 1~eall.hof the antenna, receive only noise data for estimating the noise power, calibration tone signal for monitoring the receiver gain and temperature measurements and T / R module failure in for ma^ tion transmitted via the downlink telemetry. These ancillary data and calibration site data are essential t o ilerive radiometric calibration parameters which are applied during tlie data processing t o produce calibrated image product. For geometric registration, special consideration is required in the processor design to ensure that the ootpul. images are registered in both cross-track and along~trackdimensions. The remaining paper prescnts an overview of thc Slll-C endb e n d ground d a t a processing systcm design, which includes input d a t a format specifications. system operal,ions dcsign, data products design, processing algorithm d e s i p , hardware arcliitccture design and software design At th:
ciid

quantization formats. Additionally, data will be collected in two nominal attitudcs over incidence angles fronr 17" to 63' with a variety of antenna elevation patterns controllable to provide beam spoiling a t tlie stecp incidence angles. The largc number of radar modes complicates tlic logics in l i a n ~ dling difcrent typcs of data format
aid

iiicmwcs the scope

of tcstirig thc iritegratcd processor noftivare. The second challenge to the SIR-C processor design is to
cope with the large attitude uncertainties and high attitude

of tlic

paper, we give a brief summary of the s t a t u s id Iplan for

drift rates of the space shuttle platform as shown in T:tble

the processor development.

2. The large altitude uncertainties create P R F ambiguity

problem in Doppler centroid estimation and large Doppler errors for data acquired over high terrain relief areas. Special

2. INPUT DATA FORMAT

T h e signal d a t a is recorded across four recorder channels on t h e High Density Digital Cassette (WDDC). The data rate for each recorder channel is 45 Mbps for a total of 180 Mbps. T h e signal d a t a is quantized into one of three types of format: 4-bit, 8-bit or (8, 4) block floating point quantization (BFPQ). Nominally, the data is collected over a period called data take using t h e same set of commanded radar parameters. T h e length of the data take varies from 3 minutes t o as long as 15 minutes for ocean s h e data. T h e avtrage length

techniques (multiple P R F technique and attitude steering technique, respectively) are required to resolve the problems [Chang, 92a.1, [Chang, 92b]. T h e high attitude drift rates induce fast Doppler drifts in both cross~trackand along-track dimensions. Frequent Doppler update is required to maintain the image quality, which complicates the geometric rectification proccdure to produce
R

seamless image.

processor design is t o proThe third challcnge to the S I R ~ C duce radiometrically calibrated and gcometrically registered

is estimated t o be between 4 and 5 minutes.

Table 3: SIR-C radar data aquisition modes

T h e nominal SIR-C data take consists of a turn-on sequence, followed by t h e science data collection and a turn-off sequence as shown in Figure 1. The first four seconds of the turn-on sequence consist of (one second each): receive only noise data, caltone scan data, low noise amplifier (LNA) BITE d a t a and high power amplifier (III'A) BITE data. These four second data are used for radiometric calibration. The remainder of the turn-on scquence consists of one second of PRFA data and one second of PRF8 data which together with t h e first second of PRFc data are used for resolving P R F ambiguity in estimating the Doppler centroid frequency [Chang, 92a]. The system remains on PRFc for collection of t h e science data. The turn~offsequence is simi-

-. :HA"EL>HANNEL--;HANNEL-l
1

:HANNELLHH LVH CHI WH LHH

LHk

LHV

LHV

2 3
4 5 6

7
8 9 10 11

12
13 14 15 16
_ _ I

LVH CHI CVH LHH CHI LHH LHH LVV LVH LHH LVH

LVV CHV

LVV CHV

cvv
LVV

cvv
LVV

an
LM

cvv cvv
CHI

cvv cvv
CHI

LHk
LHH CHI LVH. LHH LHH C H I

LM LVV LVH LHV LVV LVV LVH CVH

cvv
CHI CVH CHI LHV

cvv

. CVH
CHV

cvv cvv
LVV

CVH, CHH CVV. CHV LVV, LHV -LVH CVH LHV CHV
~ ~

E. cvv
LVV

lar t o the turn-on sequence in that the science data collection

is followed by one second each of PRF8 and PRFA. T h e last four seconds of the turn-of sequence are receive only
noise data.

; : : ;3 ?

---

cvv

x-3
Null-Lines at wary one second lime tick

lime
Turn-Off Sequence

LA---Turn-On Sequence
Target
uenee

fm=f
C

Fau

CAL Scan

LNA BITE
or

HPA BITE
or

PRF
A

PRF

, p F 6 ; w m m

m m
t 6 seconds

t Average

f f

4.5 minutes 4 t 6 seconds

Figure 1: S i R C input data run format. Each Segment in turn-on and lurn-off sequence is 1 second duration. RON: Receive Only Noise, LNA: Low Noise Amplifier, HPA: High Power Amplifier, CAL scan: Caltone Scan.

At every one second time tick, a null-line is inserted. T h e

null-line is obtained by setting a half of the phase array el^ ements with a 180 phase difference to create a null around the ccnter of the antenna elevation pattern. The n r i l l ~ l i n ris used to estimate the shuttle roll angle drift.

4. DATA PRODUCTS DESIGN

T h e SIR-C output data products include three image products: survey image, standard multi-look image and standard single-look image; and one reformatted signal data product. T h e throughput requirements are to produce 24 survey im-

A sinusoid waveform, called the calibration tone (caltone)

signal, is injected in the receiver electronics and recorded together with the return echo data. T h e caltone is used to estimate the receiver gain change as the temperature varies.

age products per week during the phase 1 operations and

to produce 9 standard multi-look image, 1 single-look image and 1 reformatted signal data products per week during the phase 2 operations. T h e expected processor throughput far exceeds the requirements.

3. SYSTEM OPERATIONS DESIGN

T h e survey image is a 4-look, single-polarization strip image, Operations of the SIR-C processor is comprised of two main phases: phase 1 survey processing and phase 2 standard processing, which last for a total of one year. During the phase 1 operations, a quick-look survey processor is employed to process singlefrequency band, single-polarization channel data

stored in the byte amplitude format. T h e image is deskewed to zero-Doppler and resampled to the ground range domain with a 50 meter pixel spacing. T h e resolution is approximately 100 meters. T h e length of the survey image is equal to the length of the data take. The average length is approximately 4.5 minutes or 2000 Km. T h e survey image will be recorded on Alden thermal prints and CD-ROMs. The CDROMs will be distributed to all the principal investigators

into low resolution strip images. These survey images will cover all the SIR-C ground sites albict with a single radar channel. By-products of the survey processor include unambiguous Doppler centroid estimates history and roll angle estimates history. Additionally, during the phase 1 operations, some selected data segments (covering calibration sites) will he processed into single-look, full-resolution complex imagery. These data will be analyzed to derive the parameters used for antenna pattern generation and polarimetric calibration. These parameters will be applied during the phase 2 standard processing to produce phase and amplitude calibrated~data products [Freeman, SS], [Klein, 921. The system operations schedule is planned as follows. Six weeks are allocated far processor check-out upon receipt of the first signal d a t a tape. Phase 1 operations will begin following the completion of the processor system clicckwmt and last for a period of twelve weeks. This is followed by phase 2 operations for a period of forty weeks.

(PIS).
T h e standard multi-look image is a multiple look, polarimetIic (single., dual- or quad-polarization) frame image. T h e image is deskewed to zero-Doppler and resampled to the ground range domain with a 12.5 meter pixel spacing. T h e azimuth resolution is chosen to be 25 meters. The range resolution is chosen to be 25 meters or the natural resolution if greater than 25 meters. T h e image data is stored in a compressed cross-product format [Dubok, S9). The basic frame
size is chosen to be

100 Km. The image will be recorded on

Kodak prints and CEOS formatted tapes. T h e standard singlelook complex image is a single-look, polarimetric (single., dual- or quad-polarization) frame image. T h e image is processed to full-resolution, deskewcd to zero-

Doppler and presented in the slant range domain in natural pixel spacing. T h e image data is stored in a compressed scattering matrix format. The basic frame size is chosen tc be 50 Km. T h e image will be recorded a n CEOS iormat,ted tapes and a reduccd, detected image will he printed by the Kodak printer. The reformattcd signal data contains the signal data reformatted in the range line byte format. The signal data together with the decoded radar parameters will be stored on

is performed using the spectral analysis (SPECAN) algorithm which requires fewer azimuth FFT's tha.n the traditional matched filtering algorithm. Following a:iimutlr
C O ~

prcssion, radiometric correctiaii is applied to compensate for the along-track radiometric modulation. This is followed by
a geomctric rectification step that resamplcs the slant rangc-

Dapplcr image into the ground range cross-track arid alongtrack domain. T h e rectified burst images are !IICII to producc the final multi-look st.rilj imagc. overlaid

CEOS formatted tapes.

5. P R O C E S S I N G A L G O R I T H M D E S I G N

For llic survey processor, tlic initial ilopi>lrr m n t r o i i

Si<:-

quency is determined using a clutterlock routine .and a ambiguity resolution technique t h a t requires a multiple P R F data collection a t the start of each data take [Chang, 92aJ. The unambiguous Doppler centroid frequency is then tracked by

5.1 S u r v e y P r o c e s s i n g A l g o r i t h m
. burst mode processThe SIR-C survey piocnsor utilizes x

a burst mode clutterlock algorithm during the d;rta

ing algorithm [Sack, E5], [Curlander, Sib]. The algorithm
flow chart is shown in Figure 2. The survey processor is de-

process^

ing. T h e Doppler frequency rate is solely derivt:d from the ephemeris parameters. Analysis results show that the accuracy of the ephemeris is sufficient for generation of survey products without employing the autofocus routine.

signed t o process an entire data take into a strip imagc in approximately oncscventh the real timc data collection ra.tc. To attain high throughput rate; the data is bursted i n aaimuth (slow time) with aone-quartcr duty cycle factor. T h e data volume i s further reduced by a factor of four in range (fast time) by processing the data using only orie~quartcr of the range chirp bandwidth. The azimuth compression

5.2 S t a n d a r d P r o c e s s i n g A l g o r i t h m

Prior t o standard processing, preproccssing is einployed to iteratively refine the Doppler centroid frequency and the Doppler frequency rate estimates using clutterlock and autofocus techniques [Li, 851. Doppler centroid frequency is estimated from thc azimuth spectrum by locating the energy centroid. Doppler frequency rate is estimated from the look registration crror by azimuth cross-correlating the look1 and look-4 images obtained by spectral division. Identical

segment of data

Null Lines AdJacenI Lines

Bursted

Range Lines

c
Null-Line Processing
roll

Callone Processing
callone

gain estimate

angle slimale Generation

1 4 -

Dopplcr parameters are uscd for processing all p'nlarimetric

Compression

data channels to ensure t h e phase cohcrency required for thc polarimetric data analysis. This approach will resilt in some increase in azimuth ambiguities if the antenna beams are not exactly aiigncd. T h e range-Doppler processing algorithm (;.e., the rectangu~
lar algorithm) with secondary range compression and fie^

Cross-Track

Radiometric Compensation

qiiency domain range cell migration campensation was selected by SUI-C for standard processing [Wu, 821, [Jin, 841, [Curlander, Sla]. The algorithm Rowchart is shown in Figure

3. Tha range compression and azimuth comprcssion matched

filtering operations arc performed using tho frequency domain fast convolution technique. All the signal data, sir+-look,
iii~

dependent of the final products, are initially proccssed to complex imagery using the frill azimuth processing bandwidth. This is followed by an azimuth deskew op-

S","ey Image

eration where the resulting deskewed, single-look complex image is then radiometrically corrected. Following standard processing, postprocessing is employed to generate the final image product [Curlander, 91bI. Data

Figure 2: Survey processing algorithm flowchart

8-5

reduction is the only postprocessing function for generation of single-look image products, where the data reduction function is applied to t h e scattering matrix. Major postprocessing functions for generation of multi-look image products include cross-product generation, multi-look filtering and data reduction where the multi-look filtering combines multi-looking as well as geometric rectification functions. For SIR-C, all the multi-look images will be filtered to a 25 m resolution in azimuth and a 25 m or natural resolution in range. T h e pixel spacing is selected to be 12.5 m in both range and azimuth. T h e filtering is applied to the crossproducts. T h e data reduction function is then applied to the multi-look filtered cross-products data.

6. H A R D W A R E A R C H I T E C T U R E DESIGN
Figure 4 shows the hardware architecture design of the SIR-C ground data processor. T h e entire processor system is composed of seven subsystems. T h e Data Transfer Subsystem (DTS) performs raw data reformatting and line synchronization. T h e SAR Carrelator Subsystem (SCS) processes the SAR signal data. into survey and standard image data. The
azimuth deskewed. single-look complex image

Output Products Subsystem (OPS) performs image data reformatting, recording and display. T h e Control Processor Executive (CPX) controls the processing sequence of the above three subsystems. T h e Catalog Subsystem (CAS) stores t h e information concerning the processing request and processor status into database. The Calibration Subsystem (CAL) is used for generation of calibration parameters and analysis of calibration site image quality. The Radar Data Center (RDC) archives all the output data products. T h e SCS consists of a STAR array processor with three computational modules, an Alliant FX/8 mini-supercomputer with eight compute elements and an Alliant FX/2800 minisupercomputer with twelve 860-based CPUs. The STAR array processor is the main compute engine for survey procsSOT.

Figure 3a: Standard processing algorithm flowchart.

azimuth deskewed, single-look complex image

Mulil-Loo1
Complex

Its FFT performance is measured at 120 MFLOPS using

Oversampling and Slant-toGround Range Conversion

Oversampiing and Siant-taGround Range Conversion

three computational modules. T h e Alliant FX/8 is primarily used for standard postprocessing functions. Its aggregate

FFT performance ness 20 MFLOPS. Two SKYBOLT accelerator hoards are installed to speed up the FX/8 computer, which provide additional 100 MFLOPS compute power. T h e

(scattering

(intensity

FX/ZSOO is the main compute engine for standard processor and standard preprocessor. Its aggregate F F T performance

is measured at over 300 MFLOPS. Computational tasks are

Reduction (cross-product

distributed over computers for concurrency processing in order to provide maximum processor throughput. T h e DTS consists of a high density digital recorder, a D E MUX and two data quality analyzers (DQA). The DEMUX is used for selection of recorder channel for data processing.

Figure 3b: Standard postprocessing algorithm flowchart.

T h e DQA is used for line synchronization and verifying the

x-6

Figure 4: SIR-C ground data

processing system hardware architecture.

rically registered. Another challenge to development of this large software based system is its complex interlaces among the many software programs. Clear interface definitions are essential to successfully deliver the operational system on schedule. Currently, we are in the middle of developing all the processor software. Major computer hardware will be installed by summer 1992. T h e end-to-end system integration will take place in early 1993. T h e entire system is scheduled to begin oDerations in late 1993. Development of the SIR-C processor inherits a great deal of experience from the previous and existing spaczhorne and airborne SAR processors, such as SEASAT, SIR.-B, and J P L AIRSAR. Experience accumulated from the SIR-C proces-

data quality and integrity. The OPS consists of Exabyte tape drives, Alden thermal printers and Kodak color printers. Three subsystems, OPS, CAS and CAL, run on three

separate SUN Sparc workstations. The image display and operator interface display are handled via X-terminals.

7. S O F T W A R E DESIGN
There are a variety of software packages used for developing the SIR-C processor due to the need of specific applications. T h e major part of the signal processing software is written in FORTRAN while the input and output formatting software is written in C. T h e image display software is developed using X-library routines. The operator interface software is developed using a graphics user interface software called Teleuse which runs on top of MOTIF. T h e image annotation is created using a commercially available software package called PV-WAVE. T h e catalog subsystem software uses both FORTRAN and INGRES.

sor will certainly benefit future processor design and development, such as EOS SAR and RADARSAT. ACKNOWLEDGEMENTS

8. S U M M A R Y

T h e research described in this paper was performed at the Jet Propulsion Laboratory, California Institute of Technology, under t h e contract with the National Aeronautics and

Design and implementation of the SIR-C ground d a t a processing system is quite a challenge due to the Large number of radar modes and the large attitude errorslhigh attitude drift rates. In addition to the correlation software, there are many software programs required for deriving parameters from the ancillary data in order to ensure that the output image products are radiometrically calibrated and geomet-

Space Administration.

REFERENCES
[Chang, 9 2 1 . C.Y. Chang and J.C. Curlander, "A.pplications

of the Multiple P R F Technique to Resolve Doppler Centroid

8-7

Estimation Ambiguity for Spaceborne SAR, " IEEE Transactiops on Geoscience and Remote Sensing, September 1992. [Chang, 92bI C.Y. Chang and J.C. Curlander, "Attitude Steering for Space Shuttle Based Synthetic Aperture Radar,

Transactions on Geoscience and Remote Sensing, Vol. G E

22, No. 6 , November 1984, pp. 592-597.

[Jordan, 911 R. Jordan, B. Huneycutt, M. Werner, "SIRC/X-SAR Synthetic Aperture Radar Systems,

" Proceeding

" Proceedings of

1992 International Geoscience and Remote

of the IEEE, Vol 79, No. 6, pp. 827-838, June 1991.

[Klein, 92) J.D. Klein, "Calibration of Complex Polarimetric SAR Imagery Using Backscatter Correlations,
"

Sensing Symposium, Houston, May 1992. [Curlander, Sla] J.C. Curlander and R.N. McDonougb, Synthetic Aperture Radar: Systems and Signal Processing, John Wiley and Sons, 1991. [Curlander, 91b] J.C. Curlander and C.Y. Chang, "Tecbniques in Processing Multi-Frequency Multi-Pol. Spaceborne SAR Data,
"

IEEE

Transactions on Aerospace and Electronic System, Vol. 28,

No. 1, pp. 183-194, January 1992.

[Li, 851 F. Li, D. Held, J.C. Curlander, and-C. Wu, "Doppler Parameter Estimation for Spaceborne Synthetic Aperture Radars,

'' IEEE

Trans. on Geoscience and Remote Sensing,

European Transactions on Telecammunica-

Vol. G E 2 3 , No. 1, January 1985, pp. 47-56. [Runge, 90 [ H. Runge and R. Bamler, "X-SAR Precision Processing, May 1990. [Sack, 851 M. Sack, M.R. Ita, and I.G. Cumming, "Application of Efficient Linear FM Matched Filtering Algorithms
"

tions, Vol. 2, No. 6 , pp. 605-617, November 1991. [Dubois, 891 P.C. Dubois, et al, "Data Volume Reduction for Imaging Radar Polarimetry, 1357. [Freeman, 891 A. Freeman and J.C. Curlander, "Radiometric Correction and Calibration of SAR Images, " Phatogrammetric Engineering and Remote Sensing, Vol. 55, No. 9, September 1989, pp. 1293-1301. [Jin, 841 M. Jin and C. Wu, "A SAR Correlation Algorithm which Accommodates Large Range Migration,

Proceedings of the International Geoscience

''

IEEE International Sympo-

and Remote Sensing Symposium, College Park, Maryland,

sium on Antennas and Propagation, 1989, Vol 111, pp.1354-

to Synthetic Aperture Radar Processing, '' IEE Proceedings,

Vol. 132, P t . F, No. 1, February 1985, pp. 45-57.

[Wu, 821 C. Wu, K.Y. Liu, and M. Jin, "Modeling and a

Correlation Algorithm for Spaceborne SAR Signals,

"

IEEE

Trans. on Aerospace and Electronic Systems, Vol. AES-18,

''

IEEE

No. 5, September 1982, pp. 563-575.

9-1

INVERSE SYNTHETIC APERTURE

J.P. Hardange

RADAR

Thomson-CSF
178, Ed Gabriel Peri 92242 NALAKOFF Cedex FRANCE

1. OBJECTIVES AND APPLICATIONS

ISAR is a technique, based on time and Doppler frequency analysis, which is used for imaging of targets having rotational motions with regard to the radar. In the sixties, observation of the Moon and planets by a radar located on the Earth was one of the first applications (Ref. 1, 2 chapter 3 3 ) . closely derived from these first trials, imaging of objects in terrestrian orbit by ISAR techniques is performed with interesting results (Ref.
3).

However, the most current domains of application of ISAR are now measurements of targets signatures and target recognition. The objective of the first one is to measure the complex reflectivity of each reflecting point of a target. Although the processing has to compensate for various degrading effects of the image quality, it is in general the most simple case, as far as the conditions of the experiment can be perfectly mastered. This happens if the target can be put on a turntable for the analysis (Ref. 5 ) . The result of the analysis is a two dimensional (2-D) image of the distribution of the reflecting points. The two dimensions are range and cross-range in a fixed plan. The second application has been subject to numerous studies and experiments in the field of aircraft classification from a ground based radar (Ref. 4 , 7, 8, 13, 1 4 ) . 1-D (cross-range only) and 2-D principles have been tested. ISAR has been envisaged to equip air defence radars and eventually airborne fire control radars with a non-cooperative target recognition mode. In the same domain, the most demanding configuration is obtained when the radar and the target are moving simultaneously, and when their relative position and velocitv can be described onlv with the help of.an analysis of the returned radar signal itself. This IS the case of ship imaging with an airborne radar (Ref. 11, 16). In this case, the particular behaviour of ships at sea allows the production of 3-D images: range, azimuth and height. 2. HOW TO GET THE IMAGE ? Let us consider a conventional radar, in which the receiver is matched to a single pulse return, that is to say to a very short observation time. In this case, it is

well known that the resolution in range is inversely proportional to the bandwidth of the received waveform. The coherent processing is to short in time to allow any resolution in Doppler frequency (Ref. 2 p. 3-18). If several targets are located in the beam in the same range cell, they cannot be resolved. Furthermore, there is no information at all about their range rate. acceleration, rotational motions, etc... The conventional radar is basically a mono-dimensional range-only sensor. Specific means have to be added to give to the radar the ability to provide the position in azimuth and height of the target, and all the parameters listed above. To try to get all these informations, the principle is to realize a filter which is matched to a long time observation of the target. In these conditions, the Doppler frequency can be analysed, providing an additional dimension to the image. one question is here to recognize the axis which corresponds to this Doppler analysis: IS it the azimuth? Is it the height? Is it something other? This requires some knowing about the conditions of observation. If the axis of rotation of the target is known, like in the case of an object put on a turntable, it is shown in paragraph 4 that the dimension which is measured by the Doppler analysis is a cross-range axis, in a plane orthogonal to the axis of rotation of the target. Generalizing this remark, one can realize a reception filter which is matched to a longer and longer analysis. several terms of the conventional ambiguity fonction, which are negligible for a short O K medium analysis time, become preponderant. The output of the filter is not only a function of time and Doppler frequency, but also of derivates of the Doppler frequency: Doppler rate, Doppler acceleration, etc... ~t is a generalized ambiguity function (Ref. 2 pp. 3- 14 and 3-15). The output of this long time matched filter is a function depending of 2, 3 o r more parameters. Through an a priori knowing of the behaviour of the target, it can be possible, in some particular cases, to establish a relation between each parameter of the ambiguity function and an axis of analysis of the target: Range, azimuth, height. As an example, a relation between the parameters of the generalized ambiguity

9-2

function and the two cross-range axis of a target is established in paragraph 6 in the case of ship classification.
2.1. Doppler Frequency

Equation ( 6 1 becomes: aP,(t)= 2 n f, t

- ----x

4 n R

+ eo

(8)

The expression of the Doppler frequency i.s well-known, but is established again in this paragraph, as an introduction to further developments. These developments leed to the estimation of the resolution that can be obtained for the different parameters of the generalized amhiguity function. The computation of the resolution . h e in range, using a Taylor development of t correlation function of the received signal, is not performed. But the computation of the resolution in Doppler is done, and then the computation of the resolution in Doppler rate, by the Same method. Let us consider a radar, transmitting a constant frequency continuous wave. The phase of the transmitted.signa1 is linked to the time and the carrier frequency by the expression: $*(t)
=

R is in fact a function of time. R(t) can

be developped, using its successive derivatives (up to the third order, in the above example), around its value at an initial reference time (t=O): R(t)
= R,

uo t -

yo

-2

t2
-

.
yo

-6

t3
(9)

uo

--

initial velocity of the target, relative to the dt It=O radar (positive when closing) (10)

dR

yo =

--- I
d3R 1

initial acceleration of the target, relative to the d t ' It=O radar (11)

d2R 1

YO

--- 1
d t '

derivative of the acceleration of the target It=O relative to the radar (12)

2 n f . t +

@,

(1)

Substituting ( 9 ) in ( e ) , we obtain: @,(t)

=

e o : Initial phase The signal is reflected by a target, located at a range R. The signal which is transmitted at instant t, is received at:
t,

2 n f, t -

------ + m,

4 n Ro

t +

--c

2 R (2)

+ --x
4 n
---

4 n

ug t ti
Yo

Approximation: we consider that the velocity of the target is far lower that the velocity of light. As a consequence, we admit that the range of the target doesn't vary between the time of transmission of the signal and the time of reception. At this time, the phase of the received signal is the one which had been transmitted at instant t:

--

---

4 n .

--

t3
(13)

Taking the instant of reception as reference time, equation ( 3 ) hecomes:

The three last terms of equation (13) are characteristic of the motion o f th,? target. We can define a "Doppler phase", wliich is the phase that remains after demodulation of the received signal by the transmitted signal: @,(t)
=

--- uo t + --x x

4 n

4 n
y0

t2
-+

4 n .

t3
(14)

--- Yo -x 6

Using equation ( 1 ) . we get: 2 R @,(t)

=

The Doppler frequency is deduced, being a derivative of the Doppler phase:

2 n f , (t -

---) C

+,

(5)

Introducing the wavelength:

C

We have obtained an expression of the

(7)

-' f

Doppler frequency and phase, depending of the parameters which are describing the motions of the target. It can he useful1 to settle the expression of the Doppler.phase as a function of the Doppler frequency and its derivatives:

9-3

f,(t)

f ,

f , t

.. t2

f , -2

(17)

fD

____
x
2 ____ x
2

2 uo

Average Doppler frequency (18) Doppler Slope (19)

far2 . t ..t3 exp j2n[(f+far- ----It +(f+ fa+)--+ f --ldt 2 2 6 2.4. Ambiguity function lx(~,f,f,f ,fa,fa)12 is the generalized ambiguity function of the waveform.

..

..

.
..

Yo

fD -

..

..

Yo

f , = ----

Doppler acceleration

(20)

3- RESOLUTION

P,(t)

2 n f , t

. t3 n f , t + n f, -3

(21)

3.1. Resolution in Doppler frequency To evaluate the resolution in Doppler frequency, we suppose that the derivatives of the Doppler frequency are null. That is to say, we use the range-Doppler ambiguity function. The ambiguity function is a correlation function. Its maximum is at the origin.
To estimate the width of the correlation peak on the frequency axis, we perform a taylor development around the position of the peak (0.0):

2.2.

Received signal
~ ~ ~

we rnnsider .~. now that the sianal which is

transmitted, s(t), is no more a continuous wave at constant frequency, but a narrow bandwidth waveform on a carrier frequency. To simplify the notations, the amplitude is normalized to 1. Expression of the received signal: s,(t) s,(t)
=

s(t) exp

j *,(t)

(22)

s(t) exp [jZn(f,t

t
2

f , --

.. t3

IX(0,f)l

IX(0,O)IZ

+ f

dIX(0,f)I I

---------- I
df

f, --)I
6
(23)
+

I f=O
(32)

f ___

d21X(0,f)12 I

2.3. natched Filter The filter, which is matched to a signal Of Doppler parameters fa, fa, fa, has the following impulse response:

2!

---___----I df I c=o

..

The first derivative of the ambiguity function at ( 0 . 0 ) is equal to zero, because at this point, the function is reaching its maximum:

df Notations: f=f,-f, f f

I f=O

The second derivative can be computed, as a function of the value of the ambiguity function at (0,O): (26)
(27)

f, - f, f , - fa

..
P

.. .

..
..
=

Output of the matched filter: X(~,f,f,f ,f,,f,)

f, is the half width of the peak on the Doppler axis, at 3 dB: (35) The width of the peak is then:

.. .

(28)

with:

2 f0 -

--------___-_______---_-= -- ( 3 6 )
2n
P

(J.zls(t)12 dt)1/2 tls(t)l*

dt)1/2

T ,

T , is the equivalent duration of the

signal. For the signals of finite duration, which are considered in our application, T, is very near from the duration of the signal.

Example:
For an ohservation during an aperture time T,, the resolution of equation (36) gives:

n '

f '

t '

Is(t)I' dt

IX(0,0,0)12 ( 4 2 )

The width of the peak is then:

2
f

2 f0

0.78
____=____

.
0

(37)

-----_-------___________ (43)
n (/"t41s(t)12
-p

(/+*ls(t)12

d t ) ' / '

n T ,

dt)l/'

T .

Example:
3.2. Resolution in Doppler acceleration

We make now the hypothesis that the Doppler frequency is null, and we use the same principle to evaluate the resolution in f. The Taylor development of the ambiguity function around ( O , O , O ) is done hereafter: jnft' dtl' (38)

For an observation during an aperture time Ta, the resolution of equation ( 4 3 ) gives:
8 E

5.7

4- BASIC ISAR 4.1.

Description

=IX(O,O,O)I'+f

dlX(O,O,~)I'l

----------- I . dt I f=O

The radar is fixed (point A of figure 1). The target is rotating at a consta.nt rate w. The center of rotation is C . The range between the radar and the center c 8 f rotation is R. We consider one reflecting point of the target (point M of figure 1). which is located at coordinates (x,y,O), at initial time (t=O). x and y are small compared to
R.

The coordinates of M at each time are (U.v.0): u f, is the half width of the peak on the Doppler derivative axis, at 3 dB:
v
=
=

x cos w t
x

y sin w t
cos w

(45)
(46)

sin w t + y

The range of the reflecting point is:

R,'
= (R = (R

(41)
R,'

u)' + v 2

(47)

+ x cos w t

y sin w t)'
(48)

(x sin w t + y c o s w t)'

't

Figure 1: Easic ISAR geometry

9-5

Rx2

R , '
=

2 R x cos wt 2 R y sin wt
=

is a function of only one parameter (49)

4.3.

m,.

whith R,

R , at t

Projection plan

R x cos w t

R y sin w t

It comes from previous developments that the resolution of ISAR processing is obtained on a cross-range axis, which is orthogonal to the axis of rotation of the target.
(50)

-----_----- )
Eo

The Doppler frequency of M is then: f,

The image is a projection in a plan containing the radar-to-target axis (range axis) and the cross-range axis which is orthogonal to the axis of rotation. Figure 2 gives several examples of basic projections.
- In figure Za, the target is oriented to

w (y cos u t

+ x sin

t)

(51)

For a short observation time T, around t=O, the Doppler frequency is depending only on the y coordinate:
f
0 -

- ----__ 1 ..

2 Y O

(52)

the radar and has a pitch motion. The range resolution is on the lenght axis of the target, and the cross-range resolution is the height of the target. The projection plan is a range-height vertical plan.

We have established a direct relation between the Doppler frequency and the position of the reflecting point on a cross-ranqe axis, relative to the center of rotati on. we can note that the reception filter has to be matched to signals of constant frequency and finite duration. A S a consequence, the ISAR processing is reduced to a simple Fourier transform.
4.2.

In figure 2b. the target is oriented perpendicular t o the radar range axis and has a roll motion. The range resolution is on the width axis of the target, and the cross-range resolution is still the height. The projection plan is a width-height vertical plan.

Resolution

AS a result of equation ( 3 6 ) . the resolution in Doppler frequency, for an observation time T,, is:

In figure 2c, the target is oriented to the radar and has a yaw motion. The range resolution is on the length axis of the target, and the cross-range resolution is the width. This case is comparable to SAR. The projection plane is a length-width horizontal plan.

where the value of k is near to 1. This result is well-known for every conventional spectral analysis. It could be possible to achieve a better resolution, by the mean of estimation methods. Unfortunatly, these methods require a high signal-to-noise ratio and a large amount of computation. They are not currently used in this field of application. Th@ resolution on the cross-range axis is then:

In several applications, the configuration is not so simple. The target principal axis are anv Dossible orientation ComDared to the rakg'e axis. There can be SimGltaneouly pitch, roll and yaw motions. The orientation of the instantaneous axis of rotation is not known. Consequently, the position of the projection plan is not known either.

5- RELATIONS WITR SAR, REAL APERTURE, AND TOHOGRAPBY

5.1. Relation with

SAR

Figure 3 shows a typical SAR configuration. The radar is flying at constant speed, u, and height on a straight line. It is observing a first reflecting point designed The by M,. The azimuth of the target is Doppler frequency is:

+,.

with
+a

= w

T,

(56)

0 . is the angle of rotation of the target during the observation time (figure l).It is interesting to note that the resolution

There is a second reflecting point, W 2 , which is at the same range, but with a slightly different azimuth:
02
=

mi

+ 89

(58)

9-6

Doppler Height

Rp

Range= Lenght Doppler =Height

Range = Width Dop pIer =Width

(Range = Lenght

a
Figure 2: P , , j e c t i o n plan of the image

The Doppler frequency of the second reflecting point is:

f,,
=

--x

2 u
c o s ( 4 , + 643

(59)

At the first order: f , ,

2 u f , ,
= ---

sin

$,

64

(60)

For a short observation time T,, can be considered as a constant. In this case, the signals received from the targets are pure constant frequencies, and the matched processing is done by Fourier transform. The resolution in Doppler frequency is l/Ta. The associated angular resolution in azimuth is derived:

r+

___________-2 u T , sin
+I

(62)

The projection of this angular resolution at the range of the target is:
ry =

_______-----2u T , sin
+I

X R

(63)

During the observation time, the displacement of the radar is the length of the synthetic aperture: L=uT,
(64)

Figure 3 : SAR geometry

9-9

X
Y

Figure 4: Ship imaging by an airborne radar

+ (Q,(t)sin

+ Qt(t)cos
a

a) z

(86.3)
(86.4)

+ uacos
+

+
I

+ u,cos

f,,(t)

(86.5)

The terms in Q, and Q, of equation (86.1) are negligible. f , , is the effect of spurious aircraft motion. .
f

86.5: The SpUKiOUS Doppler frequency due to aircraft motion is supposed to be sufficiently small to have no coupling effect with other terms. It is evaluated by inertial means, to allow a compensation of Doppler frequency before any further processing. Equation (86) could permit to resolve the ambiguity between y and 2, if the aperture time is sufficiently large. But this also can be done, considering equation (87). in which only z is mentioned: 87.1: It is a constant Doppler acceleration term, which is identical for every reflecting points. It is eliminated by the same way than term (86.1). 87.2: This term is proportional to z . It allows the measurement of Z . 87.3: Effect of aircraft spurious motion, which is corrected by the same way than term (86.5). To simplify the problem, we make an additional hypothesis: The ship has only a roll motion, and + r = 0. After correction of non discriminating terms, equations ( 8 6 ) and (87) become:
f
D -

2
- -

D -

(u,sin + - ussin a ) * ___-____-----_-----(87.1)

R

+ (Q,(t)sin
+ f , , ( t )

+ Q,(t)cos

a)

y )

(87.2) (87.3)

Let us give an interpretation of each term of equations (86) and (87) and explain their interest: 86.1: It is a linear term, which does not depend on y and z. It is identical for each reflecting point. It brings no discriminating information. The parameters ua, u a , 0 and OL can be estimated, by the use of tracking or other method (Ref. 16). This term is eliminated by Doppler compensation. 86.2: This term is proportional to y. It establishes the link between the Doppler frequency and the position of the reflecting point in azimuth. It is depending only on the displacement of the aircraft. It is in fact a SAR effect. 86.3: This term is proportional to 2 . It establishes the link between the Doppler frequency and the position of the reflecting point in height. It is depending on the rotation of the ship. It is an ISAR effect. 86.4: This term is proportional to the velocity of the ship relative to the aircraft. It brings no discriminating information. It is eliminated by the same way than (86.1).

--

2
[

x
+

u,sin + - ussin a - -----_______--_-_

R

sin(2nfrt) sin a

(88) (89)

fD

or cosl2nf,t) sin a z

We come here to a fundamental property of ISAR applied to ship imaging (figure 5):

The Doppler frequency is a sinusoidal function. The average value of the function is proportional to the azimuth of the reflecting point. The amplitude of variation is proportional to the height of the reflecting point. The use of a time-frequency transform is Very interesting to visualise the evolution of the Doppler frequency of each scatterer during the observation time. It can be

9-10

fD

azimuth shift

1
t
Roll moflon period

Figure 5 : Evolution and interpretation of the Doppler frequency either a wavelet transform (Ref. 17.) or a Wigner-Ville transform (Ref. 12). Example:

X
The derivative of the Doppler frequency is a sinusoidal function, centered on zero. Its amplitude is proportional to the height of the reflecting point. ua
R

= 3
= =

cm (x-band)

150 m/s 100 km

m
6.2.

= 900
=

Characteristics of the image

mr

3"/S

If we realize the filter described by equation (31), the resolutions in azimuth and height can be estimated using equations ( 3 7 ) and ( 4 4 ) , ( 8 8 ) and ( 8 9 ) . The best result will be obtained around t=O. At this time, the Doppler frequency in equation ( 8 8 ) is due only to y (the term in z is null). Considering u,>>u, (velocity of the aircraft greater than the velocity of the ship) the resolution in azimuth is:
ry

T , = 1.5 S

a = 450

Resolution in azimuth: ry Resolution in height: rs

=

5 . 2 in
1 m

Taking into account all the approximations that have been done to come to this result, these values are to be considered only as orders of magnitude. The resolution in azimuth is not so uood than the resolution in height, becauie it is proportional to the range. At far range, it is useless to try to have an exploitable top-view of a ship.
~~~~ ~~~~~~~~~

rtD

Ai7 ---------_

2 ua sin 0

Considering the height, starting from equation ( 8 9 ) . at first order we can write:
. 2
= - mr

7- SPURIOUS EFFECTS The spurious effects are identical to spurious effect in SAR, but interpretation can be different. The most important are listed hereafter.
7.1. Range-cell migration

f ,

sin a z

(92)

7.1.1. Description
Due to its rotation relative to the radar, the aspect of the target is changing during the aperture time. The range of each reflector is changing, providing the Doppler effect. If the change of range becomes greater than the range

9-7

The variation of attitude angle of the target relative to the radar is: L u T~

(69)

(70) The resolution ry can be written as a function of +,:

ry =

5.3. Relations with tomoqraphy

-__________
2 0 , sin

+,

(66)

It can be shown that some kind of ISAR can be interpreted a s a tomographic reconstruction problem (Ref. 10). Computer-aided tomography (CAT) is a technique for providing a two- or three-dimensional image of an ohject through digital processing of many 1-D projectional views taken from different look angles. It is used is the medical field for imaging with X-ray scanners. The projection-slice theorem, that is used in this technique, can be applied to ISAR processing.

The best resolution is achieved for 9, = 9 0 D (side-looking radar). In this case:

x ---2

+-

(67)

This equation is identical to equation 5 5 . SAR and ISAR are identical. In both cases, the parameter, which fixes the resolution, is the variation of attitude angle of the target, relative to the radar. To change from SAR to ISAR is only a change of reference:
- Rotation angle refered to the radar: SAR - Rotation angle refered to the target:

5.3.1. Tomography principle, projection-slice theorem and hackprojection reconstruction method Let o(x,y) be the function to map. The measurement are made by the mean of several 1-D projections on different directions. The direction of the line-of-sight for one projection is 4 . The value of the projection at each range u is:

ISAR We have established this result, considering a short observation time. We have limited the computations to first order developments:

P+(u)

In SAR technics, it is the hypothesis of the unfocused SAR. The processing is the same (Fourier transform), whatever the range of the target is: it is focused at infinite range. In ISAR technics, the equivalent name "unfocused ISAR" is not usual. But the principle is the same: second and higher order terms in the variation of the range are neglected.

I*= =

o[x(u,v,+),Y(u,v,~)~ dv (71)

with notations of figure 1: x(u,v,+) y(u,v,@) u cos 0 - v sin u sin

(72)

+ v cos 0

(73)

The projection function has a Fourier transform:

P+(U) =

For a longer apeture time, there is a difference between SAR and ISAR:

In a pure ISAR configuration, the range between the radar and the center of rotation of the target is constant.

- In a SAR configuration, this range is varying following a quadratic law.

The Fourier transform of the projection function is one slice of the 2-D Fourier transform of the image o(x,y) (projection-slice theorem). The orientation of the slice is given by the angle

:l

-j2nuu P+(u) e du

(74)

+:

However, the effect of this range variation can be compensated through an appropriate demodulation of the received signal.

P0 ( u ) =

E ( u cos +,U sin 4 )

(75)

E(X,Y) is linked to o(x,y) by the relation:

5.2. Relation with the real aperture For SAR and ISAR, the length of the synthetic aperture is: L=+,R The real aperture providing the same resolution is:
(68)
!,I.,
I . .

o(x,y) can be reconstructed by the use of a 2- D inverse Fourier transform:

GeneraUy, the reconstruction is not made by the mean of an inverse Fourier transform, because the value of Z(X,Y), in rectangular coordinates is not known. It has to be computed, using interpolations. Another method is used, based on filtering and backprojection in polar coordinates. The measurements are obtained in polar coordinate. Let us write equation (77) as a function of these polar coordinates: X
= U

2 f

s(f,4)

= P+(U),

with U

= -

(83)

The back-projection algorithm can be used, following equations ( B O ) and (81):

. Multiplication
. .

by

IuI

Inverse Fourier transform. At this step the signal is compressed on the range axis. Spreading of each point (amplitude and phase) of each profile on a Line which is orthogonal t o the line of sight of the profile, and sum of the resulting signal at each point (x,y).

cos 4 ,

U sin

(78)

*
The integral over U is identified as an inverse Fourier transform of variable ( x cos + + y sin + I :
~1

6- EXAMPLE OF APPLICATION: IMAGING OF SHIPS

6.1. Modelization

where k is the inverse Fourier transform of IUI. The processing is made of the following successive steps:

Imaaina of shius at sea is a comolex application o f ISAR. It is interesting, because it is a case where the discriminating parameters are not cnly the time and the Doppler frequency, but also the derivative of the Doppler frequency. Figure 4 is a geometrical representation. The radar is flying on-board an aircraft, at velocity ua. Its initial position is 0 and its average course is along the X-axis. The antenna is oriented toward the ship, defining an x-axis. The angle between X and x is the azimuth of the target,

. .

a high-pass filter, H(f) = I U I , which is producing a constrast enhaqcement, a sum for all the values of

+.

Other reconstruction methods are existing and are listed in Ref. 12.
5.3.2. ISAR as a tomographic reconstruction problem

The center of rotation of the target is located initialy at point 0'. Its velocity is us and its course makes an angle u relative to the x-axis. TWO cross-range axis are defined: y-axis, in the horizontal plan, orthogonal to x, and z-axis, which is the height. we consider a reflecting point M on the ship. Its position, relative to O ' , is
(0,
~~~~ ~~

we consider now a radar which is transmitting several continuous waves at constant frequencies. The different frequencies are regularly spaced and can be addressed simultaneously o r sequencialy. This type of observation is repeated for several values of the angle of view 4 of the target. The signal received for a given frequency and a given angle of view, after demodulation, is proportional to: s(f,+)
=

Y,

2).

The shin ~~c is SuDDosed to have sinusoidal

~~

pitch and roll'motions. The angular velocity in pitch and roll are:

np
n,

wp sin (2 n f , t +

$p)

(84)
(84)

= wr

sin ( 2 n f, t + 4 , )

The Doppler frequency of the signal reflected by point M, and its derivative. are: f0

Ir:.

j2n 2f/c u P+(u) e du

182)

-. - .

where 2 L is the dimension of the target along the line of sight, and p$(u) is defined by equation (71). s(f,+) is the Fourier transform of the projection p+(u):

9-11

resolution ( O K the range-cell), the reflector does not appear any more as a point scatterer on the image. There is a degradation of the range resolution. Furthermore, the processing is most of the time organized in two successive steps:

If s(t) is supposed to be a signal of constant amplitude and frequency during the aperture time, then equation (97) can be considered as the Fourier transform of:

. .

~ i r s tstep: range processing, performing a matched filtering on a short time. second step: Cross-range processing, performing a further matching on a long time. This processing is usualy made on signals at constant range of the radar, that is to say, independantly for each range cell.

It is usual, in the study of spurious aircraft motion for SAR processing, to distinguish low frequency motion and high frequency mot ion. 7.2.2.

Low frequency motions

Two main situation can happen:

a
7.1.2.

If there is a migration of the echo, greater than the range resolution cell, the processing in cross-range is mismatched. The aperture time is reduced to the time of presence of the echo in one range cell. The resolution in cross-range is degraded. Compensation methods To compensate for this effect, the principle is to perform the cross-range processing at variable ranges. There are two possibilities:

Error on the rotation rate It is evident, from equation (52). and without making any computation, that this error will modify the scale factor on the cross-range axis. If the value of the rotation rate is not known, the scale factor is anyhow unknown to.

. Slow variation of the rotation rate. This will cause a change of the value of the Doppler frequency during the aperture time. We can make the hypothesis of a linear variation. It is equivalent to introduce a quadratic phase shift.
The result is a degradation of the cross-range resolution. w(t) k,
= w0
SW

to shift, in time, the received signal,

. to shift, in time, the impulse response of the cross-range

matched filter. The main problem is that the ranae-cell mioration +~ ~ - , is not the same for every scatterer. It is proportional to the distance between the scattering point and the center of rotation, that is to say, to its Doppler frequency.
~~~ ~ ~~ ~ ~~

+ k , t

(99)

(100) T, The corresponding variation of Doppler frequency during the aperture time is: 6f

= --

a
7.2.1.

---__x

SW

(101)

7.2. unknown radar and target motions Nodelization The modelization is made for the basic case of ISAR: Short aperture time, developments limited to first order. The signal received from a point reflector is:

A good approximation is to consider that the linear variation of the frequency, Sf, provides a widening of the peak, at the output of the Fourier transform, according to the expression: rf,
= (

rfDo*+ Sf
TfD0Z

)1/2

(102)
411

rfD

= (

-------X

4 6 ,

Y
)

(103)

where m(t) is the phase generated by unknown spurious motion. The output of the matched filtering is deduced from equation ( 3 1 ) : j4(t) j2nft e dt
(97)

7.2.3.

High frequency motions It is the domain of periodic motion, having several periods during the aperture time: Vibrations.

9-12
A

reflector can have a vibration motion on the target. For example, on a ship, objects located on the mast are vibrating with frequencies depending on the natural frequencies of the mast and the engines.

7.2.4.

Random motions Random motions is a generalization of high frequency motions. :Instead of one vibration frequency, there is a complete spectrum, creating a continuous level of side-lobes. x(t) is the spurious displacement,
X is the variance of x.

The spurious motion on the range axis is: x(t)

=

xo sin (2 n f, t)

(105) The spurious phase is: 4 n +(t)

= ---

The spurious phase is: 4 n 4(t)

= ---

xo sin ( 2 n f , t)

(106)

Xlt)

(111)

Writing (106) in equation (97), it comes: x(O,f)=

The variance of the phase o f the spurious signal is:

4nx, j2nft exp(j---- sin(2nfmt)) e dt -a. x (107)

+ -

X(O,f)- 6(f) + ---- L6Lf-f.)

2nx,

The output of the processing filter is:

6(f+f,)l (108)

The usefull echo is bordered with two side-lobes, located at Doppler frequencies f , and -f, (figure 6 ) . The position of the false echos is:
y = +

,----f .

(109)

2 w

The level of these echos is:

2 n xo

IX(O,f.)l

------

(110)

(with IX(0,O)I peak)

1, level of the

- fm
y=--

fm

Doppler Frequency

h fm

2w

h fm y= 2w

Cross - Range axis

Figure 6 : Side-lobes due to vibrations of the radar the target=

9-13

UP-CONVERTER GENERATION

2
=

(118) R++(W) (119)

T, - T , @'

rzF*d
7.3.2. noise

9 '

DOWN-CONVERTER

Figure 7 : Transmission/Reception bloc-diagram

/- m The average power of signal at the ouput of the processing is:

Effect of local oscillator phase The same calculation than in paragraph 7.2.4. provides the average peak-to-sidelobe ratio:

E(IX(0,f)l')

T,S(f) -T,q'&(f)

+ E[S+(f)l
(121)

The peak-to-sidelobe ratio is:

4 n
1

BO*

Ta

(127)

8- REFERENCES (unclassified publications only)

1. 7.3. Spectral purity of the radar 7.3.1. nodelization The effect of local oscillator (L.O.) instability is identical to the effect of spurious radar and target motions: Brown W.M., Fredricks R.J., "Range-Doppler Imaging with Motion through Resolution Cells", IEEE Trans. Aerospace and Electronics Systems, ' l , Jan. 1969, pp. VOL. AES-5, N 98-102. Skolnik M., "Radar Handbook", MC Graw-Hill Book Company, 1970. Gniss H., Krucker K., Magura K., Perkuhn D., "Problems of Signal Processing in a high Resolution Radar - Synthetic Aperture Imaging of rotating Targets with narrowband and broadband Signals", SEE 1978 International Conference on Radar, Paris, Dec. 1978, pp. 243-250. Chen C.C., Andrews H.C., "Target Motion Induced Radar Imaging", IEEE Trans. Aerospace and Electronics Systems, VOL. AES-16, N ' 1 , Jan. 1980, pp. 1-14. Chen C.C., Andrews H.C., "Multifrequency Imaging of Radar Turntable Data", IEEE Trans. Aerospace and Electronics Systems, VOL. AES-16, ~ ~ Jan. 1 , 1980, pp. 15-22. Walker J.L., "Range Doppler Imaging of Rotating Objects", IEEE Trans. Aerospace and Electronics Systems, VOL. AES-16, NO1, Jan. 1980, pp. 23-53. Dike G., Wallenberg R., "Inverse SAR and its Application to Aircraft Classification". IEEE 1980 International Radar Conference Record, Arlington, April 1980. pp. 161-167.
~~

2.
3.

. .

Low rate frequency drifts: Degradation of the resolution Spurious lines: Side-lobes, isolated

. Phase noise: Average level of side-lobes

The local oscillator frequency is used for up-conversion at transmission and down-conversion at reception (figure 7). This operation is equivalent to a single delay filtering of the local oscillator signal. It is performing a weighting of the phase noise spectrum: f,(f)
= 4

4.

5.

sin2(n f T,) f(f)

(125)

6.

T, = 2 r/C is the tine corresponding to the range of the target. The effect of this weighting is an enhancement of the high frequency noise power, by a factor of 2, and an attenuation of the low frequency noise power, by a factor: p(f)
= 4

7.

n 2 f2 T , '

(126)

9-lA

8.

Prickett M.J., Chen C.C., "Principles of Inverse Synthetic Aperture Radar IISAR) Imaging", IEEE EASCON'OO Record, Sep. 1980, pp. 340-345. Mensa D.L., ~ e a n L., "High Resolution Target Imaging", Artech House, 1981. MunSOn D.C.Jr., O'Brien J.D., Jenkins W.K., "A Tomographic Formulation of Spotlight-Mode Synthetic Aperture Radar", Proceedings of the IEEE, Vol.. 71, n08, August 1983, pp. 917-925. Wehner D.R., '"High Resolution Radar", Artech House, 1987, pp. 273-339. D'Addio E., Farina A., Morabito C., "The applications of multidimensional processing to radar systems", International Conference on Radar 1989, Paris, April 1989, pp. 62-78. Bethke K.H., Rode 8.. "A fast ISAR-imaging Process and its inherent degrading Effect on Image Quality", AGARD CP-459, The Hague, 8-12 May 1989, pp. 31-1 to 31-12. Ender J., "1D-ISAR imaging of manoeuvering Aircraft", AGARD CP-459, The Hague, 8-12 Nay 1989, pp. 33-1 to 33-9. Marini S., Pardini S., Prodi F., "Radar Target Image by ISAR Case Study",AGARD CP-459, The Hague, 8- 12 May 1909, pp. 35-1 to 35-14. Fenou M., "L'Imagerie de Cibles marines i la Frontiare entre le SAR et l'ISAR", AGARD CP-459, The Hague, 8-12 May 1989, pp. 28-1 to 28-10. Daubechies I., "The Wavelet Transform, Time-Frequency Localization and Signal Analysis", IEEE Transactions on Information Theory, V01. 36, nos, Septembre 1990.

9. 10.

11.

12.

13.

14.

15.

16.

17.

10-1

Special SAR Techniques and Applications

R. Keith Raney Canada Centre for Remote Sensing
tion leads to the possibility of terrain height estimation, with a precision on the order of the system resolution. Differential interferometric techniques allow observation of changes in the scene on the order of the radar wavelength. For t i m e separation of the signal pair, the full potential of S A R MTI may be realized. Implementation in the time domain and in the frequency domain is described. Airborne versus spaceborne constraints are compared. The principle of conservation of energy of moving targets is presented, and used to estimate target velocity.
2. INTRODUCTION

1. SUMMARY

S A R systems as considered in these lectures are fully coherent, and are characterized by large time/bandwidth signal structure in both range and azimuth. These properties allow additional and specialized performance to be achieved through innovative system variations.
Using one signal sequence, resolution may be improved in azimuth through increased target Doppler bandwidth. Broad beam and Spotlight S A R are techniques used. For the Spotlight mode, high resolution in azimuth is achieved at the expense of image size, so that the Stretch range bandwidth reduction technique is useful to allow increased resolution in both dimensions. A subset of these methods is the Squint mode, whereby the side-looking antenna is pointed forward (or aft) of the zero-Doppler direction, leading to increased complexity for image processing. Wave domain or polar format techniques are required. The requirements, capabilities, and limitations of single beam moving target indication (MTI) for a SAR are reviewed. Special processing for multi-look S A R s to enhance S A R ocean wave imagery contrast and directional spectral estimation is highlighted. Since the S A R signal is coherent, signal phase comparison is possible between pairs of signals. For spatial separation of the signal pair, interferometric signal combinaCulladiall crow12 copy rig^

The special techniques considered in this lecture depend upon the coherent characteristics of S A R signals, and often of greater importance, on coherence properties encountered in the complex image domain. This section introduces the main concepts.
2.1. Coherent Signal Structure

Signals encountered in S A R analysis have phase as their most important attribute, and thus must be coherent over each sequence of received samples. Coherence in the azimuth dimension may be exploited to achieve remarkably good resolution, and may be used for special techniques as well. The classic treatment of S A R analysis is that of Harger 111, who makes the usual assumption for most of his treatment that the two S A R dimensions, range and azimuth, are uncoupled, and that the antenna

10-2

beam is so narrow that range curvature is not an important consideration. For most advanced techniques, detailed treatment of two-dimensional coupling must be included in the model. One such analysis may be found in [ 2 ] ,which is based on the generic signal received from a point scatterer as observed in a wide beam geometry. In studying the azimuth resolution capability, looking at the Doppler properties, or analyzing methods of preserving phase in the derived (complex) image space, an expression for the range signal, azimuth image domain is essential. Such an expression opens our review of special topics. Consider a large time-bandwidth signal ss(x2) from a scatterer at (x=O, y=R,,) observed in signal space (x azimuth, and y range) having wavenumber k = 2 r / X , and two-dimensional pulse envelope a[x2 - R,,(x)]according to

of the signal is given by

(4)

where C is a complex "constant",

and b'
A

1+

bR,ot 4k3(1- o:/4k2)3/2

(6)

The key result is the generalized Doppler domain expression of Eq. 4, derived in [2]. where

Then the azimuth Fourier transform

Range curvature refers primarily to the range/azimuth coupling found in the range de1a.y term of the envelope of Eq. 1 and the locus of Eq. 5 in the Doppler domain of Eq. 4. This coupling is itself range dependent. (This little subtlety is at the heart of processor design for this class of imaging systems.) Doppler properties are sist by the width and angular position of the envelope, both of which are range dependent.
Whether or not significant range curvature is present, the signal expression may be transformed and processed to derive an image. In its most fundamental form the

10-3

image is complex, having a phase structure that may be exploited for certain applications. Ideally, the complex image is simply a linear transformation of the input signal, and has no phase errors introduced by operations in the processor. (Note that very few SAR processors in use today satisfy this objective.) Selected advantages of the complex image domain for calibration purposes are explored in [3]. Considerations essential to arriving at a complex image with robust phase is considered further in section 2.3 below.
2.2. Comments on Resolution

platform, these are sufficient to establish the Doppler bandwidth of the signal. The class of sensors includes S A R , certain SONAR devices, Spotlight S A R , tomography, inverse S A R as used for imaging Earth satellite and nearby planets, etc. The azimuth resolution is given by the inverse (spatial) bandwidth, which leads to
pa

h
2 (sin p, +sinp,)

(7)

where the angles of observation are defined in Figure 1.

A nice discussion of resolution restraints for a SAR may be found in [l]. For the present purposes, we highlight the central result, presented in a form rather more general than that usually found in the literature. Resolution is generally (but loosely) defined as the width of the "point spread function", the "Green's function", or the "impulse response function", depending on whether one has an optics, a physics, or an electronic systems background. More properly, "resolution" refers to the ability of a system to differentiate two image features corresponding to two closely spaced small objects in the illuminated scene when the brightnesses of the two objects in question a r e comparable (according to Lord Rayleigh [1879]).

Sensor path

F
I

Figure 1. Geometry of viewing angles, general case.

A S A R is distinguished due to its ability to achieve "high resolution" (by which is meant, small response function width, and therefore "good") in the azimuth direction, which is parallel to the sensor platform velocity vector.
General case: It follows from Eq. 4 that azimuth resolution is determined only by the wavelength and the span of angles over which the object is viewed. For a moving

Extremum: The minimum (limiting best or finest) resolution is obtained for an effective field of view of 180". The limiting value is readily seen to be

rim
P I P,
9

- 90"

pa-

A 4

(8)

which might seem unrealistic. However, it should be noted that results approaching this value have been achieved in certain applications such as active seismics.

S A R Application: For a S A R , the subten-

ded angles of view are generally very small, and, furthermore, are oriented nominally at right angles to the velocity vector of the sensor. In this case the small angle sine approximation applies, and one obtains

where the angle expresses the total effective viewing angle. It should be noted for satellite radar systems, where one should account for the sphericity of the viewing geometry [4j, that

in which there is no dependence on range or wavelength! This is a fundamental characteristic of S A R systems, and makes them well suited to spacecraft platforms. NB: The magnitude of expression of Eq. 12 is an approximation only, as it violates the correct diffraction limited result that follows from Eq. 7. and it does not account for the benefits of orbital geometry. It is this subtlety that justifies revisiting S A R resolution in this lecture. Although the best known of the " S A R facts" to be found in the open literature, Eq. 12 remains an unacceptable approximation for many space based S A R considerations.
2.3. Comments on Processing

which expresses the fact that the angle over which the object is viewed in a satellite setting is increased over the azimuth (free space) beamwidth of the antenna pattern due to the rotation of the spacecraft as it progresses along its orbit. (The parameters VS,= and V,,,,,, refer to the velocity of the spacecraft along its 01-bit, and the velocity of the antenna beam footprint over the surface of the Earth, respectively.) Note that in virtually all available literature this factor is not included.
A common approximation, familiar to S A R people, is that

Means of handling the coupling in the context of range curvature correction in a S A R processor are well reviewed by Bamler [5]. The several cases discussed in that reference differ primarily by the method and scope of approximations employed. The consequence of most limitations used to date is to restrict applicability of any given method to narrow beamwidth around zero Doppler offset, or to narrow beamwidth about a given Doppler offset. Implementation of several traditional techniques is covered more extensively in [6]. For phase sensitive applications, particularly as encountered in interferometry (see section 4 below), requirements on phase precision in the derived complex image surpasses the ability of most traditional processing algorithms. In response, Cafforio, Prati, and Rocca [7j introduced seismic wave domain techniques to the field of radar. In the context of seismic inversion problems, typically characterized by rather large angular fields of view, and hence severe range/azimuth signal coupling, Stolt [SI had developed an effective change of variables in the two-dimensional frequency

where D,$ is the (azimuth) aperture size of the antenna. Thus, from Eq. 3, we reach the famous representation
(1%)

10-5

domain. A practical difficulty is that the Stolt change of variables requires an interpolator for its implementation, which leads to increased cost and decreased precision impacting particularly phase fidelity in the complex image. Improvements have been demonstrated (e.g., [9]), although until recently interpolators are still required. The main difficulty arises from the range dependence of the range curvature parameter, a quantity not available in the twodimensional frequency or "wave-number'' domain. For phase precision, it is desirable to be able to cope with large angular field and range depth without the need for an interpolator. There have now been processors demonstrated that satisfy this objective ([lo], [ll], and [12]). In future, the coherent structure inherent to the complex image of S A R and related system data sets should not be limited by processor fidelity. The remainder of this lecture assumes that complex image data is available with sufficient phase accuracy to satisfy processing requirements.
3. ONE SIGNAL SEQUENCE

resolution may be maintained to within about a factor of two over "perfect" sidelooking SAR resolution out to squint angles on the order of 45', and certain systems support squint up to 80" and more. In squint mode geometry, since the range and azimuth coordinate systems are no longer orthogonal, targets appear to move through many range resolution cells during their time of radar illumination. For narrow beam systems, this "range walk" may usually be approximated by a linear phase term. Processing adapted to squint mode radars is covered in [6] usually implemented digitally using interpolation. A more recent wave domain contribution described in [SI has been demonstrated at squint angles in excess of 45".
A second complexity arises in squint mode

Specialized systems have been demonstrated that depend on coherent properties of a single sequence of signals in order to improve resolution, or to derive information about moving targets.
3.1. Squint Mode

geometry. Recall that the radar is a sampled system in the azimuth dimension. According to the Nyquist sampling criterion (see, for example, [l]), the Doppler spectrum of the signal sequence must be sampled with sufficiently high radar pulse repetition frequency (PRF) so that spectral aliases, or ambiguities, are avoided. The mean Doppler frequency is set by the squint angle of the antenna according to the terms of Eq. 7. I t is in general not desirable to increase the PRF in proportion to squint, since that would reduce the unambiguous range interval available to the system. Fortunately, the bandpass version of the Nyquist sampling theorem applies, so that the PRF must be greater than only the Doppler bandwidth of the signal sequence. This leaves open the question of estimating the Doppler centroid which may be many times larger than the PRF, a common challenge in most S A R systems. (Doppler centroid estimation is covered in another lecture in this series.)

In many applications, particularly for tactical systems, it is desirable for the radar to look forward (or aft) of the zero Doppler plane by a significant amount. The attainable azimuth resolution for such a "squint mode" S A R follows from Eq. 7 and Figure 1. With suitable processing, azimuth

3.2. Spotlight Mode

proportionally more Doppler bandwidth must be obtained. Processing for Spotlight S A R s tends to be specifically designed for the task. A socalled "polar format" is often employed [13]. This had its roots in the days of optical processing, and creates a data format analogous to the polar coordinate system in which the Spotlight S A R naturally observes its target. The first unclassified work on the subject was [13], and an interesting interpretation may be found in [14]. Treatment of Spotlight S A R in the general clmtext of S A R processing is included in [6]. Modern systems use onIboard real time processing.
3.3. The Stretch Technique

If one attempts to sharpen azimuth resolution in a conventional S A R , either in a squint or side looking mode, a smaller azimuth antenna aperture must be used. This is the "ground rule" for any strip mapping mode SAR. Unfortunately, the effective gain of the antenna is reduced by the square of its aperture, so a severe and usually unacceptable penalty must be paid to push the antenna aperture to very small dimensions. The fundamental resolution expression of Eq. 7 may be exploited in another way, however. Resolution for a given illuminated region is determined primarily by the set of angles over which the data sequence is gathered. A Spotlight SAR is designed to observe a selected patch of terrain with dynamic angular pointing of the antenna pattern steered so as to maintain illumination of the desired area as the sensor passes by. Azimuth resolution much smaller than the antenna half-aperture may be achieved for one area, with the trade-off being that other adjacent areas are not imaged at all. Antenna size for a Spotlight S A R is governed by the desired size of illuminated area, and by the required signal-to-noise ratio of the received signal sequence. Both of these objectives are range dependant, leading to larger antenna size for larger ranges, all else equal. Phased array antenna technology is highly desirable for such systems, although impressive performance may be obtained with rather modest scan angles. For example, from Eq. 7, one may find that resolutions on the order of only 5 times the wavelength (15 cm at X-band) may be achieved with an illumination angle of about 3" (in the nominally side-looking Spotlight geometry). For an N-look image,

Having achieved very fine azimuth resolution through the Spotlight technique, it is natural to seek comparable resolution in the range direction. This is not easily done by direct means. For example, for 15 cm range resolution (symmetrical with the azimuth example above), the range bandwidth required is on the order of 1 GHz. It is desirable to maintain fine range resolution without paying such a penalty.

In the early 1970's the "stretch" technique was demonstrated [15]. For a linear frequency modulated (linear fm) signal, demodulation of the received signal by a delayed replica of the original resiilts in a difference signal of constant frequency. The frequency is proportional to the relative delay of the received and reference signals. Useful results occur only when the signals overlap substantially.
Stretch is perfect for the Spotlight application. It reduces the bandwidth requirements needed for all portions of the system following demodulation. It does this by a trade-off between range and bandwidth, a

10-7

convenient consequence of the linear fm waveform. It is thus restricted to scenes with relatively small range extent, which fits the Spotlight requirement. Stretch modulation applied to the Spotlight S A R case is described by Walker in [13].
3.4. MTI

An ability for moving target indication (MTI) has been for many years an objective of both strategic and tactical radar systems. It is of interest to explore the ability of a S A R in this application. Early work in this area may be found in 111, and especially [16], both of which are restricted to rates rather small compared-to that of the radar platform. For larger motions, the problem rapidly gets more complicated [17]. Attempting to simultaneously achieve both fine spatial resolution and spectral resolution flies in the face of physical principals (e.g. [IS]). The general formulation of this principle is through the ambiguity function. These fundamental limits apply to the conventional SAR configuration, suggested in Figure 2.

The solid lines in the figure represent the (two-way) power profile of stationary scatterer return as limited by the azimuth pattern of the radar antenna. The zero order Doppler response is shown together with two of the ambiguous spectra each centered on a multiple of the PRF. The width of the clutter spectrum is B, = 2pV/X, where p is the two-way half-power angular width of the antenna pattern, and Vis the radar vehicle speed. The dotted lines represent the envelope returned from a single target having radial velocity v with respect to the radar line of sight, leading to a Doppler frequency shift 2v/X which appears as an additional phase term (normalized to azimuth spatial frequency by V ) in Eq. 4. (This may be derived from from Eq. 1 by using R,, = > R,,+ vt.) We are interested in describing the way a slowly moving target appears in a SAR image, and in the possibility of detection of moving targets. For a S A R , the small (coherent) radial velocity component of a scatterer leads to a shift Ax in the mapped azimuth position of its image. This is a natural consequence of the fact that the azimuth coordinate system is derived from Doppler information, and the processor has no way of knowing that the target itself might have a Doppler component. For the aircraft case [16], this argument leads to Ax = R,v/K Since the range to a scatterer is relatively large, there is often a shift in scatterer position many times the azimuth resolution. However, unless there are tell-tale signs of where the object slzould be located (such as the wake of a ship), it is impossible to identify the return as being associated with a moving target. This is an exampIe of the spatialDoppler ambiguity inherent in the problem.

Figure 2. S A l 7 Doppler space in the presence of a moving target. (Two ambiguities explicitly shown.)

10-8

In order for a signal to be detected as moving, its Doppler shift (see Figure 2) must be greater than the Doppler bandwidth of the return from stationary scatterers, which, for satellite S A R s , may be many km/h. Furthermore, in order to detect a small moving target against a larger clutter background, and to guard against false alarms, usually the Doppler shift must be much lai-ger than the system Doppler bandwidth.

In order to achieve such a result, the Doppler MTI passband, defined as the space between ambiguous Doppler spectra free from clutter energy, must be made large, requiring a larger than normal PRF. Normal motions from vehicles and other moving targets of potential interest are tisually more complicated than these simple results consider. In general, defocusing arises when all velocity components are included and the possibility of range acceleration is allowed [16]. More complex motions and vehicle vibrations may lead to loss of signal coherence. These considerations argue against effective adaptation of a conventional SAR as an MTI device. (More niay he done, however, as noted below.)
3.5. Imaging Ocean Waves The most subtle of "conventional" SAR imaging applications is that of ocean waves. A SAR observes essentially only the surface layer of salt water which tinder typical circumstances is constantly in motion. Ships moving on the surface may reasonably be expected to follow the "rules" outlined in the preceeding section, but what about the water itself? The question has value in that SARs are promoted as potential sources of global wave climatology data. Today this promise has yet to be f~ilfilled.

A qualitative prediction of S A R wave imaging appeared in 1978 [19]. Most quantitative analysis was concerned with the effect on a S A R image by the advection of each scattering cell on the ocean's surfa'ce by the passage of longer waves such as swell [20], [21]. Since motion coherently sensed by the SAR causes image shift, systematic motions from the waves' orbital velocities leads to "velocity bunching", actually helpful (within limits) for forming a wave-like contrast pattern in a S A R ocean image. However, the detailed structure of a wave changes with time, sometimes rapidly, so that there are coherence time limitations on image formation that lead to inherent azimuth reolu tion constraints, the so-called "azimuth cut-off'. The state of the art of ocean S A R imaging at that time is summarized in [22]. Independent investigations, both theoretical [23] and experimental [24], raked the importance of non-coherent aspects of S A R wave imaging. Since waves move, and since they are imaged from a platform itself in motion, both phase sensitive and position sensitive motions should be of importance.

For the time scales encountered in airborne

SAR systems, the two types of wave motion may be exploited. The separat,: looks normally created from a S A R data set are spectrally separable in the Doppler (domain. Since the azimuth signal is of large timebandwidth product, the Doppler spectrum, for any reference time, is proportional to the time of actual data collection with respect to the reference time. It follows that looks may be separately processed, relative image shift between looks compensated, and then the shifted looks combined [23]. Properly done, this leads to an "optimum" SAR image of ocean waves, and, of more importance, to an optimized directional spectrum derived from the SAR data [25]. Furthermore, the technique naturally

10-9

leads to removal of the 180" ambiguity of wave propagation direction so typical of wave imaging systems. Imaging ocean waves from satellite altitudes and velocities is in general less than optimum. The time variable aspect of the geometry is scaled to first order by h / V which assume less favourable values for most orbits than for most aircraft S A R s . Ocean wave imaging is an active area of research.
4. SEQUENCE PAIRS

d,

Figure 3. Basic geometry for topographic interferometric imaging radars.

One could visualize a S A R as simply one example of a classical coherent imaging system. It is well known that such systems reach their pinnacle of precise performance when used in an interferometric mode, ie., one in which the scene observed is made to yield its secrets at scales of the wavelength of illumination. This is the objective of S A R configurations designed to use a pair of signal sequences, either spatially or temporally.
4.1. Spatial Interferometry A S A R image of terrain is a two-dimensional mapping of a three-dimensional surface. Unless more information is available, there is no way to quantitatively estimate terrain height. Of course, as with aerial photographs, a pair of S A R iamges each taken from a different point of view could be used as a stereo pair, but as a coherent system, S A R has much more to offer than that.

geometry are shown in Figure 3. The essential feature is that, for known antenna separation d,$, and known slant ranges R, and R, to a scattering centre, then its relative height may be estimated from the phase information in the interference pattern. In the mid-l980's, a brilliant innovation was introduced by Goldstein of JPL [27]. He suggested that an interferometric pair could be created by two relatively closely spaced but separate passes of a (single antenna) S A R satellite. Two pass interferometry was proven using selected passes of existing Seasat data, from which interference fringes were generated, and relative terrain height estimated. Large area image interferometry placed a great strain on the S A R processing capabilities then available. In response, the "wave-domain" algorithm and its derivatives were introduced: collaboration between Rocca and Goldstein led to better phase performance of S A R processors, and proliferation of SAR interferometric work to Europe and elsewhere.

In the early 1970's, Goodyear introduced a S A R system [26]that carried two antennas designed to function simultaneously. The two received signals were coherently combined analogous to an optical interferometer. The main elements of the

The underlying assumption of two pass interferometry is that the terrain being observed is essentially unchanged (with respect to phase characteristics) in the time interval between observations. lhe condition may be inverted, since the absence of interference fringes is an indication of (randomly) changed phase in the affected area. The basic SAR interferometric principle leads to terrain height estimation with a precision on the order of the resolution of the SAR [28]. Its accuracy, however, is very sensitive to the value of the baseline d,,. For applications in which a known "level" terrain feature is present, such as a shoreline, then the elevation contours may be calibrated. In principle, the technique may be used with any satellite SAR. The measurement capability of interferometric SARs has been extended through the use of differential techniques [29], [30],[31]. Through this approach, one looks for differences in the fringe pattern between two pairs of observation opportunities. In effect, in each case, one of the signal pair establishes a reference phase for each element in the scene, and the other signal provides an estimate of phase change with respect to the reference. In principle, the method is sensitive to physical changes in position of a reflecting element to less than X/4, which has been demonstrated [29]. It has been proposed as a method for estimating crustal movement either before or after an earthquake [30]. Interferometry is not without its problems. One of the most challenging is "phase unwrapping" [31], [32]. Interferometry rests on phase estimation, which, unless other information is implicitly or explicitly available, is multiple valued. Current techniques in effect rely on an assumption

of slope continuity through which progressive phase shift may be integrated across changes in excess of 27r. Use of low resolution methods such as "shape from shading" is also helpful. Within the last decade, interferometry for S A R is one of the two most significant developments. (Quadrature polarimetry is the other one.) The topic remain., an area of very active research, and may reasonably be expected to offer valuable quantitative results to the user community in the years to come. 4.2. TemDoral Interferometrv " There is an alternative way in which to build an interferometer. Fir a S A R on a moving platform, a pair of antennas could be arranged to lie along the flight vector, thus providing essentially identical views of the illuminated field, but at slightly different times. The basic geometry is illustrated in Figure 4.

b
V

mad

Figure 4. Dual antenna time-sequential interferometry.

10-11

The time delay 6t between the pair of received signals is D/V if the antennas are operated separately as transmitter and receiver. The time delay is D/2V if, as is usually the case, only one antenna is used for rransmission and both are used for reception. Scene elements having radial motion vrad cause a differential phase shift

where, neglecting noise, one may show that

rn,(t)

- mm(t)

1 - exp - j 4 n - 6 t

( vy

11

as the indication of the presence of a moving target. The detected difference image will contain only moving targets (plus differential clutter and noise), each weighted by the squared magnitude of the radially dependent phase term of Eq. 15.
Implementation of this algorithm using digital techniques in general requires interpolation since there must be continuous adaptation to variations in aircraft velocity V. There is a neat way around this objection, however, by taking the (azimuth) Fourier transform, yielding a domain characterized by a pair of signals such as given by Eq. 4. The delay operator becomes simply a phase multiply exp(-jDox/2) when written in spatial frequency, and under the assumptions that only one antenna is used for transmission, and that the PRF is maintained proportional to ground speed. The companion operation on the signal pair of Eq. 14 is to form their sum. Thus

which may be detected by interferometric combination of the signals from the two channels. For area extensive motion, such as ocean surface currents, the detection technique normally employed is correlation [33], from which excellent results have been obtained. The technique has been extended to wave spectral estimation through a variation on the basic theme [34].
4.3. MTI and Velocity Estimation

When most of the imaged field is expected to be unchanged, the more direct processing strategy of "delay and subtract" applies. This is the case for MTI. It has been known for many years that the dual antenna configuration offers a way around the Doppler lower limit imposed by the clutter spectrum [35]. Performance estimates for the SAR case are available in [16]. If ss&) is the signal from the front antenna, and ssz(t) is the signal from the rear antenna, and nim(t) is the signal from a slowly moving target with radial velocity then
RZ, ( t

ss2(t)- S S , ( ~- 6 t )

(14)

in which the effective common mode (complex) signal is doubled, and the phase modulated term is present as the complement to the differential channel. One may show [36] that an effective doubling of the image SNR follows from the first effect.

10-12

The two components of the moving target are shared proportionately between the two channels, with sin2(kv,Jjt)
in

the

difference channel, and cos2(kvrd6t) in the sum channel, where k = 27r/X. These expressions show that there is a "blind L(, at which the target motion is speed" vrod, through one (round-trip) wavelength between observations, and thus not detectable by phase shift. Blind speeds are characteristic of MTI radar systems. Energy partitioning of the moving target signal between the sum and the difference channel may be used for target velocity estimation, analogous to the same technique used for calibration of dual channel polarimetric radars [36]. Extrapolation of this principle to orbital velocities shows that along track antenna separations of several hundred metres would be required for detection of targets with small components of velocity towards the radar.
5. ADDITIONAL COMMENTS

Improvement of image quality, both through summation of extra non-coherent l.ooks, and through coherent integration i.n special circumstances, has been demonstrated using separate data sets, available, for example, from nearly spatially coincident orbits. By taking the two-pass topographic interferometric geometry to the limit, in principle one could generate multiple range offset data sets sufficient to create full holographic (three dimensional) images of the illuminated surface. In fact, this was demonstrated nearly 20 years a.go using aircraft based radars, but has certain physical limitations when presenkd as an optical analog. It could have value in the modern era of digital imaging, however. Both spatial- and time-delay interferometry have been implemented using a S A R . The classic (partially coherent) field analysis tool, however, is the mutual coherence function which utilizes simultaneous application of a spatial and a temporal interferometric baseline. Such a tool in the SAR context should have value, for example, in deeper analysis of S A X ocean wave imaging. Most of the techniques describe'd above, and the underlying capability to form a basic SAR image, all have their "bistatic" counterparts. A bistatic system is one for which the transmitter and the receiver are in different locations, sometimes widely separated. We have restricted th'e discussion to the case for which the radar transmitter and receiver have the same physical location, the so-called mono-static case. The fundamental requirement for bistatic configurations, just as for the conventional case, is that there be "system coherence". In each situation a means of maintaining phase robustness is required for all signals to be utilized.

This lecture has attempted to suggest ways in which the basic capabilities of a S A R system may be stretched to new and sometimes almost unbelievable domains. The discussion has been far from complete. Some closing comments are offered here. One of the richest areas for S A R system advanced technology is in quadrature polarimetry This topic is covered by other lectures in this series. One should note in passing, however, that the underlying coherence of the S A R channels is required to achieve full polarimetric synthyesis either in the signal domain or the image domain.

10-13

Under the constraint that phase of the signal ensemble is available, then there are many varied measurement possibilities open to the clever inventor. One may reasonably expect that in the forthcoming years, more special S A R techniques will be developed, and be adopted as needed for specific applications of this exciting device.

[7] C. Cafforio, C. Prati, and F. Rocca, Full Resolution Focusing of Seasat SAR Images in the Frequency-Wavenumber Domain, International Journal of Remote Sensing, Vol. 12, 1991, pp. 491-510. [8] R. Stolt, Migration by Fourier Transform Techniques, Geophysics, Vol. 43,1978, pp. 49-76. [9] R. K. Raney and P. W. Vachon, A Phase Preserving SAR Processor, Proceedings of the International Geoscience and Remote Sensing Symposium, Vancouver, Canada, July, 1989, pp. 2588-2591.

6 . REFERENCES

[l] R. 0. Harger, Synthetic Aperture Radar Systems: Theory and Des&, Academic Press, New York, 1970. [2] R. K. Raney, A New and Fundamental Fourier Transform Pair, Proceedings of the International Geoscienceand Remote Sensing Symposium, 26-29 May 1992, Houston, Tx. [3] R. Touzi, K. Raney, and A. Lopes, On the Use of Complex S A R Data for Calibration, Proceedings of the International Geoscience and Remote Sensing Symposium, 26-29 May 1992, Houston, Texas.
[4] R. K. Raney, Conceptual Design of Satellite SAR, Proceedings of the International Geoscienceand Remote Sensing Symposium, Strasbourg, ESA Publication SP-215, 27-30 Aug, 1984, pp. 801-807.

[lo] R. K. Raney, An Exact Wide Field Digital Imaging Algorithm, International Journal of Remote Sensing, Vol. 13, No. 5, 1992, pp. 991-998.
[Ill I. Cumming, F. Wong, and R. K. Raney, A SAR Processing Algorithm with No Interpolation, Proceedings of the International Geoscienceand Remote Sensing Symposium, Houston, Tx., 26-29 May 1992. [12] H. Runge and R. Bamler, A Novel High Precision S A R Processing Algorithm Based on Chirp Scaling, Proceedings of the International Geoscienceand Remote Sensing Symposium, Houston, Tx., 26-29 May 1992.

[5] R. Bamler, A Systematic Comparison of S A R Focusing Algorithms, Proceedings of the International Geoscience and Remote Sensing Sy~nposium,Espoo, Finland, June 1991, pp. 1005-1009.
[6] D. A. Ausherman, A. Kozma, J. L. Walker, H. M. Jones, and E. C. Poggio, Developments in Radar Imaging, IEEE Trans. on Aerospace and Electronic Systems, Vol. AES-20, NO. 4, July 1984, pp 363-400.

[13] J. L. Walker, Range-Doppler Imaging of Rotating Objects, IEEE Transactions on Aerospace and Electronic Systems, AES-16, No. 1, Jan 1980, pp 23-52. [14] D. C. Munson, J. D. OBrien, and W. K. Jenkins, A Tomographic Formulation of Spotlight-mode Synthetic Aperture Radar, Proceedings of the IEEE, No. 7, Aug 1983, pp 917-925.

IO- I 4

[IS] W. J. Caputi, Jr, Stretch: A TimeTransformation Technique, IEEE Transactions on Aerospace and Electronic Systems, Vol. AES-7, No.2, Mar 1971, pp. 269-278. [ 161 R. K. Raney, Synthetic Aperture Radar and Moving Targets, IEEE Transactioins on Aerospace and Electronic System, Vol. AESI, No. 3, May 1971, pp. 499-505.

[23] R. K. Raney and P. W. Vachon, Synthetic Aperture Radar Imaging of Ocean Waves from an Airborne Platform: Focus and Tracking Issues, Joumal of Gcophysical Research, Vol. 93, No. C10, 1988, pp. 12,475-12,486.

[17] E. J. Kelly and R. P. Wishner, Matched Filter Theory for High-velocity Accelerating Targets, IEEE Transactions on Military Electronics, Vol. MIL-9, Jan 1965, pp. 56-59. [18] J. R. Klauder, The Design of Radars Having Both High Range Resolution and High Velocity Resolution, Bell System Tec/mical Journal, July 1960, pp. 745-808. [I91 R. K. Raney and R. A. Shuchman, SAR Mechanisms for Imaging Ocean Waves, Proceedings 5th Canadian Symposium on Remote Sensing, Victoria, B.C., 1978, pp. 495-505. [20] C. T. Swift and L. R. Wilson, Synthetic Aperture Radar Imaging of Ocean Waves, IEEE Transactions on Antennas and Propagation, Vol. 27, No. 6, 1979, pp. 725-729. [21] W. R. Alpers and C. L. Rufenach, The Effect of Orbital Motions on Synthetic Aperture Radar Imagery of Ocean Waves, IEEE Transactions on Antennas and Propagation, Vol. 27, No. 5, 1979, pp. 685-690. 1221 K. Hasselmann, R. K. Raney, W. J. Plant, W. Alpers, R. A. Shuchman, D. R. Lyzenga, C. L. Rufenach, and M. J. Tucker, Theory of Synthetic Aperture Radar Ocean Imaging: A MARSEN View, Journal of Geopliysicul Research, Vol. 90, No. C3, May 1985, pp. 4659-4656.

[24] P. W. Vachon and J. C. West. Spectral Estimation Techniques for Multibok S A R Images of Ocean Waves, IEEE Transactions on Geoscience and Remote Sensing, to appear, 1992. [25] P. W. Vachon and R. K. Raney, Ocean Waves and Optimal SAR Processing: Dont Adjust the Focus!, Transactions of the IEEE Geoscience and Remote Sensing Society, to appear, 1992. [26] L. C. Graham, Synthetic Interferometers for Topographic Mapping, Proceedings of the IEEE, Vol. 62, No. 6, 1974, pp. 763-768. 1271 H. A. Zebker and R. M. Goldstein, Topographic Mapping from Interferometric Synthetic Aperture Radar Observations, Journal of Geophysical Research, Vol. 91, NO. B5, 1986, pp. 4993-4999. [28] C. Prati, F. Rocca, and A. M. Guarnieri, Effects of Speckle and .Additive Noise on the Altimetric Resolution of Interferometric S A R (ISAR) Surveys, Proceedings of the International Geoscience and Remote Sensing Symposium, Vancouver, Canada, July 1989, pp. 2469-2472. [29] A. L. Gray and P. Farris-Manning, Two-Pass Interferometry with Airborne Synthetic Aperture Radar, IEEE Transactions on Geoscience and Remote Sensing, to appear, 1992.

10-15

[30] A. K. Gabriel, R. M. Goldstein, and H. A. Zebker, Mapping Small Elevation Changes over Large Areas: Differential Radar Interferometry, Journal of Geophysical Research, Vol. 94, No. B7, July 1989, pp. 9183-9191.
[31] F. K. Li and R. M. Goldstein, Studies of Multibaseline Spaceborne Interferometric Synthetic Aperture Radars, IEEE Transactions on Geoscience and Remote Sensing, Vol. 28, No. 1, Jan 1990, pp. 88-97.

[32] C. Prati, M. Giani, and N. Leuratti, S A R Interferometry, A 2D Phase Unwrapping Technique based on Phase and Absolute Value Information, Proceedings of the International Geoscience and Remote Sensing Symposium, Washington, D.C., May 1990, pp. 2043-2046. [33] R. M. Goldstein and H. A. Zebker, Interferometric Radar Measurements of Ocean Surface Currents, Nature, Vol. 328, 1987, pp. 707-709. [34] D. R. Lyzenga and J. R. Bennet, Estimation of Ocean Wave Spectra using Two-Antenna SAR Systems, IEEE Transactions on Geoscience and Remote Sensing, Val. 29, No. 3, May 1991, pp. 463-465. @[35] H. Urkowitz, The Effect of Antenna Pattern on the Performance of Dualantenna Radar Airborne Moving Target Indicators, IEEE Transactions on Aerospace and Naval Electronics, Vol. ANE-11, December 1964, pp. 218-223. [36] R. K. Raney, A "Free" 3-dB in CrossPolarized S A R Data, ZEEE Transactions on Geoscience and Remote Sensing, Vol. 26, No. 5, Sept 1988, pp700-702.

Review of Spaceborne and Airborne SAR Systems R. Keith Raney Canada Centre for Remote Sensing
1. SUMMARY

This Lecture provides a concise summary of the state of the art in SAR systems, both spaceborne and airborne. The first civilian SAR mission in space was the United States' SEASAT (L-band), operating July-September 1978. It was followed by two Shuttle missions of one week duration each, SIR-A (L-band, November 1981) and SIR-B (L-band, October 1984). The 1990's is witnessing a flurry of orbital SAR activity, with Almaz (USSR, S-band, March 1991), ERS-1 (ESA, C-band, July 1991), J-ERS-1 (Japan, L-band, February 1992), SIR-C/XSAR (USA/Germany and Italy, L-, C-, and X-bands, three launches planned after 1993), ERS-2 (ESA, repeat mission), and RADARSAT (Canada, C-band, 1995) taking place in the decade. System parameters are included in Tables, and general characteristics of these radars are compared and contrasted. The lecture also provides an overview of airborne SAR systems, including those of NASA (USA), CCRS (Canada), Intera/MDA (Canada), DLR (Germany), and TUD (Denmark), among others. Acronyms are defined in appropriate sections of the text.
2. INTRODUCTION

idea. This lecture describes the current state of system development. All civilian Earth observing synthetic aperture radar ( S A R ) satellites are reviewed: past, present, and future. The current status of representative airborne SAR systems is outlined, and several other airborne systems mentioned. The discussion includes more than twenty S A R s , hardly an exclusive field as of this writing! Taken together, the implied financial support is on the order of 5 billion dollars. There are several published sources that are more complete than this brief lecture is able to be. The reader is referred to a special issue of the Proceedings of the IEEE [l] for articles that describe currently operating and approved future satellite S A R systems in some detail. Those papers include planetary missions as well as the Earth orbital radars of interest here. In the context of preparation for the EOS S A R mission of NASA, an excellent review article [2] has been prepared that covers several airborne SAR systems as well as orbital ones.

After decades of development since the concept of "synthetic aperture radar" was published in 1951, the 1990's is witnessing an unprecedented level of technical (and financial) activity in imaging radar techniques based on Carl Wiley's original
a Canadian Crown Copyiighf

For satellite S A R s , this lecture is organized according to epoch: previous missions; present missions; planned missions; and proposed missions. The airborne systems, typically in a constant state of change, are described in terms of present known capabilities.
As is the case for any area in which many sources might be found, there is not always agreement in those sources on the detailed specifications for any given

system. This is understandjble, particularly when technical change as well as language and terminology confound the issue. In this review, primary sources are used whenever available. Furthermore, when there is an important conflict in published data, comments are included in the text to help clarify the matter for the interested reader.
3. PREVIOUS ORBITAL SAR SYSTEM!;

to a massive short circuit in the slip ring assembly of the solar panel prima:ry power system. (There are system evaluation reports available from the committees charged with investigating this event, should any reader be interested in hardware and programmatic details.) Data provided by Seasat proved to be of high quality and immense interest to the science and applications communities, and still appears as prime material for recent papers in the professional literature.
In the Table, the antenna is described as "corporate". This is a shorthand notation for a passive antenna using power dividing techniques to distribute the signal to be transmitted to a network of radiating elements. In the case of Seasat, the antenna was a flat microstrip array, built on eight panels. The size of the ;antenna for Seasat is of interest. Its long length, and large aspect ratio, is "typical" of "standard" (satellite) S A R antennas. The reasons for this may be found in [ 3 ] , and in other lectures in this series. Typical parameter values important for image quality for satellite S A R systems, in particular resolution and number of looks, were firmly established by Seasat. Unless one is willing to give up other image aspects such as swath width, there is rather little flexibility available. The NASA objective for Seasat was digital processing, but NASA ran out of money (after the potential processor contractor had been selected, but before the contract had been awarded!). The proje.ct was launched, literally, with optical processing as the baseline. In parallel, Charles Wu at JPL (and MDA, Canada) developed digital processing techniques for Seasat. Seasat was a resounding success largely because of digital processing, :as the

For some individuals, history hinges on the wisdom of Lao Tzu (1st century, China), and for others on the inspiration of Jesus Christ (1st century, Judea). For those interested in synthetic aperture radar, however, history hinges on Seasat (20th century, North America). Virtually all Earth observing SAR satellite systems in this lecture owe both their inspiration and baseline performance standards to Seasat, and by implication its design and science teams. Of course, there are solid logical reasons for the technical similarities, but the fact remains, Seasat led the way. Seasat was initiated by an inter-agency study team, and its implementation was managed by the Jet Propulsion Laboratory (JPL) (of the California Institute of Technology in Pasadena, California) funded by NASA. Seasat was a magnificent achievement [l].
Principle parameters of Seasat are sketched in Table 1, abstracted from [3], also available from a widely distributed literature. As in the remainder of this lecture, most comments in the discusssion on this SAR are keyed to entries in pertinent columns of tables. The design lifetime of Seasat was two years. Unfortunately, the spacecraft failed after three months of S A R operation due

TABLE 1. Previous Orbital S A R Systems

Seasat General Country Agency Spacecraft Launch date Lifetime Radar Band [wavelength (cm)] Frequency (GHz) Antenna Size (m), length x height Polarization Incidence angle (") Range resolution (m) Azimuth resolution (m) Looks Swath width (km) Recorder on board? Processing (Optical, Digital)
Noise equivalent uo (dB) Nominasl altitude (km)

SIR-A

SIR-B

(Kosmos 1870)

USA

USA NASA
Shuttle

USA NASA
Shuttle

USSR
Glavkosmos Salyut 25 Jul 87
2 years

NASA
Seasat

26 Jun 78 3 mos
L [23.5]

12 Nov 81 2.5 days

L [23.5]

5 Oct 84
8 days
L [23.5]

s POI
3.0
waveguide

1.275
corporate

1.278
corporate
9.4x2.16

1.282
corporate

10.7x2.16 HH 23
25

10.7x2.16 HH 15 - 64 25 58 - 17
4

1.5~15 (2) HH 30 - 60 (-30) (-30)

(4) 20 - 45

HH

50
40 40 6

25
4

100

50
Y

10 - 60

N
O,D -24

Y
O,D

Y
0

0
-32

(-28) 350, 225 57 250-280 71.5

Mission

800 108
N

260 38
N (none)
nil -

Inclination (") Sun synchronous? Down-link data rate (MB/sec) Repeat cycle (days) Operation time per orbit (min)

N
30
nil

N
analog variable 3 min/tape

110 17, 3
10

remainder of the Tables will support Seasat used an analog downlink with a bandwidth of 20 IVlHz. The data wdS recorded, thence a/d converted (5 bits, I and Q, fixed point) to an effective data rate of 110 MB/sec. The Shuttle SAR missions extended the fouiitlations established by Seasat in the dimension of incidence angle. Both SIRA and SIR-B [2] provided valuable results, but with technical limitations. SIR-A used only optical processing, depending on airborne film recorders adapted to the Shuttle mission. (SIR-A was an achievement in its own right, however, as it was the first payload to fly aboard a space Shuttle platform.) For SIR-B, two problems interfered with the SAR data quality, including a partial failure of the signal line connector to the antenna (with severe increase in noise level), and failure of the data downlink system (TDRS antenna tracking system) so that the data recovery originally planned was curtailed. Never-the-less, much valuable data was collected within the constraints of the flight mission. For two pears at the close of the decade, the USSR operated the Kosmos 1870 radar [4] [SI, a t the time a classified system. Kosmos 1870 was one of three satellite SARs built in the late 1970's built by NPO iMachinostroenye. (The other two are known at present by the name of Almaz, discussed below.) The Soviets had a series of well known real aperture radars in space (the Kosmos 1500 series), but Kosmos 1870, in addition to Venera 16 and 17 t o Venus, was its first publicized orbital SAR. Kosmos 1870 and its Almaz relatives are the only space radars with two antenna assemblies, directed to opposite sides of the orbital plane.

After the conclusion of the Kosmos 1870 mission, Glavkosmos started to advertize its S A R products on the international market, using agreements with such organizations as Space Commerce Corporation (USA). Kosmos 1870 was a radar limited in several regards by its "analog" foundations. Image products from that radar appeared to be far inferior to expectations based on paper specifications. Optical processing in series with analog data links seem to have significantly reduced the effective bandwidth of the data, leading to fewer looks and more coarse resolution than the radar itself should have provided. (N.B. "Digital" image products available from Kosmos 1870 are digitized optically processed S A R data, not to be confused with digitally processed S A R data.)
4. PRESENT S A R SATELLITES

In contrast to the almost total absence of (civilian) space radar capability in the ten years following Seasat, the present decade All is witnessing substantial activity. known space based S A R s presently in service are listed in Table 2.
4.1. Lacrosse

Although a classified mission, there have been several open literature pubkations (e.g [6] and 171) from which the entries in the table for Lacrosse have been gleaned. The key items include resolution, down to one metre [7], and the antenna size, which may be estimated from an illustration published in (61 and Paris Match (date unknown). These two facts lead to the conclusion that Lacrosse uses the Spotlight SAR technique in its high resolution mode. This requires a steerable antenna pattern, so it follows that the rather large

11-5

TABLE 2. Present S A R Satellites

(Lacrosse)
General

Almaz

ERS-1

J-ERS-1

Country Agency Spacecraft Launch date Lifetime (design)

Radar Band [wavelength (cm)]

USA
USAF

USSR
Glavkosmos

Europe
ESA
ERS-1

Japan MITIINASDA J-ERS-1

11 Feb 92

(Atlantis)

Salyut

13 Dec 88 -

31 Mar 91 2 years

16 Jul91 2-3 years

2 years
L [23.5]

X [3.0]
9.5
array
(- 8x2)

s POI
3.0 waveguide

c [5.7]
5.25
waveguide

Frequency (GHz) Antenna Size (m), length x height Polarization Incidence angle (") Range resolution (m) Azimuth resolution (m)
Looks

1.275
corporate

15x1s (2)

10x1

11.9x2.4

HH
steerable

HH
30 - 60

vv
23 26 28 6 100
N

HH
38 18 18 3 75 Y D -20.5 568 97.7

> 1
> 1

15 - 30 15
>4
20 - 45

variable variable -

Swath width (km) Recorder on board? Processing (Optical, Digital)

Y D

Noise equivalent uo (dB) Nominal altitude (km) Inclination (")

Sun synchronous?

D -24
-780 98.5 Y 105 3, 35, 176 10

Mission

-275
57
N -

300 73
N nil

Y
30 (x2) 44 20

Down-link data rate (MBjsec) Repeat cycle (days) Operation time per orbit (min)

3 minltape

antenna nitist be a ph;ised array. Orbital parameters are based on the capability of the Shtittle Atlantis latinched from Cape Kennedy, for which 57" is the limiting inclination. (Lacrosse is nianoeuverable [7], however, so that ctirrent values of inclination and altitude may differ from those in the Table.) The press reports at least two other Lacrosse systems awaiting operations, one of which may already have been launched. Imagery from Lacrosse is not available. 3.2. Almnz Alninz ([I], [4], [SI), or "diamond i n the roiigli", is very similar to Kosmos 1S70, b u t tip-graded i n several regards. Data is storcd o n l ~ o a r d i n four parallel video recorders whose capacity limit the length of each data take. The data tiownlink is analog for both realtime data and recorded data. The most significant imp rove me n is in the A 1maz signal chain followed from conversion to digital data handling and processing. The standard processor is patterned after a VAX 780, and tiscs the conventional range/Doppler :ilgoritIiin. Irnage qtiality is variahle, but some excel leii t exn niples are wai la ble.

the heart of which is a SAR whose parameters are listed in the table. After several months in the three da:y repeat orbit used for system verification and data validation, the orbit was changed. The first two weeks of April 1992 used the "roll tilt" mode in a 35 day repeat patte:rn. The "roll tilt" accomplishes a change in incidence angle to about 35". After midApril, ERS-1 resumed normal operations, still in the 35 day repeat orbit. 'The 176 day repeat orbit will be used for an extended period later in the mission to support altimetric experiments. Data is downlinked when within range of a receiving station on X-band at 105 MB/sec using 5 bit quantization (fixed point) on the I and Q channels. Image quality from ERS-1 is excellent. With two operatibnal S A R s in orbit, it is interesting to compare their respective views of the Earth. Figure 1 shows an area near Whitecourt, Alberta as seen by both Almaz and ERS-1. The contrast between these two images is striking. EKS-1 appears to be more sensitive to topographic effects, and less sensitive to differences in vegetation. A h " , on the other hand, easily displays the clear-cuts in the forest, but is less sensitive to the terrain relief.

From a n applications point of view, Almaz is a very interesting system. Its S-band wavelength arid selectabie incidence angle make it a good choice for certain Earth resource issues. An example of Alinaz imagery may he found i n Figure 1.
3.3. ERS-I

4 . 4 . J-ERS-1
The Japanese launched J-ERS-1 ("Earth Resources Satellite") early in 1992. After early difficulty with the mechanism, the SAR antenna was successfully deployed on 9 April. At this writing, the SAR system is still undergoing inflight checkout; the initial imagery is excellent. J-ERS-1 is a joint development of the Ministry of International Trade (MITI) and the Science and Technology Agency of the National Space Development Agency

a
-

After extensive studies ant1 preparations from about 1975, the European Space Agency Iaunched ERS-I i n 1991. The primary payload [SI, [9] o n l ~ o a r dERS-1 ("European Remote Sensing" satellite) is the Active Microwave Instrument (AMI),

Figure 1. Satellite Imagery of the Forest Test Site in Whitecourt, Alberta. Top: ERS-1 image (C-band, 23' incidence angle) 4 ' incidence angle) Bottom: ALMAZ image (S-band, 4 (Processing courtesy Canada Centre for Remote Sensing)

(NASDA) of Japan [lo].

5. PLANNED ORBITAL SAR MISSIONS

J-ERS-1 is like Seasat SAR in many ways. In fact, its original performance specifications were identical to those of Seasat. It turned out that two things combined leading to the present paramaters (Table 2). First, the test rzsults for the solid state power amplifiers showed much hetter output power than had been thought possible in early design stages. Thus, greater range could he accepted. Second, the driving application for J-ERS1 is global economic geology, for which larger incidence angles are desirable in order to reduce image distortions from terrain relief. The nominal incidence angle for J-ERS-1 was changed accordingly, now at 38", rather than the nominal 23" of Seasat. (Note that the literature almost always quotes a "look angle" of 35" for J-ERS-1, which refers to the angle of the antenna beam at the spacecraft, and does not include Earth curvature. Incidence angle is the preferred specification for the antenna effective viewing angle.) As a consequence of the larger incidence angle, the ground range resolution is reduced from 25 metres to 1 8 metres, with no change in system range bandwidth. Having a range resolution of 18 metres, the azimuth resolution was chosen to be the same, achieved by a simple adjustment in the processor. Finally, the number of looks available is reduced by about 25'3'0, leading to 3 looks, or (18/25)x4. Con-. clusion? Although the specifications for J-. ERS-1 look different from Seasat, the imagery with respect to image quality should be about the same. The increased incidence angle should be of value.

In the sense used here, planned orbital missions are those for which the sponsoring agences have commi:tted the necessary resources, and the flight systems are under construction. The "approved" systems are outlined in Table 3. They include: SIR-C/X-SAR, a series of Shuttle-based missions each of relatively short duration; ERS-2, identical in most regards to ERS-1 (see above); and RADARSAT, the first remote sensing satellite mission of Canada.

5.1. SIR-C/X-SAR
The SIR-C/X-SAR mission represents a major milestone in space-based radar. Principal features of interest are suggested by the entries in Table 3 which guide the comments in this section. More details may be found in [2] and [Ill. SIR-C/X-SAR is a joint venture between the United States (NASA-/JPL), and a European consortium of Deutsche Forschungsanstalt fur Luft- und Raumfahrt e.V. (the German Aerospace R.esearch Estab-lishment known as DLR) and the Agenzia Spaziale Italiana (AX, Italy). JPL is responsible for the C- and L-band systems, and DLR/ASI is responsible for the X-band system. Through extensive collaborative efforts, these radars have been harmonized so that they may be operated simultaneously, clearly a "first" at both organizational and technical levels. Since the Shuttle has limited orbital duration, each of the three planned missions are expected to last for about 10 days, although an extended (16 day) Shuttle capability remains a possibsility for the later launches.

11-9

TABLE 3. I'lanned Orbital SAK Systems SIR-C/X-SAR

ERS-2

RADARSAT

General Coli n t ry

(See text) NASA/DLR/DARA Shuttle

Europe

Canada CSAIUSA (dedicated)

Agency Spacecraft

ESA

ERS-2
1994

La ti rich c i a te
Design lifetime
I<adar

'93, '94, '96 10 days

L,C; X

1995

2-3 years
C [5.25]

5 years
C [5.7]

Band [wavelength (cm)] Frequency (GI-Iz)

Ant en na

1.25, 5.3, 9.6

arrays

5.3
WaveGuide

5.3
WG array

WG(X)

Size (ni), length x height Polarization Incidence angle (") Range resolution (m)
Azirnti t h resolution (m)

12x(3, 0.75, 0.4)

10x1

15~1.5

quad-pol L + C ; X"" 15 - 55

vv
23 26 28 6 100
N

HH
< 20 - > 50

10 - 30
30
-4

10 - 100 9
-

100

LOOI<S Swath width (Itm) Recorder on board? Processing (Optical, Digital)

1-8 10 - 500

15 - 60

y ( + D/L) D
-40 < -28 225 57

Y
D

D
-24 -780 98.5

Noise eqtiivalent uo (dB)

Mission Altittide (Itm) Inclination (")

-23

- 800
98.6
Y

Sun synchronous?
D o w n l i n k data rate (MB/sec)

N
45 (TDRS)
nil

Y
105

14 - 105 24 20

Repeat cycle (days) Operation time per orbit (min)

TBD
10

(60 h, total)

11-10

The radar bands selected represent orbital SAR heritage (L), current operational preference (C), and extension to a new (civilian) orbital wavelength (X). The antennas required have been matched in two regards. First, in order to allow the same PRF for all three bands, dictated by the requirement to have simultaneous multi-frequency operation, each of the antennas should have the same length. Second, in order to have nominally the same elevation beamwidth, the height of the antennas must be scaled in proportion to their wavelength. The size of the total array is about 12m by 4.lm. The three antennas will be mounted on one common structure in the cargo bay of the Shuttle, vertically stacked. Antenna technology for the X-band system is slotted waveguide, horizontally polarized. Elevation steering is by physical rotation of the antenna radiating surface about its longitudinal axis. Antennas for both the C- and L-band radars use active phased array technology. 8 panels for each antenna, There are 1 each one consisting of transmit/receive (T/R) modules for E1 and V polarizations. By using phase control of the individual T/R modules, the antenna patterns can be steered in both elevation and azimuth. The H and V phased arrays, backed up by parallel receiver and data recording chains, may be cycled to achieve reception of the complex scattering matrix of the scene, the so-called quadrature polarization technique (described in another lecture in this series). SIR-C/X-SAR will be the first time that this capability is available from space. Clearly, with choice of polarizations and frequencies, nearly 20 modes are available

from this radar system. These, coupled with a degree of freedom in bandwidth and incidence angle selection, lead to very complex planning for the operations of SIR-C/X-SAR. In addition, there are options in the number of bits to be included in the signal data path, and allowance for experimental passes in the ScanSAR format (see RAD'ARSAT below), and for squinted beam data collection. (At this time, however,,there is no plan to support data processing for these special imaging geometries.) Processing for SIR-C, and particularly for X-SAR, poses its own challenges. Normal Shuttle angular motions, in combination with Earth rotation, lead to relativl-ly large and time varying Doppler centroid variations in the received SAR signals. For the Shuttle, pitch, roll, and yaw are allowed to vary within 1.5". (The attitude control system for the Shuttle is not "proportional", but depends on angular momentum impulses from gas-jet thrusters to correct angular position when the design "dead-zone'' is exc8:eded.) Tolerance by the coherent radar signal to (yaw) attitude is determined primarily by azimuth antenna beamwidth. For a given aperture size, the beamwidth of the antenna pattern decreases in proportion to wavelength. For example, the azimuth beamwidth for X-SAR is about 0.15', which is ten times smaller than the Shuttle attitude uncertainty. It follows that Doppler centroid estimation and tracking, and its compensation in time and range, is the most challenging S A R process.ing task among all others represented by the mission. This circumstance has led DLR to new frontiers in SAR data processing techniques, le., [12] and [13].

5.2. RADARSAT

On 13 September 1989,the Government of Canada announced full commitment to build and operate RADARSAT, Canada's first Earth resources remote sensing satellite [14]. It is being prepared for a launch in 1995, and is designed for five years of service in orbit. The only imaging instrument is a SAR. A variety of resolution, image swath width, and incidence angle parameters are available that may be selected through ground command. The designated agency responsible for RADARSAT is the Canadian Space Agency (CSA). The mission is the result of more than a decade of work and initiative by the Canada Centre for Remote Sensing (CCRS). The nominal configuration of the spacecraft has the SAR pointing to the north side of the orbital plane. This makes possible regular coverage of the Arctic up to the pole, but coverage of the central Antarctic region is not possible with this geometry. However, for two periods during the first two years of the mission, the satellite will be rotated 180" about its yaw axis so as to direct the radar antenna beam to the south side of the ybital plane. Each such yaw manoeuvre IS expected to be maintained for two weeks. The purpose of this manoeuvre is to obtain a complete SAR map of Antarctica at the times of maximum and minimum ice cover. (To the extent that spacecraft thermal budgets allow, this will also allow opposite side viewing for belected regions north of the Equator during these periods.) Optical sensors such as Landsat have no access to regions closer to the poles than 81", and are further compromised by clouds and darkness. Other space radars such as

ERS-1 have no ability to look to the south, or, in the case of Almaz, have orbital inclinations such that Antarctic coverage is incomplete. NASA is a major partner in the RADARSAT mission contributing services for the planned 1995 launch from Vandenburg AFB using a medium-class expendable vehicle (McDonnell Douglas Delta I1 7920-10). The satellite payload will consist of the S A R and its associated downlink transmitters, tape recorders, and command and control computer. The spacecraft is being procured from Ball Aerospace (United States), while SPAR Aerospace (Canada) has prime system responsibility under contract with the Canadian Space Agency. Data downlink for RADARSAT is at Xband, with the maximum data rate chosen to be compatible with the ERS-1 data rate. RADARSAT will use 4 bit floating point quantization, I and Q. There are two downlink channels, required to support simultaneous readout of the onboard tape recorder together with "live" data. There are three downlink transmitters, a "3 for 2" redundancy approach.

For a satellite using a radar sensor, good solar illumination of the spacecraft is more important than sunlight on the Earth's surface. For this reason, RADARSAT will use a sun synchronous dawn-dusk orbit. In this orbit, the spacecraft is illuminated by the sun throughout each orbit. (There are short periods at southern latitudes during the Austral winter during which the spacecraft is in eclipse.) Perhaps the greatest operational advantage of this orbit is that the SAR can be fully dependent on solar derived rather than stored battery power,

11-12

which means that there is no limiting distinction between ascending and descending passes from an applications point of view. Thus, nearly twice as many viewing opportunities are available to the mission than would otherwise be available. Another operational advantage is that the ground station data downlink periods for RADARSAT will not conflict with other remote sensing satellites, most of which use near mid-day orbit timing. RADARSAT will carry tape recorders with sufficient capacity for more than ten minutes of full quality SAR data. As a consequence of the orbit yaw manoeuvre and the onboard recorders, RADARSAT will be the first fine resolution satellite system capable of complete global coverage. This feature of RADARSAT is relevant to global environmental monitoring as well as being of scientific value.
5.2.1. Modes

imaging modes at a variety of incidence angles, several additional imaging modes become available at rather small marginal cost. The design philosophy for these extra modes has been to base the system specifications on the standard imaging modes, and to optimize the performance of the additional modes within the design envelope determined by the standard modes. Image quality in the additional modes is predicted to be comparable to that of the standard modes. Imaging modes for RADARSAT include Standard, Wide Swath, Fine Re:solution, Extended, and ScanSAR (see Figure 2 and Table 4). In each mode, data are collected continuously along a swath parallel to the sub-satellite path. Swath length is limited only by the duration of continuous radar operation, and may be thousands of kilometres long. Swath widths and positions are determined by the antenna elevation beam patte:rns (and the radar range gate control), a.nd have been chosen for the standard modes so that there is at least 10% overlap between adjacent swaths. Range resolutilnn when projected onto the Earth's surface varies with incidence angle and hence ground Three range bandwidths are range. available (11.6, 17.3, and 30.0 MHz) to allow choice in ground range re.solution achieved at each incidence angle. Nominal range resolution for the standard beams has been specified at ground ranges of 400 km and 675 km from i.he subsatellite locus. The additional modes are generated by appropriate choices of antenna beam and range pulse bandwidth. The fine resolution mode, for example, is achieved by selecting the widest available bandwidth, and using a narrow heam in elevation at incidence angles larger than

RADARSAT has been designed in response to user requirements that demand a variety of incidence angles (from about 20"to 50") in the standard imaging modes. An antenna with electronic elevation beam steering is part of t'he baseline RADARSAT design. Although this enables user requirements to be met, it does add further complexity to the entire system. In order to provide a (nominally) constant ground range resolution over the range of incidence angles, three different poise bandwidths are needed. It also follows that very fine control of the transmitter pulse repetition frequency (PRF) i s required. I-Iaving moved to build in an antenna and control system with the flexibility (and complexity) necessary to support standard

11-13

1
Mode

TAULE 4. RADARSAT Imaging Modes Resolution (R' x A, m) Looks' Width

(W
4 4 4 1 2-4 4-8 4 4

Incidence (degrees)

Standard Wide (1) Wide (2) Fine resolution ScanSAR (N) ScanSAR (W) Extended (H) Extended (L)

25 x 28 48-30 x 28 32-25 x 28 11-9 x 9 50 x 50 100 x 100 22-19 x 28 63-28 x 28

100 165 150

45
305 510 15 170

20 - 49 20 - 31 31 - 39 31 - 48 20 - 40 20 - 49 50 - 60 10 - 23

' Nominal;ground range resolution varies with range. ' Nominal;range and processor dependent.

" '
I \
250 km

Subsatellite Track Extended Beams (low incidence)

A\

-,-./-wG9wlth

..

-.

Fine Resolution
,

- _... .. .

Extended Beams (high incidence)

D=-al,,s

Beams Beams

Figure 2. RADARSAT Imaging Modes

II-IJ

(nominally) 45'. A narrow swath results from the requirement to minimize beamwidth in order to maintain good signal to noise ratio, and also from the necessity to maintain data rates consistent with downlink channel capacity. Wide hwath modes are supported by wider antenna beam widths than normal at steeper incidence angles, and use of the smallest available range pulse bandwidth leading to more coarse ground range resolution. Signal to noise ratio and data bandwidth arguments apply in these modes that are counterparts to those for the fine resolution modes, but with the result of broadening the usable swath width. Extended modes result from selection of beams outside of the nominal 500 km accessibility region, either closer to nadir (steeper incidence), or further away (more shallow or grazing incidence angle).
5.2.2. ScnnSAR

to image a section of one sub-swath must be from consecutive pulses in order to provide adequate sampling, and must be of sufficient length to allow formation of the synthetic aperture needed for the subswath at the required resolution. The imaging operations are therefore split up into a series of blocks of pulses, each block providing returns from one of the sub-swaths. Each block is processed to provide an image of a section. of the corresponding sub-swath. The imaging operations cycle around the full set of subswaths sufficiently rapidly that the imaged sections in any one sub-swath are adjoining or over-lapping.

RADARSAT will be the first operational satellite radar system to implement the ScanSAR technique.
6. AIRBORNE S A R SYSTEMS

a -

I n order t o allow imaging of a swath much wider than ambiguity limits would normally allow, the RADARSAT system has been designed to incorporate an alternative and less conventional mode ([l5] and [16]) known as ScanSAR. In this mode, for which rapid steering of the elevation beam pattern of the antenna is essential, extended range coverage can be obtained using a set of contiguous beams, enabling images of total swath width up to about 500 km to be produced. This is accomplished at no increase in mean data rate from the radar, b u t at the cost of degraded resolution of the resulting image.

This section provides a glimpse at civilian airborne S A R systems now in service. Since most of these radars are meant for technology development as well as applications experiments, the hardware is frequently changed. The parameters listed in Table 5 are thought to be an accurate representation of the basic perforniance of those systems. Additional comments, and a resume of other airborne SARs are included in the following sub-sections.
6.1. Comparison of Selected Systems

The principle of ScanSAR is to share

radar operational time between two or more separate sub-swaths in such a way as to obtain full image coverage over their combined swath. The set of returns used

The most widely deployed airborne S A R , and the one having the most modes, is that of NASA which carries the radar of JPL (the Jet Propulsion Laboratory). This radar, known as AirSAR ([2] and [19]), is relatively new, having been designed and built to replace its predecessor wl-iich was destroyed in an aircraft fire in 1985. The

11- 15

TABLE 5. Current Status of Selected Airborne SAR Systems

AirSAR General Country Agency Aircraft Nominal altitude (km) Nominal airspeed (m/sec) Purpose Radar Band Frequency (GHz) Antenna length (m) Antenna motion controller ? Polarization diversity Quadrature polarization Incidence angle () Range resolution (m) (Slant) Range cells Azimuth resolution (in) Looks Swath width (km) (Slant) STC ? Processor on board ? Noise equivalent uo (dB) Special modes
2

C/XSAR Canada CCRS

E-SAR

KRAS

STAR-1 Canada Intera Cessna

USA

Germany

Denmark

JPLfNASA

DLR
DO-228 3.5

TUD
Gulfstream

DC-8
8
expl

CV-580

6
130
expl

12.5 300
expl C

10 175
operatl

70
expl

c , L, p

x,c
-1.2 Y Y
C

x, c, L
9.6, 5.3, 1.3 0.15, 0.24, 0.85
N

X
9.6 -0.8
Y

5.3, 1.25, 0.44 9.3, 5.3 1.3, 1.6, 1.8

N

5.3
1.2
Y

HH, VV
No

vv
No
20 - 80 2,4,8 8192

HH
No

c, L, p
20-60 7.5 ( S )

0 - 85
6 - 20 4096
:1 - 1 0

15 - 60 2 (SI 2048 2

45 -

so

6, 12 4096 6 7
40, 60 (S)

2,4,8
2 - 16 9 - 48 (S)
Y

1-1
18 - 63
Y

1-8
3 (S)

7-13

Y Y (QL)
-40, -30, -35

Y (1 ch)
InSAR, At

Y
-30, -40
InSAR

(Y) -42

Y
-30

11-16

quatlrature polarimetric capability at the three frequencies of this radar offers a unique and very rich data source that is made available by NASA to investigators arotiiid the world. AirSAR also offers two other special niotlcs that merit attention, a s noted i n the Table. InSAR is an interferonietric mode created by using data from two antennas, one mounted above the other on the side of the aircraft. The interference pattern between them may be used to deduce terrain height information. For AirSAR, this capability is at C-band only. The other special mode also requires two separate antennas, this time spaced along the line of flight. Interferometric measurements with these two antennas may be tised t o observe phenomena in the scene that change over the short interval & between observations, such a s currents on the oceans surface. The SAR flown by CCRS (the Canada Centre for Remote Sensing) is on ii Convair-580. Both X- and C-bands are fully supported by onboard real time digital processors [20],and have a variety of modes and data combinations available. The standard image products are pro-diicctl at seven looks. The system has recently been modified to incorporate full quatlrature polarimetry on C-band, and may be operated in an InSAR mode. Signal data is recorded so that ground processing may be tised for specific experimental purposes, such as investigations requiring access to separate looks for optimized oceanographic SAR wave imagery. The airborne radar of DLR (defined in section 5.1) continues to be upgraded with new modes and capabilities [21], [22], and [23]. It is designed primarily for high resolution and technology development, hence it has a narrower swath width than

do the previous two systems. Its recent extension to X-band is in support of DLRs vested interest in X-band for the X-SAR radar. The system includes a quick-look (QL) onboard processlx having 50 m x 50 m resolution. The radar of the Technical University of Denmark (TUD) has been designed to offer a variety of incidence and image parameter values within the conr,traint of being a single polarization C-band system [24], [ 2 5 ] . Within the limits set by their respective imaging geometries, the radar is matched to the ERS-1 S A R , aided by the use of an aircraft capable of high speed and high altitude. The leading civilian S A R that is used for mapping surveys is STAR-1, owned and operated by Intera, of Calgary, Alberta [26]. This X-band system has performed more than 75% of all of the airborne radar mapping done for commercial clients world-wide since 1986. Data products from this system are digitally rectified to map accuracy standards, and, with the recent use of GPS, allows accurate mapping with no need for surveyed control points. Under satellite navigation control, the radar map itself is the most accurate source.
6.1. Overview of Other Systems

There are several other airborne SARs that deserve mention. The Netherlands for many years has been supporting the development of an advanced system Currently, known as PHARS [27]. PHARS is a C-band S A R mounted in a Swaeringen Metor 11, a twin engined business plane. Early results meet or exceed specifications. Nominal resolution is about 5 metres with about 6 looks. The antenna is VV polarized to :;upport

11-17

experiments with ERS-1. The program is committed to implementing a full quadrature polarimetric version within a few years. One of the pioneers in the field of S A R is the Environmental Research Institute of Michigan (ERIM), whose civilian experimental SAR is managed by the United States Navy Air Development Center (NADC), and is mounted on a P-3 aircraft. The radar operates at X-, C-, and L-bands, and is fully polarimetric [28]. Nominal resolution is in the 3 metre range. The French have been actively involved in S A R for many years. Their S A R (a civilian version of a former Thomson-CSF classified X-band system) is known as VARAN-S [29], and has supported several remote sensing experiments in Europe. Polarizations are HH and VV. Nominal resolution is on the order of 5 meters, with four looks, nominal. In addition to these relative well known systems, others might be mentioned for completeness. The Chinese Academy of Sciences developed their own X-band S A R several years agb. The system is called CASSAR, and has seen limited deployment in the Far East. The USSR maintained an airborne S A R capability as a test-bed for satellite programs, both Earth oriented and planetary. India has developed its own airborne S A R as well as an indigineons digital S A R data processing capability.
7. PROPOSED SAR SATELLITES

I1 has already been built, and is virtually a twin to Almaz. There is an intention to launch and operate this radar, hut the changing infra-structure of the former USSR space segment places the future of Almaz I1 in doubt. Progammatic responsibility has been assumed by Russia for PRIRODA ("nature"), a complex payload that includes a S A R as one of several instruments [17], [18]. Much of the hardware for the PRIRODA payload has already been built.

The EOS-SAR (Earth Observing System SAR) [2] is at a different stage. There is an active design and science team at work at JPL on EOS-SAR, but the program has yet to receive funding approval from NASA. The Proposal by the EOS-SAR team to NASA is to be in 1993. The EOS-SAR as presently conceived would be a most ambitious system. It would carry the multi-frequency, quadrature polarization capability of SIRC / X - S A R into a long design life satellite implementation. In addition to these SARs, there are other initiatives being developed. France for several years has been promoting S A R 2000, an X-band space radar intended to complement the SPOT program. Recently, France and Canada have entered a bilateral agreement to do joint studies of a combined satellite radar system building on the SAR-2000 and the RADARSAT 1 1 1 planning. RADARSAT 1 1 , essentially a replacement for RADARSAT, is under consideration by the Canadian government. The European Space Agency has sponsored several studies of S A R concepts that look beyond ERS-2. Undoubtedly, other space agencies have exploratory space radar studies under way.

There are three SAR satellites likely to be approved for completion and launch in the next decade, as noted in Table 6. Nmdz

TABLE 6. Proposed SAR Satellites

ALMAZ I1

EOS-SAR

PRIR.ODA

General
COLI n try

Agency Spacccraft
Latinc h cla te

(TBD) TBD
Salyut
(1994?)

USA NASA
(dedicated)

Russia

TElD

MI:R
(1W4?)

2000

Life t i ine
R R dR r

2 years
S WaveGuide

15 years
L, c, x arrays quad-pol

Band
A ii t en na

L, s WG, array
VV
01' HH

Polarization Incidence :ingle (") Range resolution (in)

Azinititli resolution (m)

HI3
30 - 50 15 - 30 1s

15 - 45

35
100

20 - 250
8 - 250
1 - 10
30 - 360

SO, 150
80

Look5
Swatli width ( k m )

>4

20 - 45

Recorder on h o a r d ?

TBD

11-19

8. REFERENCES

[I] F. L. Li and R. K. Raney, Prolog to Special Section on Spaceborne Radars for Earth and Planetary Observations, Proceedings of the IEEE, Vol. 19, No. 6, June 1991, pp. 773-176.

[lo] Y. Nemoto, H. Nishino, M. Ono, H. Mizutamari, K. Nishikawa, and K. Tanaka, Japanese Earth Resources Satellite-1 Synthetic Aperture Radar, Proceedings of the IEEE, Vol. 79, No. 6, June 1991, pp. 800-809. [ l l ] R. L, Jordan, B. L. Huneycutt, and M. Werner, The SIR-C/X-SAR Synthetic Aperture Radar System,Proc. of the IEEE, Vol. 79, No. 6, June 1991, pp. 827-838. [12] H. Runge, Benefits of Antenna Yaw Steering for S A R , Proceedings of the International Geoscience and Remote Sensing Symposium, IGARSS '91, Espoo, Finland, 3-6 June 1991, pp. 257-261. [13] R. Bamler and H. Runge, PRFAmbiguity Resolving by Wavelength Diversity, IEEE Transactions on Geoscience and Remote Sensing, Vol. 29, NO. 6, NOV1991, pp. 997-1003. [14] R. K. Raney, A. P. Luscombe, E. J. Langham, and S. Ahmed, RADAR-SAT, Proceedings of the IEEE, Vol. 79, No. 6, June 1991, pp. 839-849.

[2] J. Way and E. A. Smith, The Evolution of Synthetic Aperture Radar Systems and their Progression to the EOS S A R , IEEE

Transactions on Geoscience and Remote Sensing, Vol. 29, No. 6, November 1991, pp. 962-985.
[3] R. L. Jordan, The Seasat-A Synthetic Aperture Radar System, IEEE Journal of Oceanic Engineering, Vol. OE-5, NO. 2, April 1980, pp. 154-163.
[4] W. B. Wirin and R. A. Williamson,

Satellite Remote Sensing in the USSR: Past, Present, and Future, Remote Sensing Yearbook 1990, Chapter 3, Burgess Scientific Press, Basingstoke, UK, pp. 49-64.

[5] S. Chenard, Soviet Earth Obser-vation gets Less Remote, Interavia Space Markets, Vol. 6, No. 1, January/February 1990, pp. 11-22.

a 171 J. T. Richelson, The Future of Space

Reconnaisance, Scientific American, Vol. 264, No. 1, January 1991, pp. 38-45. [8] ERS-1 Special Issue, ESA Bulletin, European Space Agency, 8-10 rue MarioNikis, 75738 Paris Cedex 15, No. 65, February 1991. [9] E. P. W. Attema, The Active Microwave Instrument on-Board the ERS1 Satellite, Proceedings of the IEEE, Vol. 79, No. 6, June 1991, pp. 791-199.

[6] USA Today, 14 December 1988.

[15] R. K. Moore, J. P. Claasen, and Y. H. Lin, Scanning Spaceborne Synthetic Aperture Radar with Integrated Radiometer, IEEE Transactions on Aerospace and Electronic Systems, Vol. AES-17, NO. 3, 1981, pp. 410-421.
[16] K. Tomiyasu, C o n c e p t u a l Performance of a Satellite Borne, Wide Swath Synthetic Aperture Radar, IEEE Transactions on Geoscience and Remote Sensing, Vol. GE-19, No. 2, ApriI 1981, pp. 108-116. [17] Glavkosmos, Orbital Station "MIR: Complex of Remote Sensing of the Earth "PRIRODA", Moscow, 1990.

11-211

[18] N. A. Armand, A. A. Kalinkevich, B. G. Kutuza, and S. M. Popov, SAR Facilities for the "PRIRODA' Mission, Proceedings of the International Geoscience and Remote Sensing Symposium, IGARSS'91, Espoo, Finland, 3-6 June 1991, pp. 1790-1792.
[19] J. J. van Zyl et al., AIRSAR Reference Manual, Jet Propulsion Laboratory, California Institute of Technology, 1991. [20] C. E. Livingstone, A. L. Gray, R. K. Hawkins, and R. B. Olsen, CCRS C/XAirborne Synthetic Aperture Radar: an R and D Tool for the ERS-1 Time Frame, IEEE AES Magazine, October, 1988, pp. 11-16. [21] R. Horn, M. Werner, and B. Mayr, Extension of the DLR Airborne Synthetic Aperture Radar, E-SAR, to X-band, Proceedings of the Inter-national Geoscience and Remote Sensing Syniposium, IGARSSPO, Washington, DC, 20-24 May 1990, p. 2047. [22] R. Horn, C-Band SAR Results Obtained by an Experimental Airborne SAR System, Proceedings of the International Geoscience and Remote Sensing Symposium, IGARSS'89, Vancouver, Canada, 10-14 July 1989, pp. 2213-2216. [23] C. Dahme, R. Horn, D. Hounam, H. Ottl, and R. Schmid, Recent Achievements of DLR's Airborne Experimental SAR System and Image Processing Equipment, Proceedings of the Intemational Geoscience and Remote Sensing Symposium, IGARSS'YI, Espoo, Finland, 3-6 June 1991, pp. 245-246. [24] S. N. Madsen, E. L. Christensen, N. Skou, and J. Dall, The Danish S A R

System: Design and Initial Tests, IEEE Transactions on Geoscience and Remote Sensing, Vol. 29, No. 3, May 1991, pp. 417426. [25] J. Dall, J. H. J@rgensen, E. L. Christensen, and S. N. Madsen, A Realtime Processor for the Danish C-Band S A R , Proceedings of the International Geoscience and Remote Sensing Symposium, IGARSS'91, Espoo, Finland, 3-6 June 1991, pp. 279-282. [26] A. D. Nichols, J. W. Wilhelm, T. W. Gaffield, D. R. Inkster, and S. P:. Leung, A S A R or Real-time Ice Reconnaissance, IEEE Transactions on Geoscience and Remote Sensing, Vol. GE-24, No. 3, May 1986, pp. 383-389. I271 P. Hoogeboom, P. J. Koomen, H. Pouwels, and P. Snoeij, First Results from The Netherlands S A R Testbed "PHARS", Proceedings of the Intemational Gcoscience and Remote Sensing Symposium, IGARSS'91, Espoo, Finland, 3-6 June 1991, pp. 241-244. [28] A. D. Kozma, A. D. Nichols, R. F. Rawson, S. J. Shackman, C. W. Haney, and J. J. Shanne Jr., Multi-frequency, polarization S A R for Remote Sensing, Proceedings of the Intemational Geoscience and Remote Sensing Sym.posium, ICARSS'86, Zurich, Switzerland, 8-11 September 1986, ESA Publication SP-254, pp. 715-719. [29] R. Albrizio, P. Blonda. A. Nlazzone, F. Pasquali, G. Pasquariello, F. Posa, and N. Veneziana, Digital Processing of XBand VARAN-S Airborne SAR Images, Proceedings of the Intemational Geoscience and Remote Sensing Symposium, IGARSS'89, Vancouver, Canada, 10-14 July 1989, pp. 2203-2208.

R- I

BIBLIOGRAPHY This Bibliography was compiled to support Lecture Series 182 by the Fachinformationszentrum, Karlsruhe, D-7514 Eggenstein-Leopoldshafen 2 (Fiz Karlsruhe), Germany, in association with the Lecture Series Director, Dr W.Keydel.

TYPE 1/4/1 Quest Accession Number : 91A52169 91A52169 NASA IAA Journal Article Issue 22 Radarsat (AA)RANEY, R. K . ; (AB)LUSCOMBE, ANTHONY P.; (AC)LANGBAM, E . J.; (AD)AHMED, SHABEER (AA)(Canada Centre for Remote Sensing, Ottawa); (AB)(Spar Aerospace, Ltd., sainte-Anne-de-Bellevue, Canada); (AD)(Canadian Space Agency, Ottawa, Canada) 79, June 1991, p. 839-849. IEEE, Proceedings (ISSN 0018-9219). "01. 910600 p. 11 refs 32 In: EN (English) p.3832 Radarsat, the first Canadian remote-sensing spacecraft, is designed to provide earth observation information for five years. The Satellite is scheduled for launch in 1 9 9 4 . The only payload instrument is a (C-band) synthetic aperture imaging radar (SAR). 5.G-cm-wavelength Radarsat will gather data an command Cor up to 28 mi" during each cycle of it5 800-km (nominal) near-polar orbit. Image resolutions from 10 to 100 m at swath widths of 45 to 500 km will be available. The Radarsat mission is reviewed, and the design, characteristics, and implementation of the radar are introduced. Technical problems addressed include calibration, rapid data processing, the phased array antenna that provides controlled beam steering, and the first satellite implementation of a special radar technique knOwn as ScanSAR. I.E.

TYPE 1/4/2 Quest Accession Number : 91A52168 91A52168* NASA IAA Journal Article Issue 22 The SIR-C/X-sAR synthetic aperture radar system (=)JORDAN, ROLAND0 L.; (ABIHUNEYCUTT, BRYAN L.; (AC)WERNER, MARIAN (AB)(JPL, Pasadena, CA); (AC)(DLR, Institut fuer Hochfrequenztechnik, nberpfaffenhofen, Federal Republic of Germany) Jet Propulsion Lab., California Inst. of Tech., Pasadena. (JJ574450) IEEE, Proceedings (ISSN 0018-9219). VO1. 79, J u n e 1 9 9 1 , p. 827-838. Research supported by DLR. 910600 p. 12 refs 16 In: EN (English) p. 3832 SIR-C/X-sAR, a three-frequency radar to be flown on the Space Shuttle in September 1993, is described. The SIR-C System is a two-frequency radar operating at 1250 MHz (L-band) and 5300 MHz (C-band), and is designed to get four-polarization radar imagery at multiple surface angles. The X-SAN system is an X-band imaging radar Operating at 9600 M H Z . The discussion covers the mission concept; System design: hardware; RF electronics; digital electronics: command, timing. and telemetry; and testing. I.E.

TYPE 1/4/3 Quest Accession Number : 91A52166 91A5216G NASA IAA Journal Article Issue 22 Japanese Earth Resources Satellite-1 synthetic aDerture radar
~~~~ ~~

~~~,

~.

= - . I

TOkYOj ?AD)(Mitsubishi Electric Corb., Kamaiura Works, Japan); (AE)(NEC corp., space and Laser Communications Development niv., Yokohama, japan); (AF)(NEC Corp., Guidance and EleCtrO-Optics Div., Tokyo, Japan) IEEE, Proceedings (ISSN 0018-92191, VO1. 79, June 1991, p. 800-809. p.3832 910600 p. 10 refs 8 In: EN (English) The spaceborne L-band SAR to be launched on the Japanese Earth Resources Satellitel is described. The mission is mainlv dedicated to aeoloaical applications. The ground resolution of the processed image is designed as 18 m, and the off-nadir angle required to meet geological applications is 35 deg. The design and performance of the key system parameters are discussed, along with the reasons for choosing such design parameters. The antenna is a thin-flat-foldable configuration that has a 11.9-m by 2.2-m aperture when extended in orbit. The transmitter, receiver, and signal processor are all of the solid-state type to achieve high reliability of operation. They transmit an 1100-w peak (minimum) chirp pulse and receive the return echos and process the echo Signals into in-phase and quadrature data streams. I.E.

TYPE 1/4/4 Quest Accession Number : 91A52164 91A52164* NASA IAA Journal Article Issue 22 Magellan imaging radar mission to venus (AA)JOHNSON, WILLIAM T. K . (AA)(JPL, Pasadena, CA) Jet Propulsion Lab., California Inst. of Tech., Pasadena. ( ~ ~ 5 7 4 4 5 0 ) IEEE, Proceedings (ISSN 0018-92191, VO1. 79, June 1991, p. 777-790. 910600 p. 14 refs 21 In: EN (English) p.3831 The Magellan imaging-radar mapping mission has collected and processed data from the spacecraft in an elliptical orbit around Venus. A brief description is given of the mission and the spacecraft, followed by a more detailed description of the radar system design, which used earth-orbiting SAR experience and several innovations in its design to operate from an orbit around another planet. The radar sensor, ground processing, +data products are described. This multimode radar is the onlv science instrument on the mission and has the objective of mapping a; least 70 percent of the planet Surface. It has three modes: S A R , altimetry, and passive radiometry. The radar System has produced maps of almost all of the Venusian Surface with a resolution better than 600-m equivalent Optical line pair, and the best resolution obtained is equivalent to less than 300 m. Some of the early radar images are shown. I.E.
~ ~~~

TYPE 1/4/5 Quest Accession Number : 91A49343 91~49343 NASA IAA Journal article Issue 21 Speckle reduction in multipolarization, multifrequency SAR imagery (=)LEE, JONG-SEN; (AB)GRUNES, MITCHELL R.; (AC)MANGO, STEPHEN A. (AC)(U.S. Navy, Naval Research Laboratory, Washington, DC) (IGARSS ' 9 0 - IEEE International Geoscience and Remote sensing SympoSillm on Remote Sensing Science for the Nineties, 10th. University of Maryland, College Park, May 20-24, 1990) IEEE Transactions on Geoscience and Remote sensing (ISSN 0196-2892), vo1. 29, July 1991, p. 535-544. 910700 p. 10 refs 12 In: EN (English) p.3728 ~n algorithm to take advantage of this polarization diversity to suppress the speckle effect with much less resolution broadening than using spatial filtering is discussed. The coupling between polarization channels is minimized by using local intensity ratios. The degree of speckle reduction is similar to two-look or three-look processing. The same algorithm can also he used to process multifrequency polarimetric SAR. For three-frequency aircraft SAR data Speckle reduction equivalent to six-look processing can be achieved, and further reduction i s possible by applying speckle filters in the spatial domain. A vector Speckle filter which Operates simultaneously in the polarization and spatial domains is also tested. Experimental results with simulated polarimetric SAR as well as one-look and multilook parametric SAR data demonstrate the effectiveness Of these speckle reductions, with minimum resolution broadening and coupling between polarimetric and frequency channels. I.E.

TYPE 1 / 4 / 6 Quest Accession Number : 91A47233 gin47233 NasA IAA ~ o u r n a larticle I S S U ~ 20 Measurement of surface microtopography (AA)WALL, S. D.; (AB)FARR, T. G.; (AC)MULLER, J.-P.; (AD)LEWIS, P.; (AE)LEBERL, F. W. (AB)(JPL, Pasadena, CA); (AD)(University College, London, England); (AE)(Vexcel Corp., Boulder, CO) Jet Propulsion Lab., California Inst. of Tech., Pasadena. (55574450) Photogrammetric Engineering and Remote Sensing (ISSN 0099-1112). vol. 57, AUg. 1991, p. 1075-1078, Research supported by NASA, University Colleqe London, and vercel corp. 910800 p. 4 refs 24 In: EN (English) p.3472
~ ~~~ ~

Acquisition of ground truth data for use in microwave interaction modeling requires measurement of surface roughness sampled at intervals comparable to a fraction of the microwave wavelength and extensive enough to adequately represent the statistics Of a Surface unit. sub-centimetric measurement accuracy is thus required over large areas, and existing techniques are usually inadequate. A technique is discussed for acquiring the necessary photogrammetric data using twin film cameras mounted on a helicopter. I" an attempt to eliminate tedious data reduction, an automated technique was applied to the helicopter photographs, and results were compared to those produced by conventional stereogrammetry. Derived root-mean-square (RMS) roughness for the Same stereo-pair was 7.5 Cm for the automated technique versus 6.5 cm for the manual method. The principal source of error is probably due to vegetation in the Scene, which affects the automated technique but i s ignored by a human operator. Author

TYPE 1/4/7 Quest Accession Number ; 91A43129 91A43129 NASA IAA Conference Paper Issue 18 Complex phase error and motion estimation in synthetic aperture radar imaging (AA)SOUMEKH, M.; (AB)YANG, H. (aB)(State University of New York at Buffalo, Amherst) NSF MIP-90-04996 IN: Image processing algorithms and techniques 11; Proceedings of the Meeting, San Jose, CA, Feh. 25-Mar. 1. 1991 (A91-43127 18-35). Bellingham, WA. society of Photo-Optical Instrumentation Engineers, 1991. p. 104-113. 910000 p. 10 refs 7 In: EN (English) p.
3116

TYPE 1/4/8 Quest Accession Number : 91A42468 9 1 ~ 4 2 4 6 8 NASA IAA Journal Article I S S U ~17 Estimation of ocean wave SDeCtra usina two-antenna SAR SvStemS (AA)LYZENGA, D. R.; (AB)BENNETT, J. R : (AB)(Michigan, Environmental Research Institute, Ann Arbor) N00014-90-C-0071 IEEE Transactions on Geoscience and Remote (ISSN 0196-2892). Vol. 29, May 1991, I ) . 463-465. 910500 13. 3 In: EN (English) p.2983 Data from synthetic aperture radar (SAR) systems can be Used to estimate ocean wave directional spectra, but the method is limited by nonlinearities associated with the velocity bunching mechanism and the azimuth falloff effect, which reduces the range of azimuth wavelengths that can he observed. A theoretical analysis Which Suggests that the use of two or more receive antennas, Spaced in the along-track direction, may reduce these limitations is presented. Specifically, the hand of usable azimuth wavenumbers is Shifted by an amount proportional to the antenna spacinq, so that a broader range of wavenumbers can he covered by combining the spectrum obtained from the two-antenna siqnals with the conventional image spectrum. The angular dependence Of the velocity modulation mechanism is also modified to include purely range-traveling waves in the two-antenna case. I.E.
~

Attention is given to a SAR wave equation-based system model that accurately represents the interaction Of the impinging radar signal with the target to be imaged. The model is used to estimate the complex phase error across the synthesized aperture from the measured corrupted SAR data bv combinina the two wave eouation models qoverninq the collected SAR data at two temporal frequencres of the radar signal. The SAR System model shows that the motion of an abject in a static scene results in coupled Doppler shifts in both the temporal frequency domain and the spatial frequency domain Of the synthetic aperture. The velocity of the moving object is estimated through these two Doppler shifts. It is Shown that once the dynamic target's velocity is known, its reconstruction can be formulated via a squint-mode SAR geometry with parameters that depend upon the dynamic target's velocity.
O.G.

TYPE 1/4/9 Quest Accession Number : 91A42467 91A42467 NASA IAA Journal Article Issue 17 Terrain influences in SAR backscatter and attempts to their correction (AA)BAYER, THOMAS; (ABIWINTER,RUDOLF; (AC)SCHREIER, GUNTER (AC)(DLR, Oberpfaffenhofen, Federal Republic Of Germany) IEEE Transactions on Geoscience and Remote sensing (ISSN 0196-2892). VO1. 29, May 1991, P. 451-462. 910500 p. 12 refs 19 In: EN (English) $2.2970
SAR images reveal radiometric image distortions that are caused by terrain undulations. 'The authors present the results of a study extracting and investigating the various components of these terrain influences. An imaging model is set up for the geometric rectification of the SAR image and for a reconstruction of the imaging geometry. A prerequisite for the setup of this model is the use of a digital elevation model. Eight different geometric parameters are derived and investigated for their influence on grey-value variations in the geocoded SAR image. Image grey-value variations of three major land-use Classes-forest, agricultural land, and urban/suburban areas-are examined. Empirical models of the SAR-backscatter variations are used to describe the relations between image grey values and various geometric parameters. The models are used for a numerical estimation of the terrain influence on the backscatter variations in the SAR image. The models allow the derivation of 13 different functions for the correction of the relief-induced radiometric image distortions. These functions are applied to test areas within the SAR scene under investigation, and their correction effects are described and compared numerically as well a5 visually. I.E.

TYPE 1/4/10 Quest Accession Number : 91A42463 91A42463' NASA IRA Journal Article Issue 17 The Danish SAR system - Design and initial tests IAAlMADSEN, SOREN N.; (AB)CHRISTENSEN, ERIK L.: (AC)SKOU, NIELS; l___,_. (ADIDALL, SORGEN (AA)(JPL, Pasadena, CA); (AD)(Technical University of Denmark, Lyngby) Jet Propulsion Lab., California Inst. of Tech., Pasadena. (55574450) IEEE Transactions on Geoscience and Remote sensing (ISSN 0196-2892), vol. 29, May 1991, p . 417-426. Research supported by Thomas B. Thriges " : EN (English) p.2909 Foundation. 910500 p. 10 refs 15 I
In January 1986, the design of a high-resolution airborne C-band sm started at the Electromagnetics Institute of the Technical University of Denmark. The initial system test flights took place in November and December 1989. The authors describe the design of the system, its implementation, and its performance. They Show how digital technology has been utilized to realize a very flexible radar with variable resolution, swath-width. and imaging geometry. The motion-compensation algorithms implemented to obtain the high resolution and the Special features built into the system to ensure proper internal calibration are outlined. The data processing system, developed for image generation and quality assurance, is sketched, with special emphasis on the flexibility of the system. Sample images and a preliminary performance evaluation are presented, demonstrating that the design goals have been met. The ongoing system upgrades and the planned Scientific utilization of the C-band SAR are described. I.E.

TYPE 1/4/11 Quest Accession Number : 91A42461 91A42461 NASA IAA JOUrnal Article Issue 17 Doppler frequency estimation and the cramer-~aobound (AA)BAMLER, RICHARD (AA)(DLR, Oberpfaffenhofen, Federal Republic of Germany) IEEE Transactions on Geoscience and Remote Sensing (ISSN 0196-2892). VOl. 29, May 1991, P. 385-390. 910500 p. 6 ref5 19 In: EN (English) p.2909 The authors address the problem of Doppler frequency estimation in the presence of Speckle and receiver noise. ~n ultimate accuracy bound for Doppler frequency estimation is derived from the Cramer-Rao inequality. It is shown that estimates based on the correlation of the signal power an arbitrary weighting function are approximately spectra with Gaussian-distributed. Their variance is derived in terms of the weighting function. It is shown that a Special case of a correlation-based estimator is a maximum-likelihood estimator that reaches the Cramer-Rao bound. These general results are applied to the problem of Doppler centroid estimation from SAR (synthetic aperture radar) data. Different estimators known from the literature are investigated with respect to their accuracy. Numerical examples are presented and compsred with eiperimieiitdi results. I.E.

TYPE 1/4/12 Quest Accession Number : 91A41831 91A41831 NASA IAA Conference Paper Issue 17 Detection of slowly moving targets with airborne radar Entdeckunq langsam bewegter Ziele mit luftgetragenem Radar (AA)ENDER, JOACHIM; (AB)XLEMM, RICHARD (AB)(Forschungs9esellschaft fuer angewandte Naturwissenschaften, FOrSchungSinStitUt fuer Funk und Mathematik, Wachtberg-Werthhoven, Federal Republic Of Germany) IN: Radar Symposium. 7th. Ulm, Federal Republic of ~ermany, oct. 10-12, 1989, Reports (A91-41822 17-32). Duesseldorf, DeutSche Gesellschaft fuer Ortung und Navigation, 1989, p. 161-186. I n German. 890000 p. 26 refs 17 In: GM (German) p.2902 It is shown here that recent airborne radar systems can detect moving targets amid clutter by using optimal filtering with an Order of magnitude improvement over previous systems. One-channel and multichannel methods are compared in terms of theoretical limits. The optimal clutter suppression method in multichannel systems is investigated by analysis in the frequency and time regions. Methods for obtaining Suboptimal filter procedures with reduced computational expense are determinedC.U.

TYPE 1/4/13 Quest Accession Number : 91A41830 91A41830 NASA IAA Conference Paper ISSUe 17 Calibration of SAR systems for remote sensing using X-SAR/sIR-C a5 an example Die Kalibrierung von Synthetik-ApeTtUT-Radar-Systemen der Fernerkundung am Beispiel des X-sAR/sIR-C (AA)KIETZMANN, H.; (AB)BLOETSCHER, H. (AB)(DLR, Institut fuer Hochfrequenztechnik, oberpfaffenhofen, Federal Republic of Germany) IN: Radar Symposium, 7th. Ulm, Federal Republic of Germany, Oct. 10-12, 1989, Reports (A91-41822 17-32). Duesseldorf, DeUtSChe Gesellschaft fUer Ortung und Navigation, 1989, p. 147-159. In German. 890000 p. 13 refs 9 In: GM (German) p.2902 Calibration of an SAR system for quantitative signature analysis in remote sensing is illustrated here in step-by-step fashion f o r the X-SAR/sIR-C sensor. Both internal and external calibration are addressed. The determination of operational antenna diagrams is shown. C.D.

TYPE 1/4/15 Quest Accession Number : 91A41828 91A41828 NASA I A A Conference Paper ISSUe 17 Concept and results of the Azimuth Quick-Look Proce5sor f o r aircraft SAR of the DLR Konzept und Ergebnisse d,es Azimut Quick-Look P I O L B S S O ~ S fuer das FlugZeUg Synthetik Apertur Radar der DLR (AA)MOREIRA, ALBERT0 (AA)(DLR, Institut fuer Hochfrequenztechnik, Wessling, Federal Republic of Germany) IN: Radar symposium, 7th. Ulm, Federal Republic of Germany, OCt. 10-12, 1989, Reports (A91-41822 17-32). Duesseldorf, DeutSChe Gesellschaft fuer ortung und Navigation, 1989, p. 113-126. In German. 890000 p. 14 refs 11 In: GM (German) p.2861 A real time azimuth processor has been built for the aircraft-borne E-SAR System (Experimental Radar System with Synthetic Aperture). The processor uses an unfocused correlation procedure which greatly simplifies the data processing and permits operation in real time without much use of hardware. A new algorithm has been developed which improves image parameters except for resolution and is comparable with available focused processor procedures. A number of real time images with postprocessing are Shown which meet expectations and validate the algorithm. C.D.

TYPE 1/4/17 Quest Accession Number : 91A41825 91A41825 NASA IAA Conference Paper Issue 17 Experimental X-band synthetic aperture radar in aircraft Experimentelles X-Band Synthetik Apertur Radar im Flugzeug (AAIHORN, RALF; (AB)WERNER, MARIAN (AB)(DLR, InStitut fuer Hochfrequenztechnik, Oberpfaffenhofen, Federal Republic of Germany) IN: Radar Symposium, 7th. Ulm, Federal Republic of Germany. Oct. 10-12, 1989, Reports (A91-41822 17-32). DueSSeldorf, DeUtSChe Gesellschaft fuer ortung und Navigation, 1989, p. 59-63. i n German. 890000 p. 5 I": GM (German) p.2861 An experimental aircraft SAR for the X-band is described. The overall system, technical specifications, and experimental results are briefly reported. A block Switching diagram of the system is shown. C.D.

TYPE 1/4/18 Quest Accession Number : 91A39775 91A39775 NASA IAA Journal Article Issue 16 An efficient S A R parallel processor (AA)FRANCESCHETTI, GIORGIO; (AB)SCHIRINZI, GILDA; (AC)PASCAZIO, VlTO; (ADIMAZZEO. ANTONINO: (AEIMAZZOCCA. NICOLA . . ' (AB)(CNRi Istituto di Ricerca' per 1'Elettromagnetismo e i Componenti Elettronici, Naples, Italy); (AC)(lstituto Universitario Navale, Naples, Italy); (AEl(Napoli, Universita, Naples, Italy) IEEE Transactions On Aerospace and Electronic Systems (ISSN 0018-9251). V O 1 . 27, March 1991, p. 343-353. 910300 p. 11 ref5 16 In: EN (English) p.2703 A parallel architecture especially desiqned for a synthetic-aperture-radar (SARI proceising algorithm- based on an appropriate two-dimensional Cast Fourier transform (FFT) code is presented. The algorithm is briefly summarized, and the FFT code is given for the one-dimensional case, although a l l results can be immediately generalized to the double FFT. The computer architecture, Which consists of a toroidal net with transDUters on each node. is described. Parametric expressions for the comp&tional time of the net v e r s i s the number of nodes are derived. The -architecture allows drastic reduction of the processing time, preserving elaboration accuracy and flexibility. I.E.
~~~~~ ~~~~~

TYPE 1/4/19 Quest Accession Number : 91A38977 91A38977# NASA IAA Conference Paper Issue 16 Present and future imaging radar systems 1AA)RANEY. R. KEITH (AAICRADARSAT Project office, Ottawa, Canada) IN: Space commercialization: Satellite technology; symposium on Space Commercialization: Roles of Developing Countries, Nashville, TN, Mar. 5-10, 1989. Technical Papers (A91-38976 16-12), Washinmton. - . DC. American Institute of Aeronautics- and Astronautics, Inc., 1990. p . 1-10. 9ooooo p . 10 In: EN (English) p.2677 Synthetic aperture radar (SARI Systems available for the use of planners in developed and developing countries for the purpose Of meeting future resource-observation requirements are reviewed, and attention is given to nominal image parameters, global system development, and access to data. FOCUS is placed on such airborne SAR Systems available for civilian use as the STAR 1 (Canada), and Varan-S (France). Expected SAR-carrying Spacecraft including ERS-1 (ESA), J-ERS-1 (Japan), and Radarsat (Canada), as well as research-oriented SAR Systems to be flown on the Shuttle and on polar-Orbiting platforms. It is noted that SAR data is well suited for the quantitative observation of critical national and global resources such as tropical forests and will be a primary source of information for many resource-monitoring and analysis responsibilities. V.T.

TYPE 1/4/20 Quest Accession Number : 91A33211 91A33211# NASA 1AA Journal Article Issue 13 The unveiling of Venus - Magellan's synthesis radar penetrates the cloud Die Entschleierung der Venus - MagellanS Synthese-radar durchdringt die Wolkendecke (AA)FISCHER, DANIEL Sterne und Weltraum (ISSN 0039-1263). v01. 30, April 1991, p. 226-230, 232, 233. In German. 910400 p. 7 In: GM (German) p.2255 The revelation of the Surface of Venus by the Magellan Synthetia radar is discussed. The highlights of the discoveries are shown and desoribed, including the long strips called 'noodles', the complex geological formation called the Phoebe region, the mountainous Lakshmi region which contains evidence Of Plate tectonics, and the Themis Regio highland region, which may have formed by processes analogous to those which made the Hawaiian islands. Mysterious phenomena, like the apparent youth of many of the Craters, are addressed. C.D.
C".,Sr

TYPE 1/4/21 Ouest Accession Number : 911132322 - 91A32322 NASA IAA Journal Article 1SSUe 12 The modified beta density function as a model for synthetic aperture radar clutter statistics (AA)MAFFETT, ANDREW L.; (ABIWACKERMAN, CHRISTOPHER C. (AB)(Michigan, Environmental Research Institute, Ann Arbor1 IEEE Transactions on Geoscience and Remote Sensing (ISSN 0196-2892), v01. 29, March 1991, p. 277-283. 910300 p. 7 refs 17 In: EN (English) p.2019 The authors show that the modified beta distribution function is an adequate model for the underlying distribution function of the random variable used to model SAR imaqe data. The model represents a range of SAR returns from different ice types by using a simple change in its parameter. The ability is explained by describing the distribution ~ the modified beta functions in width, modified skewness S D ~ Cwhere function covers a 'region, while the other, more common, distribution functions cover Only a curve. A procedure for comparing simple distribution functions with analytical functions specifically for digitized SAR data is presented, and the modified beta model is tested on 166 subsets drawn from three SAR collections o v e r different ice types and over open water. The modified beta function can model essentially all of the SAR subsets, where the other more common densities Cannot. Some classification of ice tYDes usins the parameters from the modified beta function are provided. I.E.
~ ~~~~ ~ ~~

TYPE 1/4/22 Quest Accession Number : 91N30395 91N30395# NASA STAR Technical Report Issue 22 Cross-calibration between airborne SAR sensors (AA)ZINK, MANFRED DeutSChe FOrSChUngSanStalt fuer Luft- und Raumfahrt, Oberpfaffenhofen (00776452) Abt. HF-SySteme. (Germany, F . R . ) . DLR-FB-91-10; ISSN-0939-2963; ETN-91-99791 910200 p. 57 In: EN (English) Avail: NTIS HC/MF A04; DLR, Wissenschaftliches Berichtswesen, VB-PL-DO, Postfach 90 60 58, 5000 Cologne, Fed. Republic of Germany, HC 22 DM p.3663 synthetic Aperture Radar (sAR) system performance and experience in signature evaluation increase, quantitative analysis becomes more and more important. Such analyses require an absolute radiometric calibration of the complete sAR system. To keep the expenditure on calibration of future multichannel and multisensor remote sensing systems (e.g., x-sAR/sIR-c) within a tolerable level, data from different tracks and different sensors (channels) must be cross calibrated. The 1989 joint E-SAR/DC-8 S A R calibration campaign gave a first opportunity for such an experiment, including cross sensor and cross track calibration. A basic requirement for succesful cross calibration is the stability of the SAR systems. The calibration parameters derived from different tracks and the polarimetric properties of the uncalibrated data are used to describe this stability. Quality criteria f o r a successful cross calibration are the agreement of alpha degree values and the consistency of radar cross sections of equally sized corner reflectors. Channel imbalance and cross talk provide additional quality in case of the polarimetric DC-8 SAR. ESA
AS
SAR

TYPE 1/4/23 Quest Accession Number : 91A29141 91A29141 NASA IAA Journal Article ISSUB 11 Accuracy of using point targets for SAR calibration (AAIULANDER. LARS M. H. i A A j (Ciiida'centre for Remote sensing, Ottawa1 IEEE Transactions on Aerospace and Electronic Systems (ISSN 0018-9251). "01. .. . . 7 - .7 , ~ .ian. 1991. n. 139-148. Research SuUDOrted bv Swedish Board for Space Activities. 916100 p . 10 refs 22 1 n : ' E N (English) p.1744
~~~~~ ~~~~~~

The peak and integral methods for radiometric calibration of a Synthetic aperture radar ( S A R ) using reference point targets are analyzed. Both calibration methods are Shown to be unbiased, but the peak method requires knowledge of the equivalent rectangle system resolution which is sensitive to system focus. Exact expressions for the RMS errors of both methods are derived. It is shown that the RMS error resulting from the peak method is always smaller than or equal to that from the integral method for a well-focused system. However, for robust radiometric calibration of SAR, or when nonlinear phase errors are present, the integral method is recommended, because it does not require detailed knowledge Of the impulse response and the resulting RluS error is not dependent on system focus. I.E.

TYPE 1/4/24 Quest Accession Number : 91A25943 91A25943 NASA IAA Journal Article Issue 09 The radiometric quality Of AgriSAR data (AA)QUEGAN, s . : (AB)YANASSE, C.; (AC)DE GROOF, H.; (AD)CHURCHILL, P. N.; (AE)SIEBER, A. J. IABl(Sheffie1d. universitv. , ,. .. Enalandl: . lAE)(CEC. . .. . Joint Research Centre. 1spra. Italy) (Remote Sensing Society, Workshop on S A R Processing and Information Extraction from SAR Images, University Of Sheffield, England, Mar. 2 2 , 1989) International Journal of Remote Sensinq (ISSN 0143-1161). vol. 12, Feb. 1991, p. 277-302. Research sponsored by the Conselho Nacional de Desenvolvimento Cientifico e Tecnologico. 910200 p. 26 refs 21 In: EN (English) p.1418
~~

The AgriSAR '86 data set for the Feltwell, U.K. test site is assessed for its ability to provide information on the spatial and temporal variation of backscatter from growing crops. The image data is shown to contain Several sources of radiometric distortion which affect any attempt at image calibration. Of these, the simplest to correct is that caused by range interpolation in the amplitude data. All interpolated pixels should be discarded from the amplitude data supplied by CNES. Correction from Slant ranae to sround range is ComDaratively straiqhtforward. Effects due to variations in antenna pattern, in the range-and azimuth direction, require assumptions about the angular variation of the backscatter of crops and statistical homogeneity of ground cover before they can be removed. The properties of system noise appear complicated and are not yet properly characterized, but interfere with corrections for the antenna pattern. Offsets in the two channels of the complex data also interfere adversely with antenna pattern corrections. The size of these offsets cannot be estimated reliably from the AgriSAR amplitude data, and complex data, and complex data should be supplied as a standard product. NO sound basis for inter-image comparison has been found. Author

M
4

TYPE 1/4/25 Quest Accession Number : 91A25942 91A25942 NASA IAA Journal Article issue 09 SAR motion compensation using autofocus (AAIBLACKNELL,D.: (AB)QUEGAN, s. (AA)(GEC-Marconi Research Centre, Chelmsford, England); (AB)(Sheffield, University, England) (Remote Sensing Society, Workshop on SAR Processing and Information Extraction from SAR Images, University Of Sheffield, England, Mar. 22, 1989) International Journal Of Remote Sensing (ISSN 0143-11611, vo1. 12, Feb. 1991, P. 253-275. Research Supported by the Ministry of Defence Procurement Executive. 910200 P. 23 refs 19 In: EN (English) p.1370
~~ ~~

TYPE 1/4/26 Quest Accession Number : 91A25941 91A25941 NASA IAA Journal . ~ . . \. r. + _ i_ r i. e TCWO .--__ no A new approach to range-Doppler SAR processing (AA)SMITH, A. M. (AA)(SD-Scicon. London, England) (Remote Sensing Society. Workshop on S A R Processing and Information Extraction from SAR Images, University of Sheffield, England, ~ a r . 22, 1989) International Journal Of Remote Sensing (ISSN 0143-1161). "01. 12, Feb. 1991, p. 235-251. Research Supported by the Ministry of Defence. 910200 p. 17 refs 6 In: EN (English) p.1370
~ ~~

__

m
2

Conventional motion compensation Schemes Correct for unwanted synthetic aperture radar (SARI platform motions using information from an inertial measurement unit (IMU). Autofocus techniques, which focus SAR images, produce an 'autofocus parameter' which is related to the platform motion. In this paper, Strong evidence is presented to Support the assumption that the Contrast optimization autofocus algorithm behaves as a least-squares quadratic fitting to the SAR platform trajectory. Using this assumption, the relationship between the autofocus parameter and aCrOSS-traCk accelerations of the SAR platform is derived. This allows the SAR platform motion to be estimated from the autofocus parameter measurements and incorporated in a motion compensation instead of IMO measurements. Three implementations of motion compensation using autofocus are compared and the achievable image quality is quantified. author

This paper presents a general analysis Of frequency-domain SAR Processing based on the relationship between the phase of the two-dimensional Fourier transform of a ooint resoonse to its ranno-time history. The paper demonstrates how this pkovides an appropriate basis for the design Of a coherent strip-mode Processor, free of geometric or phase distortion and artefacts. and Without excessive computational cost. The relevance Of the analysis to ambiguity estimation and the processing of very long-integration-time SAR data is indicated. Author
~~

TYPE 1/4/27 Quest Accession Number : 91A25413 91825413 NASA IAA Conference Paper Issue 09 Estimating the residual error of the reflectivity displacement method for aircraft motion error extraction from S A R raw data (AAIMOREIRA. JOAO (AA)(DLR, Institut fuer Hochfrequenztechnik, Oberpfaffenhofen, Federal Republic Of Germany) IN: IEEE 1990 International Radar Conference, Arlington, VA, M ~ Y 7-10, 1990, Record (A91-25401 09-32). New York, Institute Of Electrical and . 70-75. 900000 p. 6 refs 9 ~ n : EN Electronics Engineers, I n c . , 1990. p (English) p.1320 The performance Of the reflectivity displacement method is reported. The reflectivity displacement method extracts all the necessary motions of the aircraft from the radar backscatter signal using a new radar configuration and new methods for evaluating the azimuth spectra of the radar signal. Hence, an inertial navigation system is unnecessary for many applications. An error analysis of this method is carried out, and a comparison Of two processed images with and without motion Compensation is shown, proving the estimated performance. I.E.

TYPE 1/4/28 Quest Accession Number : 91.425411 91A25411 NASA I A A Conference Paper Issue 09 An improved multi-look technique to produce SAR imagery (AAIMOREIRA. ALBERT0 i-1 ~ D L R , institut fuel Hochfrequenztechnik, Oberpfaffenhofen, Federal Republic of Germany) IN: IEEE 1990 International Radar Conference, Arlington, VA, May 7-10, 1990, Record (A91-25401 09-32). New York, Institute of Electrical and Electronics Engineers, Inc., 1 ~ 9 0 ,p. 57-63. 900000 p. 7 refs 20 ~ n : EN (English) p.1361
A multilook technique for improving the radiometric resolution in SAR image formation without altering the geometric resolution of the impulse response is proposed. This technique is based on the formation of looks with different bandwidths. The final image is formed by giving each look a proper size and weighting and then adding them incoherently. The looks with larger bandwidth contribute to an improvement of the overall geometric resolution, while the looks with smaller bandwidth improve the overall radiometric resolution. The equivalent number of looks is more than 2.3 times the number Of independent looks and is superior to conventional multilook processing with overlapping. An algorithm for efficient Processing ,using the proposed technique is presented, and its valirlity is proved :mq* empaeisun and analysis. I.E.

TYPE 1/4/29 Ouest Accession Number : 91A25410 - 91A25410 NASA IAA Conference Paper Issue 09 Feasibility of a synthetic aperture radar with rotating antennas (ROSAR) (AA)KLaUSING, HELMUT; (AB)KEYDEL, WOLFGANG (AA)(MBB GmbH, Ottobrunn, Federal Republic Of Germany); (AB)(DLR, Institut fuer Hochfrequenztechnik, Wessling, Federal Republic Of Germany) IN: IEEE 1990 International Radar conference, Arlington, VA, ~ a 7-10, y 19%. Record lA91-25401 09-321. New York. Institute of Electrical and ...., Electronics Engineers, Inc., 1996, p. 51-56. Research sponsored by DLR and MBB GmbM. 900000 p. 6 refs 5 In: EN (English) p.1360
~

TYPE 1/4/30 Number : 91A20526 Quest Ac&ssion 91A20526* NASA IAA Journal Article IsSue 07 Multiole imaae SAR shaoe-from-shadins (AA)THOMAS. 5.; (AB)KOBER, w.; (ACILEBERL. F. (AC)(Vexcel Corp., Boulder, CO) Vexcel Corp., Boulder, co. (V0987612) JPL-957955: JPL-958594 Photoarammetric Ensineerins and Remote sensins (ISSN 0099-1112). V O ~ . 57, 1991, p. 51-59. s i o i o o p. 9 refs 26 In: EN (English) p.1068

an.

ROSAR (rotor-SARI is a synthetic aperture radar concept for pilot sight target detection and target localization with high resolution. ROSAR is based on illuminating/receiving antennas placed at the tips of helicopter rotor blades. The ROSAR concept has potential benefits for civil and military helicopter-borne imaging applications. The concept has two main potential benefits: the imaging field of view is 360 deg, and there is no need for a forward velocity of the carrier platform. As opposed to SAR systems baaed on linear movement of the antenna, ROSAR imaqing is based on jynthetic apertures of a circular shape. Thus, the image formation process requires a polar format processing architecture. I.E.

A technique for combining radar image shape-from-shading with stereo radargrammetry to produce terrain Surface models using multiple SAR images is described. This technique is expected to be of use to reconstruct surface Shape from Magellan images of planet Venus, and to refine the results of terrestrial radar measurements. Local variation in pixel Shading is an indicator Of terrain slope changes. This variation in pixel Shading offers an opportunity for increasing the detail of terrain mapping over that which is available from stereo radargrammetry alone. Shape-from-shading can potentially provide a relative change in height at each pixel. This leads to a dense set of height measurements and a more local terrain shapes. However, faithful rendition of the shape-from-shading needs some type Of boundary values or external terrain low-frequency information to Succeed. These can be obtained from stereo or from altimeter measurements. Author

TYPE 1/4/31 Quest Accession Number : 91A15997 91A15997 NASA IAA Conference Paper Issue 04 Further results of radiometric calibration of a multifrequency airborne sAR system (AA)KASISCHKE, ERIC S . ; (AB)GINERIS, DENISE J. (AB)(Michigan, Environmental Research Institute, Ann Arbor) IN: Quantitative remote sensing: An economic tool for the Nineties; 8 9 and Canadian symposium on Remote Sensing, 12th. Proceedings of IGARSS ' Vancouver, Canada, July 10-14, 1989. Volume 5 (A91-15476 04-43). New York, Institute Of Electrical and Electronics Engineers, 1989, p. 2897-2900. 890000 p. 4 refs 7 In: EN (English) p.500 Further results from field experiments to radiometrically correct a multifrequency airborne SAR system are presented. Three frequencies of SAR data (x- , C-, and L-band) at w polarization were collected over a target array of calibrated trihedral Corner reflectors over a 2-week period in summer 1988. For all three frequencies, data collected on one date are utilized to absolutely calibrate data on a second date. The rms errors for this calibration procedure are shown to be less than 1 dB for all three frequencies. In addition, for the L - w channel, it is shown that the System stability over the entire period of calibration flights (which is the limiting factor for absolute and relative between-scene calibration) was also less than 1 dB. I.E.

TYPE . .~1/4/32 Quest Accession Number : 91A15996 91A15996* NASA IAA Conference Paper Issue 0 4 calibration of quadpolarization SAR data using backscatter Statistics lAAlKLEIN. JEFFREY D.
~

~I

~I
~~

. ~ ~ ~ ~ ,, (AA)( ~ p ~ , ' p a s a d e n aCA)

~~~~

Jet Propulsion Lab., California Inst. of Tech., Pasadena. (55574450) IN: Quantitative remote sensing: A n economic tool for the Nineties; Proceedings of IGARSS '89 and Canadian Symposium on Remote Sensing, 12th. Vancouver, Canada, July 10-14, 1989. Volume 5 (A91-15476 04-43). New York, Institute of Electrical and Electronics Engineers, 1989, p. 2893-2896. 890000 p. 4 refs 11 In: EN (English) p.500 A new technique is described for calibration Of complex multipolarization SAR imagery. scatterer reciprocity and lack of correlation between like- and cross-polarized radar echoes for natural taraets are used to remove cross-oolarized contamination in the radar data c h a k l s without the use of known ground targets. If known targets are available, all data channels can be calibrated relative to one another and absolutely as well. The method is verified with airborne SAR data. I.E.

TYPE 1/4/33 Quest Accession Number : 91A15994 91A15994 NASA IAA Conference Paper Issue 04 Comparison of SAR polarimetric calibration techniques using clutter (AA)SHEEN, D. R.; (ABIKASISCHKE. E. S . (AB)(Michigan, Environmental Rekearch Institute. A " " Arbor) IN: Quantitative remote sensing: ~n economic tool for'the Nineties: Proceedings of IGARSS '89 and Canadian symposium on Remote Sensing, 12th. VancoUVer, Canada, July 10-14, 1989. Volume 5 (A91-15476 04-43). New York, Institute Of Electrical and Electronics Engineers, 1989, p. 2885-2888. Research supported by the Environmental Research Institute of Michigan. 890000 p. 4 refs 8 In: EN (English) p.500 Polarimetric calibration can require several reference targets. The polarimetric scattering properties of clutter can be exploited in calibration. Because of the cost of reference targets and the effort required to deploy them, using clutter data is desirable for some aspects Of calibration if it can be readily used. Polarimetric L-band imagery collected by the NADC/ERIM P-3/sAR system over a forested region is examined. Reference reflectors deployed for calibration included trihedrals, dihedrals at a variety Of orientations, and active radar calibrators. The imaged areas Consisted of forest and grassy fields, as well as some crop land. Phase calibration methods using these clutter regions can be verified by comparison to methods using reference targets. Once verified, the clutter statistics are used to extend calibration Spatially over the image. After the image is calibrated, the signatures of various targets are compared to expected signatures, showing good aareement. I.E.

TYPE 1/4/34 Quest Accession Number : 91A15942 91A15942 NASA IAA Conference Paper Issue 04 A phase preserving sAR processor (AAIRANEY, R. KEITH: (AB)VACHON, PARIS w. (AB)(Canada Centre for Remote Sensing, Ottawa) IN: Quantitative remote sensing: An economic tool for the Nineties; Proceedings of IGARSS ' 8 9 and Canadian symposium on Remote Sensing, 12th, Vancouver, Canada. July 10-14, 1989. Volume 4 (A91-15476 04-43). New York, Institute of Electrical and Electronics Engineers, 1989, p. 2588-2591. 890000 p. 4 refs 6 In: EN (English) p.499 SAR image phase information is necessary to Support many advanced sAR applications. The phase information in the complex image for conventional DD~~= Drocessors is not a h n = t estimate of scene phase. A SAR range D o ~. - m ..-___ .. processor specikically designed to preserve phase informatbn is being developed at the Canada Centre for Remote Sensing. In addition to preserving vital phase information, this processor can support large degrees of range curvature and rancle mioration. Therefore, it is possible, in^ principle; to use this processor for satellite SAR data, high-resolution airborne SAR data, and both squint-mode and spotlight-mode S A R data. The theory is summarized, and early results are presented. .~

3i

~~~

l~...~

1.E.

'TYPE 1/4/35 Quest Accession Number : 91A15937 91A15937* NASA IAA Conference Paper Issue 0 4 Doppler centroid estimation ambiguity for synthetic aperture radars (AA)CHANG, C. Y.; (AB)CURLANDER, J. C. (AB)(JPL, Pasadena, CA) Jet Propulsion Lab., California Inst. of Tech., Pasadena. (55574450) ~~~. . IN: Quantitative remote sensino: An economic tool for the Ninetie=. Proceedings of IGARSS ' 8 9 and Canadian Symposium on Remote Sensing, 12th. Vancouver, Canada. July 10-14, 1989. volume 4 (A91-15476 04-43). N ~ W York, Institute of Electrical and Electronics Engineers, 1989, p. 2567-2571. 890000 p . 5 refs 7 In: EN (English) p.499
~~ ~~ ~~~

TYPE 1/4/36 ,~~ Quest Accession Number : 91A15882 91A15882 NASA IAA Conference Paper Issue 04 C-band SAR results obtained by an experimental airborne SAR sensor (AAIHORN, R. (AA)(DLR, InStitut fuer Hochfreouenztechnik, Oberpfaffenhofen, Federal Republic of Germany) IN: Quantitative remote sensing: An economic tool for the Nineties: Proceedings of IGARSS ' 8 9 and Canadian Symposium on Remote Sensing, 12th. York, Vancouver, Canada, July 10-14, 1989. Volume 4 (A91-15476 04-43). ~ e w Institute of Electrical and Electronics Engineers, 1989, p. 2213-2216. 890000 p . 4 In: EN (English) p.498
I

~~~~

A technique for estimation of the Doppler centroid of an SAR in the presence of large uncertainty in antenna boresight pointing is described. Also investigated is the image degradation resulting from data processing that uses an ambiguous centroid. TWO approaches for resolving ambiguities in Doppler centroid estimation (DCE) are presented: the range cross-correlation technique and the multiple-PRF (pulse repetition frequency) technique. Because other design factors control the PRF selection for S A R , a generalized algorithm is derived for PRFS not containing a common divisor. AD example using the SIR-c parameters illus.trates that this algorithm i ; c=p-b:e of resolving the C-band OCE ambiguities for antenna pointing uncertainties of about 2-3 deg. I.E.

Described is an airborne experimental SAR currently under development by the DLR. It allows the Study Of the SAR method and its problems, Such as motion error correction and overall System calibration. The sensor is designed to Owerate aboard a DO 228 aircraft . . . ~ in either L- or C-band. A first series of-flight experiments in L-band was completed successfully in spring 1988. The C-band System installation onboard the aircraft was completed in October 1988, and first flight experiments were carried out over test areas in Southern Germany. The C-band front-end represents a first step towards an active array. The amount of quantization Bnd saturation noise is minimized by adapting the received-signal power variation to the dynamic range of the A/D converters. Platform-attitude and navigation data are collected and recorded on high-density tape. The ground-based data processins and the results obtained with the radar are examined. I.E.
~~~

TYPE 1/4/37 Quest Accession Number : 91A15831 91A15831 NASA IAA Conference Paper Issue 04 Discrete target recognition in polarimetric SAR data 1AA)HEAL. . J. RUSSELL; (ABICUMMING, I A N G. (AB](British Columbia,.University,Vancouver, Canada) IN: Quantitative remote sensing: An economic tool for the Nineties; Proceedings of IGARSS '89 and Canadian Symposium on Remote Sensing, 12th, Vancouver, Canada, July 10-14, 1989. Volume 3 (A91-15476 0 4 - 4 3 ) . New York, Institute of Electrical and Electronics Engineers, 1989, p. 1836-1840. Research supported by NSERC, science council of British Columbia, and MacDonald Dettwiler and Associates, Ltd. 8 9 0 0 0 0 p. 5 refs 13 In: EN (English) P.498 Whether
n Y ~ + - ~ ~

TYPE 1/4/38 Ouest Accession Number : 91A15812 91A15812* NASA IAA conference Paper Issue 0 4 Squint mode SAR processing algorithms (AA)CHANG, C. Y.; (AB)JIN, M.; (AC)CURLANDER, 3 . C. (AC)(JPL, Pasadena, CA) Set Propulsion Lab., california Inst. of Tech., Pasadena. (JJ574450) IN: Quantitative remote sensing: An economic tool for the Nineties; Proceedings of IGARSS ' 8 9 and Canadian Symposium on Remote Sensing, 12th. Vancouver, Canada, July 10-14, 1989. Volume 3 (A91-15476 04-43). New York, Institute of Electrical and Electronics Engineers, 1989, p. 1702-1706. 890000 p. 5 In: EN (English) p.497
~~

the extra information content in polarimetric features will m the e difficult oroblems of taraet recosnition in SAR (synthetic

as

~~

are mainly the resilt-of Speckle and receiver noise adversely affecting the single channel of data. TO meet this obiective. a suoervised Scheme to rlaqaifv discrete _____.. polarimetric propertie; of scatterers-has been implemented. Target classes are defined by their polarization signature and available ground truth data. The authors examined a large number of polarimetric features and were able to determine a set that gives good classification performance. Comparing these results with trials performed on single-channel SAR data synthesized from the same data set clearly demonstrates a significant performance benefit of polarimetric radar. I.E.
~ ~~~

The unique characteristics of a spaceborne SAR (synthetic aperture radar) opefating in a squint mode include large range walk and large variation in the Doppler centroid as a function of range. A pointing control technique to reduce the Doppler drift and a new processing algorithm to accommodate large range walk are presented. Simulations of the new algorithm for Squint angles up to 20 deg and look angles up to 44 deg for the Earth observing system ( E O ? . ) L-band SAR configuration demonstrate that it is capable of maintaining the resolution broadening within 20 percent and the ISLR within a fraction Of a decibel of the theoretical value. I.E.

TYPE 1/4/39 Quest Accession Number : 91A15794 91A15794 NASA IAA Conference Paper Issue 0 4 Value added geocoded SAR products (AA)SCHREIER, G.; (AB)KNOEPFLE, W.: (AC)KOSMANN, D. (AC)(DLR, Oberpfaffenhofen, Federal RRpUbliC Of Germany) IN: Quantitative remote sensing: An economic tool for the Nineties; Proceedings of IGARSS '89 and Canadian symposium on Remote sensing, 12th, Vancouver, Canada, July 10-14, 1989. Volume 3 (A91-15476 04-43). New York, Institute of Electrical and Electronics Engineers, 1989, p . 1613-1616. 890000 p . 4 refs 11 In: EN (English) p.550 In the framework of the definition of higher-level SAR (synthetic aperture radar) products for the ERS-I and the SIR-C/X-SAR missions, DLR will deliver to the user geocoded data in several presentations. These data comprise ellipsoid-corrected and terrain-corrected information, as well as additional data Sets which will aid the thematic interpretation. 1n addition to these Standard products some investigations have been performed to demonstrate the feasibility and usability of value-added geocoded products, mainly derived from terrain-corrected, one-layer high-precision data. Among these proposed products are three-dimensional views in several presentations, such as perspective and anaglyph views, digital elevation model slope and incidence angle data, simulated SAR views, and vector data overlav. The basic aeneration alaorithms and some exampies, mainly performed on seasat data, ;re given. Th; current work on facilities, software, and databases to perform value-added product generation is explained, and an outlook for merging SAR geocoded data with GIs vector data is given. I.E.

TYPE 1/4/40 Quest Accession Number : 91A15781 91A15781 NASA IAA Conference Paper Issue 04 Absolute calibration of the CCRS C-band SAR during BEPERS-88 (AA)ULANDER, L. (AA)(Chalmers Tekniska Hogskola, Goteborg, Sweden) IN: Quantitative remote sensing: ~n economic tool for the Nineties; Proceedings of IGARSS '89 and Canadian symposium on Remote sensing, 12th. Vancouver, Canada, July 10-14, 1989. Volume 3 (A91-15476 04-43). New York, Institute of Electrical and Electronics Engineers, 1989, p. 1528-1531. Research supported by the Swedish Board of Space Activities and E S P . . 890000 p. 4 refs 7 In: EN (English) p.497

A Drocedure for Calibratins the real-time processed imase data from the CCRS &band synthetic apertur; radar ( S A R ) is-presented. Miasurements made during the Bothnian experiment in preparation for ERS-1 (BEPERS-88) of the returned sianal from radar reflectors. Of iniected noise to confirm STC
discussed. -The maximum toti1 error is estimated to he +/-3.0 dB for absolute calibration and +/-2.0 dB and +/-0.6 dB for relative calibration across and alona track. resoectivelv. The calibration Drocedure 1s illistrated by coiputing the backscatte; coefficient between io and 70 deg of incidence angle for data obtained over two sea-ice classes in the Gulf Of Bothnia.
I.E.

L I

TYPE 1/4/41 Quest Accession Number : 91A15667 91A15667* NASA IRA Conference Paner Issue 0 4 Recent advances in airborne terkestrial remote sensing with the NASA airborne visible/infrared imaging spectrometer (AVIRIS), airborne synthetic aperture radar (SAR), and thermal infrared multispectral scanner (TIMS) (AA)VANE, GREGG; (AB)EVANS, DIANE L.; (AC)KAHLE, ANNE 8. (AC)(JPL, Pasadena, CA) Jet Propulsion Lab., California Inst. of Tech., Pasadena. (53574450) IN: Quantitative remote sensing: An economic tool for the Nineties: Proceedings of IGARSS ' 8 9 and Canadian symposium on Remote Sensing, 12th. Vancouver, Canada, July 10-14, 1989. Volume 2 (A91-15476 04-43). New York, Institute of Electrical and Electronics Engineers, 1989, p. 942, 9 4 3 . 890000 p. 2 In: EN (English) p.479
~~~

TYPE Quest1/4/42 Accession Number : 91A15628 91A15628* NASA IAA Conference Paper Issue 04 Ice classification algorithm development and verification for the Alaska S A R Facility usina aircraft imaaerv (AA) HOLT, BENJAMIN ; ( A B ) KWOK,-RO~ALD; (AC) RIGNOT, ERIC (AC)(JPL, Pasadena, CA) Jet Propulsion Lab., California Inst. Of Tech., Pasadena. (JJ574450) IN: Quantitative remote sensinq: An economic tool for the Nineties: ~~~. Proceedings of IGaRSS '89 and Canadian Symposium On Remote Sensing, 12th. Vancouver, Canada, July 10-14, 1989. Volume 2 (A91-15476 0 4 - 4 3 ) . New York, Institute Electrical and Electronics Enaineers. 1989. n. 751-7'14. 8 9 0 0 0 0 p. of 4 In: EN (English) p.588
~ ~~

?j

..

Significant progress in terrestrial remote sensing from the air has been made with three NASA-developed sensors that collectively cover the solar-reflected, thermal infrared, and microwave reqions of the electromagnetic spectrum. These sensors are the airborne visible/infrared imaoina SDectrometer (AVIRISI. the thermal infrared manoina soectrometer (TIMS). ani the airboine synthetic aperture radar ( S k k ) , *respectively. AVIRIS and SAR underwent extensive in-flight engineering testing in 1987 and 1988 and are scheduled to become operational in 1989. TIMS has been in operation for several years. These sensors are described. I.E.

The Alaska SAR Facility (ASF) at the University of AlasKa, Fairbanks i s NASA program designed to receive, process, and archive SAR data from ERS-1 and to Support investigations that will use this regional data. AS part of ASF, Specialized subsystems and algorithms to produce certain geophysical products from the SaR data are under development. of particular interest are ice motion, ice classification, and ice concentration. This Work focuses on the algorithm under development for ice classification, and the verification of the algorithm using C-band aircraft SAR imagery recently acquired over the Alaskan arctic. I.E.
a

TYPE 1/4/43 Quest Accession Number : 91A15557 91A15557* NASA IAA Conference Paper Issue 0 4 Limex's7 ice surface characteristics and their effect upon C-band SAR sigantures (&&)DRINKWATER. MARK R. (AA)(JPL, Pasadena, CA) Jet Propulsion Lab., California Inst. Of Tech., Pasadena. (55574450) IN: Ouantitative remote sensina: An economic tool for the Nineties: Proceedings of IGARSS 89 and canaiian symposium on Remote- sensina, 12th.

TYPE 1 / 4 / 4 4 Quest Accession Number : 91A15529 91~15529* NASA IAA conference paper I S S U ~ 04 Results of the 1988 NASA/JPL airborne S A R calibration campaign (-)FREEMAN, A . ; IABIWERNER, C . : (ACIKLEIN, 3 . D. (AC)(JPL, Pasadena, CA) Jet Propulsion Lab., California Inst. of Tech., Pasadena. (JJ574450) IN: Quantitative remote sensinq: An economic tool for the Nineties; Proceedings of I G A R S S '89 and Canadian symoosium on Remote sensina. 12th. Vancouver, Canada, July 10-14. 1989. Volume-l (A91-15476 0 4 - 4 3 ) . N & York; Institute of Electrical and Electronics Engineers, 1989, p. 249-253. 890000 p. 5 refs 7 In: EN (English1 p.537
of 1988, the NASA/JPL multifrequency, During the spring multipolarization S A R (synthetic aperture radar1 flew in series of calibration experiments Over the Goldstone drv lake bed. An arrav o f calibration de;ices was deployed, including dihedpal and trihedral c&n& reflectors, polarimetric active radar calibrators, passive receivers, and CW (continuous wave) tone aenerators. The aim of the Camnainn w a s to ~ ~ .=-->-. -calibrate both amplitude and phase~ofthe ;esulting S a R images, over long and Short time scales. The results of the analysis of the calibration data collected in the spring of 1988 are presented. Trihedral corner reflector signatures and certain image background measures are used to externally calibrate relative ampl.iturle and phiss betwcsn pslarii6tions at a g i v e n frequency, and to calibrate across frequencies. Assessments are made of the calibration accuracy as a function of image frame position for each frequency, and the stability of the radar calibration over long and short time-scales. I.E.
~~~ ~~~~~~ ~~ ~~~~~

Ice Surface characterization data were recorded during March 1987 in the Labrador Sea marginal ice zone, at the Onset of Spring melt. Measured data are used as input parameters in a simple Scattering model to simulate the effects of temporal variations in material properties upon C-band scattering signatures. Snow moisture and Surface roughness have a significant effect upon sigma(0)HH and large differences are predicted between undeformed floe surfaces and deformed or rubbled ice areas. The model reproduces a calibrated synthetic-aperture-radar (SARI-derived signature Obtained during the experiment with a reasonable degree of certainty. Predictions also simulate a trend observed in S a R images of I Inrm.ci"" __ _..~ h*-b---b*"L --..---V I I L ~ brtwrrn a ~ ~ deformed and undeformed ice over period of surface warming. I.E.
I

I I I . I \ _ U . F L

TYPE 1/4/45 Quest Accession Number : 911115527 91A15527 NASA IAA Conference Paper Issue 04 Calibration for airborne SAR IAAIHAWKINS, R. K.: (AE)LUKOWSKI, T. I.; (AC)GRAY, A. L.; (A6)LIVINGSTONE. C. E. (AD)(Canada Centre for Remote Sensing, Ottawa) IN: Quantitative remote sensing: ~n economic tool for the Nineties; Proceedings of IGARSS '89 and Canadian symposium on Remote Sensing, 12th. vancouver, Canada, July 10-14, 1989. Volume 1 (A91-15476 04-43). New York. Institute Of Electrical and Electronics Engineers, 1989, p. 238-242. 890000 p. 5 refs 28 In: EN (English) p.537 Relative and absolute radiometric calibration of the CCRS (Canada Center for Remote sensing) airborne SARS (synthetic aperture radars) 1s It is noted that airborne SAR has unique calibratiol discussed. difficulties due its large swath/height ratio when compared with the - manifests itself in stronqlv varyinq terms in satellite-borne case. This imolications for image inteGpretation . the radar equation, with imoortant and analysis. Calibration for airhoke SARs may therefore lead to data products dissimilar from their satellite counterparts. Recommendations for . _ . usino the CCRS facility are presented, data acquisition and calibration along with CCRS olans for c ontinuing calibration Strategy,-which~emphasize .~~~ calibration on data utility. the implications.~if~ I.E.
~~~ ~~ ~~ ~~ ~~~~~~~ ~~~ ~

TYPE 1/4/46 Quest Accession Number : 91A15525 91A15525* NASA IAA Conference Paper ISSUe 0 4 The need for S A R calibration (AA)FREEMANJ a. (AA)(JPL, Pasadena, CA) Jet Propulsion Lab., California Inst. of Tech., Pasadena. (55574450) IN: Quantitative remote sensing: An economic tool for the Nineties; Proceedings of IGARSS '89 and Canadian symposium on Remote sensing, 12th. Vancouver, Canada, July 10-14, 1989. Volume 1 (A91-15476 04-43). New York, institute of Electrical and Electronics Engineers, 1989, p. 230-233. 890000 p. 4 refs 17 In: EN (English) p.537
~~~~ ~~~~

TYPE 1/4/47 Quest Accession Number : 91815451 91N15451# NASA STAR Conference Paper Issue 07 Antenna subarrays and GaAs T/R modules for X-SAR on EOS (AAIERUNNER, A.; (AB)LANGER. E. siemens A.G.. Munich (Germany. F.R.). (SKO32Ol2) Radio and Radar systems/Semiconductors. In DLR. Symposium on Applications of M u l t i f r e q u e n c y / M u l t i p o l a r i z a t i o n SAR in View of X-EOS (X-SAR for EOS) CGS p 259-271 (SEE N91-15434 07-32) 900500 p. 13 original contains color illustrations In: EN (English) Avail: NTIS HC/MF A12; DLR. VB-PL-DO. Postfach 90 60 58, 5000 Cologne, p.993 Fed. Republic of Germany, HC 105 Deutscbe marks An alternative antenna concept to the printed patch principle is proposed for the x-SAR (Synthetic Aperture Radar) quad polarization missmn. The active phased array system consists of a doubly polarized WaVeCtuide arrav fed v i a active antenna modules. The high slotted pilirization Durity required is achieved by Orthogonal slots in the broad side of two integrated groups of rectangular waveguides. A waveguide section of 10 (20) slots can be used as a subarray for one active module. That means 64 (32) subarrays along the 16 m Of the flight direction and 18 along the 0.38 m in range. The 1152 (576) subarrays or 1152 (576) active modules fill out the 6 sq m aperture. A number of 1152 active modules corresponds with 4.3 w output RF-power per module, for 5 kW peak power is to be radiated. ~ecauseof a better efficiency of a 4.3 w module compared to a 8.6 moaule out of 576 active modules for 576 subarrays, the 4.3 W output power module is recommended. Light weight Structures for module boxes and suhstrates are discussed. Principles of module components like Low Noise AmDlifier ILNA) . . or Sinqle-Pole-Double Through (SPDT) Switches are explained and shown. ESA
I~~~~~~

TYPE 1/4/48 Quest Accession Number : 91N15446 91N15446ff NASA STAR Conference Paper Issue 07 SAR interferometry can give 1 meter altitude resolution 1AAIPRATI. C.: IABIROCCA, F . : IACIMONTIGURNIERI. A. i A< (Consiglio Nazionale delle Riierche, Florence, Italy ) POI tecnico di Milano (Italy), (PX565076) Dipt. di Elettronica. I" ~ L R , Symposium on AppliCations of MultifreqUency/Multipolarization SAR n View of X-EOS (X-SAR for EOS) CGS p 197-212 (SEE N91-15434 07-32) 900500 p. 16 In: EN (English) Avail: NTIS HC/MF A12: DLR, VE-PL-DO, Postfach 90 60 58, 5000 Cologne, Fed. Republic Of Germany, HC 105 DeutSche marks p.992 In Synthetic Aperture Radar ( S A R ) interferometry, the altimetric information of the terrain can be obtained from the phase difference of two focused complex images. These images can be gathered by the Same sensor in two passes along different orbits or by two sensors mounted on the Same platform. The altimetric resolution of such a system improves when the Sensors displacement is increased in the cross track direction. The maximum allowed displacement, limited by Speckle noise, increases with the spatial resolution of the SAR image. Excluding the additive noise, it is shown that the achievable vertical resolution is better that the slant range resolution. As an example, an altimetric map of the Panamint Valley area is calculated using repeated passes of the Seasat satellite. The effect of the additive noise is visible only when the cross track distantof the two orbits is low. ESA
W

TYPE 1/4/49 Quest Accession Number : 91N15441 91815441% NASA STAR Conference P a ~ e rIssue 07 Multiparametric radar data for land. applications (AAIMAUSER. W.: J. ~. (ABINITHACK. , iABi (Deutsbhe Forschun~sanstalt fuer Luft- und Raumfahrt. Oberpfaffenhofen, Germany, F.R. ) Freiburg univ. (Germany, F.R.I. (F68175101 Dept. Of Hydrology. In DLR, Symposium on Applications of MUltifreqUency/MUltipolarization SAR in View of X-EOS (X-SAR for EOS) CGS D 115-138 (SEE N91-15434 07-321 900500 p. 24 original contains color illustrations I n : EN (Englishj Avail: NTIS HC/MF A12: DLR. VB-PL-DO, Postfach 90 60 58, 5000 Cologne, Fed. Republic of Germany, HC 105 DeutSChe marks p.992
~~ ~~~ ~

TYPE 1/4/50 Quest Accession Number : 91N15440 91N15440# NASA STAR Conference Paper Issue 07 Applicability of radar data for geoscientific purposes and expected improvements by use of multifrequency and multipolarization data (AA)JASKOLIA, F.; (ABIBODECHTEL, J. Technische UniV., Munich (Germany, F.R.I. (TJ492950) Inst. for General and Applied Geoloqy. In ~ D L R , sympoiium on Applications of M u l t i f r e q u e n c y / M u l t i p o l a r i z a t i o n SAR in View Of X-EOS (X-SAR for EOS) CGS p 95-114 (SEE "31-15434 07-32) 900500 p. 20 original contains color illustrations I": EN (English) Avail: NTIS HC/MF A12; DLR, VB-PL-DO, Postfach 90 60 5 8 , 5000 Cologne, Fed. Republic of Germany, HC 105 Deutsche marks p.991 Application of Synthetic aDertUre radar data in the field of n . ~. n ~~~-c r i o n r ~ is discussed. High potential especially for mapping purposes in geology is demonstrated. The differentiation possibilities of lithological units are recognized as poor due to small variations of dielectric properties of rocks and the poor availabilitv of multiDarameter radar data. wavc, in..-, - which the interpretation of strictural features, could be significantly improved are outlined. Delineation of surface morphology, delineation of Structural features surface and subsurface phenomena and determination of Surface moisture content are areas in which feature interoretation can be improved. These subjects include manifold parameters which are indispensable for an Optimized inclusion of radar data to different fields of interest. The most significant and promising applications besides classical applications (e.g., lithological and structural mapping) are identified. ESA
~~ ~ ~~~~

The influence of different sensor parameters and different surface conditions On the backscattering of the microwaves are Studied. Campaigns Conducted within the member countries both of ESA and the European community are described. The AGRISAR' 86 campaign, the first European campaign to study the temporal change of backscattered signal of agricultural units in the X-band using an imaging SAR (Synthetic Aperture Radar) System, the AGRISCATT '87 and '88 camDaiqns, multiband scatterometer-campaigns, flown to study the influenie of i variety of recording parameters on the backscattering signal are described. All three campaigns are flown over a test site west of the citv of Freiburo in the upper m i n e valley. ESA

__

TYPE 1/4/51 Quest Accession Number : 91N15439 91N15439# NASA STAR Conference PaDer Issue 07

In DLR, symposium - o n ApplicationS of~Multifrequency/Multipolarization SAR in View Of X-EOS (X-SAR for EOSI CGS p 77-94 ( S E E N91-15434 07-32) 900500 D . 18 In: EN (Enalishl Avail: NTIS HClMF Al?: DLR. , VR-PL-PL. - .Postfach 9 0 60 58, 5 0 0 0 CoioqGe, F;d.~ReDublic-if Germanv. HC 105 Deutsch; marks p.991
I
~~~~ ~~ ~~

TYPE 1/4/52 Quest Accession Number : 91N15438 91N15438# NASA STAR Conference Paper IsSue 07 Application of multifrequency and multipolarization SAR Systems in remote sensing for agriculture (AAIKUEHBAUCH, W. Bonn univ. (Germany. F.R.). (BT209639) In DLR, Symposium on Applications of M u l t i f r e g u e n c y / M u l t i p o l a r i z a t i o n SAR in view of X-EOS (X-SAR for EOS) CGS p 58-76 (SEE "3-15434 07-32) 900500 p. 19 original contains color illustrations 1n: EN (English) Avail: NTIS HC/MF Al2i DLR, VB-PL-DO, POStfaCh 90 60 58. 5000 Cologne, Fed. Republic of Germany. HC 105 Deutsche marks p.991 The deficiencies of remote sensing as applied in agriculture are discussed. They are due to the fact that optioal sensors depend on day light and on cloudfree coverage and. even under suitable weather conditions. miss important characters of the plant canopy that are related to crop species and yield. The all weather capability of radar especially in humid regions Offers considerable advantages in agricultural application. SAR (Synthetic Aperture Radar) potential for species recognition and yield prediction of agricultural craps is reviewed both as related to muiGicemporal single frequency and Single temporal multifrequency-multipolarization observations. The potential of variable SAR Systems and of combined optical and microwave sensors is discussed. ESA

Although various forms Of remote sensing have been used extensively for several decades, it is only recently that, through the development of new instruments and techniques, the versatility and potential of microwave remote sensing for the qualitative and quantitative measurements of hydrological processes have shown consistent improvements. Particularly, Multifrequency (MF), Multipolarization (MP), Multi-incidence Angles (MA) SAR ISynthetic Aperture Radar) Systems provided for EOS (Earth Observation System) polar platforms will open new applicative horizons in the study of hyrlmlngisal prscesses over large ais-= in e. vide range oi rnvironmentai conditions. Decisive contributions to applicative perspectives and system requirements have come out Of international experimental activities carried out in the last decade, and from multidisciplinary evaluation of the collected data. SAR campaigns Planned for the next few years, and further evaluation of SAR images to match quantitative estimations required for hydrological process-based modeling are discussed. ESA

TYPE 1/4/53 Quest Ackession Number : 91A14841 91A14841* NASA IAA Journal Article ISSUe 03 Incorporation Of polarimetric radar images into multisensor data sets (AAIEVANS. DIANE L . : (AB)VAN ZYL, SAKOB J.: (AC)BURNETTE, CHARLES F . iACj (J~~,'Pasadena, CA) set Propulsion Lab., California Inst. of Tech., Pasadena. (55574450) IEEE Transactions on Geoscience and Remote sensing (ISSN 0196-28921, v01. 28, Sept. 1990, p. 932-939. 900900 p. 8 refs 12 In: EN (English) p.390
A technique is presented for registering polarimetric SAR data to other aircraft and spaceborne data Sets. Resampling is done on the full Stokes matrix, allowing full polarization synthesis on the coregistered data set. Analysis Of data acquired over Pisgah Crater in the Mojave Desert, CA. as part of the Mojave Field Experiment shown that the resampling does not seriously affect the pedestal heights of polarization signatures or affect estimates of RMS heights for Smooth to moderately rough surfaces. I.E.

TYPE 1/4/54 Quest Accession Number : 91A14839 91A14839 NASA IAA Journal Article Issue 03 Determination of antenna elevation pattern for airborne S A R using the rough target approach (AA)HAWKINS, ROBERT K. (AA)(Canada Centre for Remote Sensing, Ottawa) IEEE Transactions on Geoscience and Remote sensing (ISSN 0196-2892). vol. 28, Sept. 1990, p. 896-905. 900900 p. lo refs 23 In: EN (English) P.358 Data from a forested region of northern Ontario are analyzed to yield an estimate of the antenna elevation pattern for the Canada Center for Remote sensing airborne SAR. The method relies on the imaged area being uniform on an average basis hut not necessarily uniform on a pixel-scale basis. The extended uniform area was imaged as a series of short flight segments in which the antenna depression angle was systematically stepped, all other acquisition parameters of the aircraft and SAR keeping essentially fixed. Subsequent analysis of the real-time imagery Was then performed, dividing average image powers for discrete bands of pixels across the swath to yield the relative gain of the antenna corresponding to the antenna angles for the Center of these bands. Combining the total set of these measurements generates the entire elevation pattern. Results are given for the C-band, HH-pattern over en angular range of 50 deg and dynamic range of over 30 dB and compared to antenna range measurements taken before installation. It is concluded that this method has good potential for airborne SARS with dynamic antenna pointing capability. I.E. 'TYPE 1/4/56 Quest Accession Number : 91N14642 91N14642* NASA STAR Issue 06 Method for detecting Surface motions and mapping Small terrestrial or / Patent planetary surface deformations with synthetic aperture radar (AA)GABRIEL, ANDREW K.: (AB)GULDSTEIN, RICHARD M.: (AC)ZEBKER, HOWARD A. (AA)inVentor (to NASA): (AB)inventor (to NASA): (AC)inventor (to NASA) (AC)(Jet Propulsion Lab., california Inst. of Tech., Pasadena.) National Aeronautics and Space Administration. Pasadena Office, CA. ( NU8946941 NASA-CASE-NPO-17831-1-cu: US-PATENT-4,975,704; US-PATENT-APPL-SN-470665; US-PATENT-CLASS-342-25; INT-PATENT-CLASS-GOlS-13/90 901204 p. 9 Filed p. 26 Sa". 1990 In: EN (English) Avail: US Patent and Trademark Offlce 840 A technique based on synthetic aperture radar (SAR) interferometry is used to measure very small (1 cm or less) surface deformations with good resolution (lo m) over large areas (50 km). It can be used for accurate measurements of many geophysical phenomena, including swelling and buckling in fault zones, residual, vertical and lateral displacements from seismic events, and prevolcanic swelling. Two SAR images are made of a scene by two Spaced antennas and a difference interferogram of the scene is made. After unwrapping phases of pixels of the difference interferogram, surface motion or deformation changes of the surface are observed. A second interferogram of the same Scene 1s made from a different pair of images, at least one of which is made after Some elapsed time. The Second interferoqram is then compared with the first interferogram to detect changes in line of sight position of pixels. By resoiving line Of sight observations into their Vector components in other -eto f intarferoarams alonil at least one other direction, lateral motions _ . . . .. . . . may be recovered in their entirety. since in general, the SAR images are made from flight tracks that are separated, it is not possible to distinguish surface changes from the parallax caused by topography. WOWF.VCT. R ~~, - third imaae mav be used to remove the topography and leave only the surface changes. Official Gazette of the U.S. Patent and Trademark Office
~ ~~~~ ~ ~~~~~ ~~ ~

TYPE 1/4/55 Quest Accession Number : 91A14838 91A14838 NASA IAA Journal Article Issue 03 SAR observations and modelina of the C-band backscatter variability due to multiscale geometry and soil*moisture (AA)BEAUDOIN, A,: (AB)GWYN, Q. H. J.: (AC)LE TOAN, THUY (AB)(Sherbrooke, Universite, Canada): (AC)(centre d'Etude Spatiale des Ravonnements. To~louse.France1 ?EEE Tran;actionS in Geoscience and Remote sensing (ISSN 0196-2892), vol. 2 8 , Sept. 1990, p. 886-895. Research supported by NSERC. 900900 p. 10 refs 15 In: EN (English) p.377 The effect of the multiscale surface geometry on the sensitivity of C-band ynthetic aperture radar (SAR) data to soil moisture is studied. The experimental data consist of C-band SAR images of an agricultural site. The backscatter variability due to Surface roughness has been analyzed. The effect of random roughness associated with soil clods is never less than 2 dB, and the effect of a row pattern can be as strong as 10 dB. In addition, the periodic drainage topography induces a backscatter variability due to soil moisture variation and drainage relief. Parameters accounting- for multiscale geometry have been modeled and used in theoretical surface scattering models. Good agreement was found between theoretical r e s u l t s and exoerimental data for the backscatter anuular ;&iation due to random roughness and row structure. The validated models have been used to extend the observations beyond the available range Of incidence and azimuth angles provided by airborne SARs. The results indicate clearlv that airborne C-band SAR data cannot be eaSilV inverted into soil moisture data. H O W ~ V ~with ~ , ERS-1 or Radarsat bata at an incidence angle of about 20 deg, the effect of random and periodic roughness can be reduced to about 2 dB if the look angle is leSS than 50 dea. I.E.
~~~~~ ~

TYPE 1/4/57 Quest Accession Ncmber : 91N14638 91N14638*# NASA STAR Technical Report Issue 06 Multiband radar characterization of forest biomes / Final Report, 1 Mar. 1985 28 Feb. 1990 (AA)DOBSON. M. CRAIG; (AB)ULABY. FAWWAZ T. Michigan univ., Ann Arbor. (MX270710) Radiation Lab. NASA-CR-185101; NAS 1.26:185101: UM-022486-1-F NAGW-733 900200 p. 12 In: EN (English) Avail: NTIS HC/MF A 0 3 p.839

TYPE 1/4/58 Quest Accession Number : 91N13595 91N13595*# NASA STAR Issue 0 5 Pipeline Synthetic aperture radar data compression utilizing systolic Patent binary tree-searched architecture for vector quantization ADalication .. (AA)CHANG, CHI-YUNG; (AB)FANG, WAI-CHI; (AC)CURLANDER, JOHN C. (AA)inventor (to NASA); (AB)inventor (to NASA): (AC)inventor (to NASA) (AC)(Jet Propulsion Lab., California Inst. of Tech., Pasadena.) National Aeronautics and Space administration. Pasadena Office, CA. (
Mr'R"dF."d,

The utility of airborne and orbital SAR in classification, assessment, and monitoring of forest biomes is investigated through analysis of orbital Synthetic aperature radar (SAR) and multifrequency and

____,

NASA-CASE-NPO-17941-1-CU; NAS 1.71:NPO-17941-l-CU; US-PATENT-APPL-SN-550775 NAS7-918 900710 p. 37 In: EN (English) Avail: NTIS HC/MF A03 p.648

A system for data compression utilizing Systolic array architecture for Vector Quantization (VQ) is disclosed for both full-searched and tree-searched. For a tree-searched VQ. the Special case of a Binary Tree-Search VQ (BTSVQ) is disclosed with identical Processing Elements (PE) in the array for both a Raw-Codebook VQ (RCVQ) and a Difference-Codebook VQ (DCVQ) algorithm. A fault tolerant System is disclosed which allows a PE that has developed a fault to be bypassed in the array and replaced by a spare at the end of the array. with codebook memory assignment shifted one PE past the faulty PE of the array. NASA

TYPE 1/4/59 Quest Accession Number : 91N13594 911313594*# NASA STAR Issue 0 5 Method for providing a polarization filter for processing synthetic aperture radar imaqe data / Patent ADvlication

TYPE 1/4/60 Quest Accession Number : 901151385 90A51385* NASA IAA Journal Article Issue 23 svnthetic-aDerture-radar imaaina of the Ocean surface usina the slightly-rough facet model and a.fuii surface-wave spectrum (AAIWEST, JAMES C.; (AB)MOORE, RICHARD K.: (AC)HOLTZMAN, JULlAN C. (AC11Universitv Of Kansas Center for Research. I n c . ., 1.awrenral , Kansas Univ. cinter 0; Research, I n c . , Lawrence. (KF728369) N00014-79-C-0533: NAGW-1278 International Journal of Remote Sensing (ISSN 0143-1161). "01. 11, Aug. 1990. p . 1451-1480. 900800 p. 3 0 refs 48 In: EN (English) p.3771
~~~~~ ~~~~ ~~~ ~~~~~~~~~~~

~~~~~

National ND894694) p.648

Ae;onautics

and Space Administration. Paiadena office, CA.

NASA-CASE-NPO-17904-l-CU; NAS 1.71:NPO-17904-1-CU; US-PATENT-APPL-SN-544293 NAS7-918 900625 p. 27 In: EN (English) Avail: NTIS HC/MF A03

A polarization filter can maximize the signal-to-noise ratio of a polarimetric synthetic aperture radar ( S A R ) and help discriminate between targets or enhance image features, e.g., enhance contrast between different types of target. The method disclosed is based on the Stokes matrix/ Stokes vector representation, so the targets of interest can be extended targets, and the method can also be amlied to the case of histatic polarimetric radars. NASA

A new model of synthetic-aperture-radar (SAR) imaging of ocean waves is described. The model is based on mapping individual, slightly-rough surface facets through the SAR processor into the image and responses of the facets in the image domain are added together coherently to give the composite image. A windowing technique allows both the orbital motion and LIE phase veiocity of m e long waves to be included. It is determined that the azimuthal cut-Off is due to a smearing of the response of the facets in the image induced by the random orbital motion of the intermediate large-scale waves and that the focus adjustment that gives the greatest image contrast is half the phase velocity of the dominant long wave. The optimal processing technique, however, may consist of spatially Offsetting the multiple looks on the image domain to ComDenSate the orovaoation of long waves-during time of the s k . L.K.S.
L..

TYPE 1/4/61 Quest accession Number : 90~50723 90~50723 Nasa IAA Journal article I S S U ~23 Synthetic aperture radar imaging Of ship wakes in the Gulf of Alaska (AA)SHEMDIN, OMAR H. (=)(Ocean Research and Engineering, Pasadena, CAI Journal of Geophysical Research (ISSN 0148-0227), vol. 95, Sept. 15, 1990, p. 16319-16338. Research supported by the U.S. Navy. 900915 p. 20 refs 11 In: EN (English) p.3704 The Gulf of alaska SAR experiment was conducted during March 9-14, 1984 to investigate SAR imaging of narrow-V wakes in a deep mixed layer environment so that surface manifestations of ship-generated internal waves would be made small. Five SAR flights were executed over large ships of opportunity in deep water where the mixed layer depth exceeded 100 m in all cases. SAR images were obtained in sea States 1-4. Range and azimuthally traveling ships were imaged. The incidence angles of azimuthally traveling ships ranged from 24 to 53 deg. The following results are reported: (1) the half angles associated with narrow-V wakes are consistent with first-order Bragg surface wave theory described by Case et a l . (1984); (2) the decay rate along the bright arms of the narrow-V wake is consistent with the combined viscous and radiation decay of short surface waves with first-order Bragg wavelengths; ( 3 ) narrow-V wakes are observed in sea states 1-3 at incidence angles less than 45 deg, with the longest narrow-V wakes bright-arm observed being 3 . 3 km; ( 4 ) turbulent wakes (dark band between the bright arms) are observed in sea States 1-4 at incidence angles less than 53 deg, with the longest turbulent wake length observed being 41 km; and ( 5 ) bright boundaries along one side of the turbulent wake are observed. author

TYPE 1/4/62 Quest accession Number : 90A50720 90~50720 Nasa IAA J O U ~ W article ~ ~ s s u e23 cross sections and modulation transfer functions at L and KU bands measured during the Tower Ocean Wave and Radar Dependence Experiment (AA)KELLER, WILLIAM C.; (AB)PLANT, WILLIAM 3. IAAIIU.S. Navy, Naval Research Laboratory, Washington, OC); (AB)(WoodS Hole oceanographic Institution, MA) Journal of Geophysical Research (ISSN 0148-0227). VOl. 95, SePt. 15, 1990, p. 16277-16289. Research supported by the U.S. Navy. 900915 P. 13 refs 26 In: EN (English) p.3771 Normalized radar cross sections and modulation transfer functions (MTFs) for microwave backscattering from the sea Surface have been measured at both L- and Ku-bands during the Tower Ocean Wave and Radar Dependence Experiment. Long waves during the experiment were usually not generated by the local wind, so a unique opportunity was afforded to investigate the effects of arbitrary wind, wave, and antenna angles on the backscatter. Cross sections at L-band are shown to be isotropic with respect to wind-antenna angle and nearly independent of wind speed except at the lowest wind speeds. Ku-band cross sections, on the other hand, show the expected wind-antenna angle anisotropy and wind speed dependence. The Ku-band cross sections agree well in magnitude with previous wave-tank and satellite measurements, casting doubt on the dependence of cross section on antenna height which has been suggested in the literature. at both frequencies the data suggest that cross sections may be lowered slightly when long waves propagate at large angles to the wind. author

TYPE 1/4/63 Quest accession Number : 90~50487 90~50487 Nasa IAA ~ournalarticle ISSU~23 Incidence-anale deoendence in forested and non-forested areas in Seasat s a i data (AA)RAUSTE, YRJO (AA)(Technical Research Centre of Finland, Espoo) lCEC and JPL. International Forest Siqnature Workshop, Ispra, Italy, Sept. 7-9, 1988)'International Journal of Remote Sensing (ISSN 0143-1161), v01. 11, July 1990, p. 1267-1276. Research supported by the Technical Research Centre of Finland, academy of Finland, and Neste, Ltd. 900700 p. 10 refs 19 In: EN (English) p.3754 Quantitative evaluation of the topography-induced variation in a Seasat scene is presented. Variations in the slope of the backscatter curve (sensitivity of radar to alterations in incidence angle) between spruce-dominated mixed forests, pine-dominated mixed forests, deciduous forests, and regenerated (pine plantations) areas are defined. The contribution of the corner reflector backscatter mechanism to the total backscatter is analyzed. analytical results Show that 65 percent of the total variation in land pixels can be attributed to terrain topography. R.E.P.
SAR

TYPE 1/4/64 Quest accession Number : 9 0 ~ 5 0 4 7 9 9oa50479* NASA m a Journal article Issue 23 The effect of changing environmental conditions on microwave signatures of forest ecosystems - Preliminary results of the March 1988 Alaskan aircraft SaR experiment 1aaIWaY. J O B E A : IABIPARIS, JACK; (AC)K?iSISCHKE,ERIC; (AD)SLAUGHTER, CHGLES; (AE)VIERECK; LESLIE (AB)(JPL, Pasadena, ca); (aC)(Michigan, Environmental Research Tn a. t_ ik ut a of Northern FOreStrv, _ . _ . . ,. Ann Arbor): . . Fairbanks, A K ) , . iAEIlInstitute ,~ , , Jet Propulsion Lab., California Inst. of Tech., Pasadena. (55574450) (CEC and JPL, International Forest Signature Workshop, Ispra, Italy, sept. 7-9, 1988) International Journal of Remote Sensing (ISSN 0143-1161). vol. 11, July 1990, p. 1119-1144. 900700 p. 26 refs 29 In: EN (English) p.3753
~~~~~~

In preparation for the ESA ERS-1 mission, a series of multitemporal, multifrequency, multipolarization aircraft SAR data sets were acquired near Fairbanks in March 1988. P-, L-, and C-band data were acquired with the NaSa/JPL Airborne SAR on five different days over a period of two weeks. The airborne data were augmented with intensive ground calibration data as well as detailed simultaneous in situ measurements of the geometric, dielectric, and moisture properties of the snow and forest canopy. During the time period over which the SaR data were collected, the environmental conditions changed significantly: temperatures ranged from unseasonably warm ( 1 to 9 C) to well below freezing ( - 8 to -15 C), and the moisture content of the snow and trees changed from a liquid to a frozen state. The SAR data clearly indicate the radar return is sensitive to these changing environmental factors, and preliminary analysis of the L-band SAR data shows a 0.4 to 5.8 dB increase (depending on polarization and canopy type) in the radar cross section of the forest Stands under the warm conditions relative to the cold. These SAR observations are consistent with predictions from a thebretical scatterins model. author

TYPE 1/4/65 Quest Accession Number : 90A49700 90A49700" NASA IAA Journal Article Issue 22 Assessment of tropical forest stand characteristics with multipolarization SAR data acquired over a mountainous region in Costa RiCa ~~(AAIWU, SHIH-TSENG (AA)(NASA, John C. Stennis Space Center, Bay Saint Imuis, MS) National Aeronautics and Space Administration. John c. stennis space Saint Louis, MS. (ND103456) Center, ~ a y (IEEE, Canadian Remote Sensing Society, URSI, et al., Quantitative remote sensing: An economic tool for the Nineties - 1989 International Geoscience and Remote Sensing Symposium and Canadian Symposium on Remote sensing, 12th. (IGARSS'891, Vancouver, Canada, July 10-14, 1989) IEEE Transactions on Geoscience and Remote Sensing (ISSN 0196-28921, "01. 28, July 1990, p. 752-755. 900700 p. 4 refs 8 In: EN (Enqlish) p.3571
~

TYPE 1/4/66 Quest Accession Number : 90A49694 90A49694 NASA I A A Journal Article Issue 2 2 Textural filtering for SAR image processing (AA)WANG, LI; (AB)HE, DONG-CHEN; (ACIFABBRI, ANDREA (AB)(Sherbrooke, Universite, Canada): (AC)(International Institute for Aerospace Survey and Earth Sciences, Enschede, Netherlands) (IEEE, Canadian Remote sensing Society, URSI, et ai., Quantitative remote sensing: An economic tool for the Nineties 1989 International Geoscience and Remote sensing symposium and Canadian symposium on Remote Sensing. 12th. (IGARSS'891, Vancouver, Canada. July 10-14, 1989) TREE Transactions on Geoscience and Remote Sensing (ISSN 0196-2892). "01. 28, July 1990, p. 735-737. 900700 p. 3 refs 12 In: EN (English) p.3571

P 2

A digital, terrain elevation data set was coregistered with radar data for assessing tropical forest stand characteristics. Both raw and topographically corrected L-band polarimetric radar data acquired over the tropical forests Of Costa Rica were analyzed and correlated with field-collected tree parameter data to study the stand characteristics. The results of analyses using 18 OUt Of 81 plots for sites A and B indicated that per-plot bole volume and tree volume are related to SAR data, particularly at Site A . The topographically corrected SAR data appear to produce the Same findings as those Of uncorrected data. I.E.

Examples of the application of a new approach to the textural filtering and enhancing of digital images are presented. Satisfactory results are obtained in processing images from both natural textures and airborne SAR scenes. Textural filtering Of SAR images can be useful in improving the discrimination between lithologic units with different surface-roughness characteristics. one application example is discussed in which textural features show different discrimination performances before and after textural filtering. I.E.

TYPE 1/4/65 Quest Accession Number : 90A49681

TYPE 1/4/68 Quest Accession Number : 90A49680 90A496808 NASA IAA Journal Article Issue 2 2

(AB)(stanforb university; CAI Stanford Univ., CA. (50380476) NAGW-419 (IEEE, Canadian Remote Sensing Society, URSI, et al., Quantitative remote sensing: An economic tool for the Nineties 1989 International Geoscience and Remote Sensing Symposium and Canadian Symposium on Remote Sensing, 12th, (IGARSS'89). Vancouver. Canada, July 10-14, 1989) IEEE Transactions on Geoscience and Remote Sensing (ISSN 0196-2892), vol. 28! July 1990. p. 669-673. Research supported by the Center for Aeronautics and Space Information Sciences and U.S. Navy. 900700 p. 5 refs 11 In: EN (English) p.3602

~~

~~~~~

I ,

jIEEE, Canadian Remote Sensing Society, URSI, et al., Quantitative remote sensing: An economic tool for the Nineties - 1989 International Geoscience and Remote Sensing Symposium and Canadian Symposium on Remote Sensing. 12th. (IGARSS'89), Vancouver. Canada, July 10-14, 1989) IEEE Transactions on Geoscience and Remote Sensing (ISSN 0196-2892). vol. 2 8 , July 1990, p. 662-668. 900700 p. 7 ref5 12 In: EN (English) p.3570 An algorithm for the global detection of coastlines based on a sequence of basic image-processing procedures and a new edge-tracing algorithm is described. The application of the proposed procedure to Seasat SAR and Shuttle Imaging Radar B images demonstrates that with only a modest computational burden it produces a good visual match between the detected coastline and the coastline of the original image. Additionally, the separation of land from water achieved by this algorithm permits clean pseudocoloring of coastal area images. I.E.

~~~~~~

--,

A method for finding cuzves in digital images with speckle noise is described. The solution method differs from standard linear convolutions followed ay thresholds in that it explicitly allows curvature in the features. Maximum a posteriori (MAP) estimation is used, together with statistical models for the speckle noise and for the curve-generation process, to find the most probable estimate of the feature, given the image data. The estimation process is first described in general terms. of the specific neighborhond system a i i d B Then, incorporation multiplicative noise model for speckle sllows derivation of the solution, using dynamic programming, of the estimation problem. The detection of curvilinear features is considered separately. The detection results allow the determination of the minimal size of detectable feature. Finally, the estimation of linear features, followed by a detection sten. is shown for computer-simulated images and for a S A R image of sea ice. I.E.

TYPE 1/4/69 Quest accession Number : 90A49677 90a.49677 NasA IaA Journal Article Issue 22 Interoolation and sampling in SAR images

~~~

y 1990, p. 641-646.
SAR

900700

p. 6

refs

In: EN (English)

The implications of interpolation and resampling for the statistics of images are analyzed. Interpolation of the complex data conserves the statistical distribution and all moments if a condition involving the autocorrelation functions of the SAR and the interpolating filter is met; in the ideal case (uncorrelated samples) this reduces to the requirement that the interpolating filter has Unit energy. Interpolation of the intensity data does not conserve the distribution. Expressions for moments up to the fourth for an arbitrary interpolation scheme applied to correlated exponential data are derived, and conditions for conservation of these moments displayed. any finite resampling or interpolation scheme (other than nearest neighbor) will change the autocorrelation functions of the complex and intensity data and may introduce nonstationarity. I.E.

TYPE 1/4/70 Quest Accession Number : 90A49675 90A49675 NASA IAA Journal Article Issue 22 a new method of aircraft motion error extraction from radar raw data for real-time motion compensation (AA)MOREIRA, JOAO R. (AA)(DLR, lnstitut fuer Hochfrequenztechnik, Oberpfaffenhofen, Federal Republic of Germany) (IEEE, Canadian Remote sensing society, URSI, et al., Quantitative remote sensing: An economic tool for the Nineties - 1989 International Geoscience and Remote Sensing symposium and Canadian Symposium on Remote Sensing, 12th, (IGARSS'~~), Vancouver, Canada, July 10-14, 1989) IEEE Transactions on Geoscience and Remote Sensing (ISSN 0196-2892). "01. 28, July 1990, p. 620-626. 900700 p. 7 refs I2 In: EN (English) p.3491

a method for real-time motion compensation is presented. The method extracts all the necessary motions of the aircraft from the radar backscatter signal using a new radar configuration and new methods for evaluating the azimuth spectra of the radar signal. Hence an inertial naviaation ="stem becomes unnecessary for many applications. The motion-compenskion parameters for real-time motion error correction are the range delay, the range-dependent phaseshift, and the pulse repetition frequency. The motions of the aircraft to be extracted are the displacement in line-of-sight direction, the aircraft's yaw and drift angle, and the forward velocity. Results show that a three-look image with i/n alimlith rpsolution of 3 m in L-band usinq a small aircraft ic . . . . _ achievable, and the implementation of this method in real time using a n array processor is feasible. I.E.
I__
~~~~~

TYPE 1/4/11 Quest Accession Number : 90A49673 90A49673 NASA IAA Journal Article ISSUe 22 Performance of a proposed SDaceborne synthetic aperture radar with variable antenna height (AA)TOMIYASU, KIYO (AA)(GE Valley Forge space Center. Philadelphia, PA) (IEEE, Canadian Remote Sensing Society, URSI. et al., Quantitative remote sensing: an economic tool for the Nineties - 1989 International Geoscience and Remote Sensing Symposium and Canadian Symposium on Remote Sensing, 12th. (IGARSS'89), Vancouver, Canada. July 10-14, 1 9 8 9 ) IEEE Transactions on Geoscience and Remote Sensing (ISSN 0196-2892). vol. 28, July 1 9 9 0 . p. 609-613. 900700 p. 5 refs 5 In: EN (English) p.3509 Broadside mapping with high area rates over a wide range of grazing angles is possible with a variable antenna area for a spaceborne SAR. The rectangular antenna length is fixed by the azimuth resolution. and the antenna area must exceed a minimum value to avoid ambiguous responses in both range and Doppler. The minimum antenna height is established by the ambiguity constraint for each mapping geometry, and then perhaps increased in height to meet an SAR power limit. With a variable-height antenna. mapping area rates up to 2142 sq km/s were computed for a 5.3-GHz SAR in an 800-km-altitude orbit. Hardware design issues are addressed, and an active-element phased array with discrete heights is suggested. I.E.

TYPE 1/4/72 Quest Accession Number : 90A45352 90A45352 NASA IAA Journal Article T S S U e 20 A state-of-the-art review in radar polarimetry and its applications in remote sensing (AR)BOERNER, WOLFGANG-M.; (AB)YAMAGUCHI, YOSHIO (AB)(Illinois, University, Chicago) (1989 International Symposium on Noise and Clutter Rejection in Radars and Imaging Sensors, Kyoto, Japan, NoV. 14-16, 1989) IEEE Aerospace and Electronic Systems M a g ~ - 2 (la>," " o o > - " "=I E T ,nnn ,_ azine o*J o > J , __^, ""I. 171", y . 6. 900600 p. 4 refs 8 In : EN (English) P.3195
" " " Z

I, Y Y l l F

The authors assess the state of the art, focusing on their awn contributions. Covered areas are the electromagnetic inverse problem in radar polarimetry, coherent polarization r a d a r theory, partially coherent radar theory, vector (polarization) inverse scattering polarization approaches, the polarimetric matched filter approach, polarimetric Doppler radar applications in meteorology and oceanography, and image fidelity in microwave vector diffraction tomographic imaging. I.E.

TYPE 1/14/73 Quest Accession Number : 90A44319 90A44319 NASA I A A Journal Article Issue 20 The role of spaceborne synthetic aperture radar in global wave forecasting (AA)BEAL, ROBERT C. (AA)(Johns Hopkins University, Laurel, MD) Johns Hopkins Univ., Laurel, MD. (55767253) Johns Hopkins APL Technical Digest (ISSN 0270-5214). vol. 11, Jan.-june 1990, p. 54-62. Research supported by Johns Hopkins University, N O M , ESA, NASA, and U.S. Navy. 900600 p. 9 refs 14 In: EN (English) p.3234 The importance of improved global directional wave forecasts is outlined, and the contingency of these forecasts upon improvements in global wind fields, model physics, and global directional wave measurements and related assimilation schemes is Stressed. Information for global numerical models must come from satellite platforms and synthetic aperture radar ( S A R ) is the only radar technique potentially useful for remote ocean wave monitoring which has actually flown in space. The geometry and functions of a spaceborne SAR are discussed and research highlights of the past decade are reviewed. Future plans for global ocean spectral sampling to be conducted by the NASA SIR-C and Subsequent experiments in the areas of ecology, geology, hydrology, oceanography, radar calibration, and electromagnetic scattering theory are outlined. The development of an experimental on-hoard SAR processor to produce nearly continuous real-time Ocean Wave spectra from the C-band channel is described. L.X.S.

TYPE 1/4/74 Quest Accession Number : 90A41671 90A41671 NASA I A A Journal Article Issue 18 Synthetic aperture radar calibration using reference reflectors (AA)GRAY, A. LAURENCE; (AB)VACHON, PARIS W.; (AC)LIVlNGSTONE, CHARLES E. ; (AD)LUKOWSKI, TOM 1. (AD)(Canada Centre for Remote Sensing, Ottawa) IEEE Transactions on Geoscience and Remote Sensing (ISSN 0196-2892). VOl. 28, May 1990, p. 374-383. 900500 p. IO refs 23 In: EN (English) p.2879
A simple expression for the terrain backscatter coefficient is derived in terms of the integrated power of an adjacent known radar reflector i n an S A R image. It is shown that this S A R image-calibration technique is independent Of the radar system focus Or partial coherence and thereby possesses an important advantage over the usual technique, Which relies on an estimate of the peak of the reflector impulse response. Results from airborne SAR overflights of comer reflectors and active radar calibrators are used to demonstrate the validity and consistency of the method and to show that the method is robust under defocus caused by an incorrect FM rate or inadequate motion compensation Of data collected during turbulence. It 1s also shown that the fading errors associated with the integral method are comparable to or slightly worse than those associated with the peak estimation method. However, this small disadvantage is outweighed by the fact that the integral method is independent of actual resolution. I.E.

TYPE 1/4/75 Quest Accession Number ; 90A41054 90~41054 NASA IAA conference paper issue 1 8 Improved filters for moving target indication with Synthetic aperture radar (AA)MEDLIN, GREGORY W.; (AB)ADAMS, JOHN W. (AA)(South Carolina, University, Columbia); (AB)(Hughes Aircraft co., Radar systems Group, LOS Angeles, CA) IN: International conference on Radar 8 9 , Paris, France, Apr. 24-28, 1989, Proceedings. Volume 2 (A90-40951 18-32). Boulogne-Billancourt, France, RADAR 89, 1989, p. 392-397. 890000 p. 6 refs 6 In: EN (English) p.2882 A digital filter design technique is proposed for the pre-filter moving target indication (MTI) method for synthetic aperture radar ( S A R ) systems. The filters have exactly linear phase and are based on a multiple Stopband design where the integrated aliasing error is minimized and the filter passband is maximally flat. The proposed filters show significant improvement in clutter cancellation and overlap between MTI bands when compared to previous designs. A design example is included Which demonstrates the effectiveness of the technique. Author

TYPE 1/4/76 Quest Accession Number : 90A41044 9 0 ~ 4 1 0 4 4 NASA IAA conference paper I S S U ~18 Autofocusing synthetic aperture radar images (AA)PRATI, CLAUD10 (AA)(Milano, Politecnico, Milan, Italy) IN: International Conference on Radar 89, Paris, France, Apr. 24-28, 1989, Proceedings. Volume 2 (A90-40951 18-32). BOUlo~ne-Billancou~t, France, RADAR 89, 1989, P. 314-319. Research Supported by the Stanford Exploration Project. 890000 p. 6 refs 8 In: EN (English) p.2875 The correct focusing of synthetic aperture radar (SAR) relies on knowing precisely the geometric and transmission parameters of the system. Transmission parameters are usually defined specifically, while the geometrical parameters can he derived from the Satellite data. The only two geometric parameters involved in the focusing process are the sensor-target relative Velocity and the sensor-target nearest approach distance. A technique to extract these two geometrical parameters from the data themselves (autofocusing), in order to achieve the best possible focusing of SAR images, is discussed. R.E.P.

TYPE 1/4/77 Quest ic&ssion Number : 90A41043 90A41043 NASA IAA Conference Paper Issue 18 Processina of soaceborne and airborne S A R data - An eXDerimenta1 activity (AA)VINELLI, F.; (AB)MORABITO, C.; (AC)TOMA, M. R.; (AD)D'ADDIO, E.; (AE)FARINA, ALFONSO (AE)(Selenia S.p.A., Rome, Italy) IN: International Conference on Radar 89, Paris, France, Apr. 24-28, 1989, Proceedings. Volume 2 (A90-40951 18-32). Boulogne-Billancourt, France, RADAR 89, 1989, p. 301-308. 890000 p. 8 refs 11 In: EN (English) p.2875
~~~~ ~~~

TYPE 1/4/78 Quest Accession Number : 90A41042 90A41042 NASA IAA Conference Paper Issue 18 30 synthetic aDeZtUre radar Surveys CLA~JDIO (AA)(Milano, Politecnico, Milan, Italy) IN: International Conference on Radar 89, Paris, France, Apr. 24-28, 1989, Proceedings. Volume 2 (A90-40951 18-32). BOuloqne-BillancOUTt, France, RADAR 89, 1989, p. 295-300. Research supported by the Stanford Exploration Project. 890000 p. 6 refs 5 In: EN (English) p.2875

(AA)PFATI,

test-bed for processing recorded and simulated SAR data is studied. SAR processor, which synthesizes images from spaceborne and airborne sAR sensors, is based on the range-Doppler algorithm. The process used to simulate S A R data is described. The SAR processor is evaluated using recorded and simulated data. The automation feature extraction algorithms based on the Hough transform technique are examined. I.F.

A The

It is shown that by utilizing multiple passes of a SAR system, terrain elevation can be measured. For greater altimetric resolution, the satellite's displacement in the cross-track direction is adapted to the slope of the terrain. A coarse altimetric map of the Panamint Valley area in Eastern California is constructed by utilizing three repeated passes of the Seasat SAR satellite. The technique avoids the foreshortening effect which IS clearly visible on hilly areas. The interference fringe images presented show an acceptably high signal-to-noise ratio. It is pointed out that the same analysis may be applied to tethered-satellite SAR systems, thereby avoiding the effects of the changing of the observed areas with respect to time. S.A.V.

TYPE 1/4/79 Quest Accession Number : 90A41000 90A41000 NASA IAA Conference PaDer Issue 18 Parallel DFT algorithms for radar Signal processing (AA)SORAGHAN, JOHN J.; (AB)GREEN, RICHARD C. (AA)(Strathclyde, University, Glasgow, Scotland); (AB)(Active Memory Technoloav. Readinu, Enaland) IN: Internationai Conference on Radar 89, Paris, France, Apr. 24-28, 1989, Proceedings. Volume 1 (A90-40951 18-32). BOUlogne-BillancOU~t, France, RADAR 89, 1989, p. 323-327. 890000 p. 5 refs 7 In: EN (English) p.2925 OFT implementation on a distributed array of processors is studied. A mapping strategy is presented and its complexity analyzed. Synthetic Aperture Radar ( S A R ) azimuth compression is Chosen as a radar problem that demands significant computational capabilities. It is shown that the AMT DAP-510 not Only matches this computational requirement but also that it efficiently copes with the range migration associated with SAR. Example processing times are aiven for simulated SEASAT and ERS-1 SAR data. Author

TYPE 1/4/80 Quest Accession Number : 90A39952 90A39952 NASA IAA Conference Paper ISSUe 17 Target cluster detection in cluttered synthetic aperture radar imagery (AA)I.?+NDOWSKI, JAMES G.; (AB)LOE, RICHARD S. (AB)(Lockheed Research and Development Laboratories, Pal0 Alto, CA) IN: Advances in image compression and automatic target recognition; Proceedings of the Meeting, Orlando, FL, Mar. 30, 31, 1989 (A90-39951 17-63). Bellingham, WA. society of Photo-Optical Instrumentation Engineers, 1989, p. 9-16. 890000 p . 8 refs 9 In: EN (English) p.2766 A technique for detecting clusters of objects in noisy, cluttered, moderate resolution imagery is discussed. The algorithm is demonstrated on synthetic aperture radar (SARI data. The approach is based on the use of a nonlinear spatial highpass or 'antimedian' filter, the complement of the median filter. The filter is coarsley tuned to produce maximum response for structures the size of or smaller than the expected object size. The filter is followed bv histouram thresholdinu and connected reuion processing. Knowledge about the-object's shape aid the cluster deployment patterns is then used to eliminate false detections. This detection technique is suitable for any imagery where the objects Of interest produce sensor responses that form contiguous regions. False clusters due to edge leakage are discussed and a solution formulated Author

?
N

TYPE 1/4/81 Quest Accession Number : 90A38345 90A38345 NASA IAA Journal Article Issue 16 The slightly-rough facet model in radar imaging of the ocean surface (AA)WEST, JAMES C.; (AB)MOORE, RICHARD K.; (AC)HOLTZMAN, JULIAN c. iACIlUniverSitv of Kansas Center for Research. Inc.. 1.awrenceI ~~, N000;4-79-C-05;3 Inteinational Journal 'of Remote Sensing (ISSN 0143-1161). vol. 11, April 1990, p. 617-637. 900400 p . 21 refs 49 In: $2.2591 EN (English)
~ ~~~~

~~~

~~

TYPE 1/4/82 Quest Accession Number : 90A38343 90A38343* NASA IAA Journal Article Issue 16 Global digital topography mapping with a synthetic aperture scanning radar altimeter (AA)ELACHI, C.; (AB)IM, K . E.; (AC)RODRIGUEZ, E. (ACIIJPL. . ~,Pasadena. CAI ~~~, Jet Propulsion Lab., California Inst. of Tech., Pasadena. (JJ574450) International Journal of Remote Sensing (ISSN 0143-1161). VOl. 11, April 1990, p. 585-601. 900400 p . 17 refs I1 In: EN (English) p.2580

IY

,.

The slightly-rough facet model of the ocean surface, an extension of the two-scale radar scattering model, is well suited for investigating SAR imaging of the surface. Several statistical properties of the facets that are important in an imaging model are derived. The two-scale scattering model is extended to include both fixst-order and second-order large-scale effects (tilt and curvature) using physical optics, showing that a spectruln ot small-scale ripples, rather than a single ripple given by the Bragg resonance condition, contributes to the backscatter from a facet. The bandwidth of the resonant ripple spectrum depends on the radar wavelength, large-scale Curvature and illumination widths. The resonant ripple spectra Of adjacent facets overlap, so the backscatter from adjacent facets is correlated. The backscatter from individual facets temporally decorrelates due to dispersion of the ripples in the resonant spectrum. Depending on the conditions, the decorrelation time may be on the order of the integration times of SAR processors. Author

Global digital topography data of the land surface is of importance in a variety of geoscientific and application disciplines. Such a. database, with a spatial resolution of 150 to 500 m and height accuracy of 5 m or better can be acquired from an orbiting platform using a synthetic aperture scanning radar altimeter. Near-global coverage can be achieved within 14 days from an orbiting platform in a polar or near-polar Orbit. Author

TYPE 1/4/63 Quest Accession Number : 90A37750 90A37750 NASA IAA Journal Article Issue 16 Motion compensation of airborne synthetic aperture radars using autofocus (A?.)BLACXNELL, D.; (AB)QUEGAN, S. (aa)(General Electric Company, PLC, Marconi Research Centre, Chelmsford, England); (AB)(Sheffield, University, England) GEC J o u r n a l of Research (ISSN 0264-9187). "01. 7, no. 3, 1990, p. 168-182. Research supported by the Ministry of Defence Procurement Executive. 900000 p . 15 refs 18 In: EN (English) p.2530 conventional motion compensation schemes correct for unwanted SAR platform motions using information from an inertial measurement unit (IMU). Autofocus techniques, which focus SAR images, produce an 'autofocus parameter' which is related to the platform motion. In this paper, Strong evidence is presented to support the assumption that the contrast optimization autofocus algorithm behaves as a least-squares quadratic fitting to the SAR platform trajectory. Using this assumption, the relationship between the autofocus parameter and across-track accelerations of the S A R Dlatform is derived. This allows the SAR Dlatform motion to be estimated- from the autofocus parameter measurements and incorporated in a motion compensation scheme, instead of IMU measurements. Three implementations of motion compensation using autofocus are compared, and the achievable image quality is quantified. Author

TYPE 1/4/84 Quest Accession Number : 90A34922 90A34922# NASA I A A Journal Article I S S U ~ 14 Radio and optical remote sensing: Satellite and ground observations. 11 SIR-c/x-SAR experimental plan (AA)FUJITA, MASAHARU; (AB)MASUKO. HARUNOBU; (AC)OKAMOTO, KEN'ICHI: (AD)URATSUKA, SEIHO communications Research Laboratory, Review (ISSN 0914-9279). "01. 35, Dec. 1989, p. 515-521. In Japanese, with abstract in English. 891200 p. 7 refs 14 In: JA (Japanese) p.2141

In June 1986, NASA/DFVLR/PSN issued an announcement of opportunity for using the SIR-C/X-SAR, which will be launched in early 1992 on the space Shuttle. I" response. two experimental proposals were submitted and were accepted in 1988. The titles of the proposals are 'Like- and cross-polarization calibration, topographic mapping and rice field experiments by SIR-C/X-SAR' and 'Remote sensing of sea pollution and sea ice by SIR-C/X-SAR'. This paper gives a brief explanation of hoti of the experiments together with an Outline of the SIR-C/X-SAR hardware. Author

0
TYPE 1/4/85 Quest accession Number : 90~34147 90A34147 NASA I A A Conference PaWer Issue 14 SvnthetiC aDertUre radar - A Kelvin wake imaqe artifact (~A)HARGER ,-ROBERT 0 . (AA)(Maryland, University, College Park) IN: Millimeter wave and svnthetic aDerture radar: Proceedinas Of the Meeting, Orlando, FL, Mar. 27, 5 8 , 1989 (A90-34126 14-32). Bellingham, WA, society of Photo-optical Instrumentation Engineers, 1989, p. 225-234. Research supported by the Environmental Research Institute Of Michigan. 890000 p. 10 refs 10 In: EN (English) p.2212
~ ~~

TYPE 1/4/86 Quest Accession Number : 90~34145 90A34145* NASA IAA Conference PaWer Paper ISSUe Issue 14 SAR ground data system The Earth Observing system ( E O S ) SAi (AA)CURLANDER, JOHN C. (AA)(JPL, Pasadena, Ca) Jet Propulsion Lab., California Inst. of Tech., Pasadena. (JJ574450) IN: Millimeter wave and synthetic aperture radar; Proceedings of the 27, 28, 1989 (ago-34126 14-32). Bellingham, wa, Meeting, oriando, FL, ~ a r . Society of Photo-Optical Instrumentation Engineers, 1989, p. 210-220. 890000 p. 11 refs 7 In: EN (English) p.2143 NASA, in association with ESA and NASDA, will launch the Space Station Freedom in 1993. A5 a complement to the Space Station, several unmanned Polar-Orbit Platforms (Pops) will be developed, built and launched with suites of instruments devoted to remote-sensing for earth surface and atmosphere observations or to planetary and deep-space Studies. Attention is presently given to the POPS-associated Earth Observing System SAR Ground Data System, which encompasses a SAR prOCeS50rr a postprocessing geophysical processor, L)rocessor. and a data management and control subsystem, a qeoDhhysical . . subsystem.
O.C.

A possible physical basis is presented for some of the narrow-vee artifacts occurring in SAR images of ship wakes. It is suggested that this pattern will emerge when the ship and SAR platform directions are nearly parallel; when these directions are less well aligned, one arm of the narrow-vee pattern artifact will be missing. Then, when the directions have become sufficiently misaligned, no image artifact will be visible. Other curious artifacts in SAR images, Such as parallel lines, have been observed in SAR images but appear to require a different physical mechanism. O.C.

TYPE 1/4/87 Quest Accession Number : 90A33035 901133035 Article Issue 90~33035 NASA IAA Journal article I S S U ~13 Wind and Ocean swell measurements from space using microwaves Wind- und seegangsmessungen mit Mikrowellen vom Weltraum aus IAAIAT.PERS. (AA)hLPERS, WERNER (AA)(Bremen, Universitaet, Federal Republic Of Germany) ueber Atmosphaere, &eane, Ozeane, Kontinente, (Deutsche Meteorologen-Tagung ueberKiel, Federal Republic Of Germany, May 16-19, 1989) Annalen der Meteorolooie 26. 1989. Meteorologie (ISSN 0072-41271. 0072-4122). no. 26, 1989, w. p. 63-66. In German. 8 9 0 0 0 0 p. 4 refs 8 In: GM (German) p.2058 Satellite-based measurements of wind and ocean swell using active microwaves are reported. The scatterometer and the SAR used to make the measurements are described. Normalized radar cross sections are shown as a function of wind direction for various wind velocities.
C.D.

TYPE 1/4/88 Quest Accession Number : 904431297 90~31297# NASA IAA Journal Article Issue 12 Theoretical studies for ERS-1 wave mode (AA)CORDEY, R. A,; (AB)MACKLIN. J. T.; (AC)GUIGNARD, J.-P.; (AD)ORIOL-PIBERNAT, E. (AB)(GEC-Marconi Research Centre, Chelmsford, England); (AC)(ESTEC, Noordwijk, Netherlands); (AD)(ESA, European Space Research Institute, Frascati, Italy) ESA Journal (ISSN 0379-2285). Vol. 13, no. 4 , 1989, p. 343-362. Research supported by ESA. 8 9 0 0 0 0 p. 20 refs 30 In: EN (English) p.1886 The theory of how the sea surface is imaged by synthetic-aperture radar (ShR) and inverse methods for recovering quantitative information about ocean waves are considered, with emphasis on the Operation of the wave mode of ERS-1. The two-scale representation of radar backscattering from the sea surface and its limitations are discussed. The description of the sea surface is reviewed, covering short-wave spectra, wave/wave interactions (hydrodynamic modulations), and breaking waves. Linear and nonlinear descriptons Of ocean-wave imaging are ermained. A new inverse method using complex imagery is developed and is shown to offer improved prediction of the speckle component in Sm-image power spectra. Azimuthal smearing from random wave motions imposes a fundamental limitation on the recovery of wave information under nonlinear imaging conditions. Recommendations for ERS-1 wave mode, including the requirements for validating the imaging theory, are discussed. Author

? N
w

TYPE 1/4/09 Quest Accession Number : 90A30653 90A30653* NASA IAA Journal Article Issue 12 Phase calibration Of imaging radar polarimeter Stokes matrices (AAIZEBKER, HOWARD A.: (ABlLOU. YUNLING (ABIIJPL, Pasadena. CA) Jet Propulsion Lab., California Inst. of Tech., Pasadena. (JJ5744501 IEEE Transactions on Geoscience and Remote Sensing (ISSN 0196-2892), 01. 28, March 1990, p. 246-252. 900300 p. 7 refs 11 In: EN (English) p.1829 It is shown that the Stokes matrices measured by an imaging radar polarimeter provide enough information for the accurate phase calibration of the observed polarimetric characteristics Of a surface. This is important because it allows the data to be reduced in volume in an operational synthetic aperture radar correlator with no prior knowledge of the conditions at the surface, and the end user can later select the particular region where he or she is comfortable with making an assumption regarding the relative phases of the hh and vv signals. No ground calibration eauioment is necessarv. as all imoortant naramaters are

TYPE 1/4/90 Quest Accession Number : 9OA30648 90A30648 NASA IAA Journal Article Issue 12 SAR imaae statistics related to atmosaheric draa over sea ~~. i c ~ (AA)BURNS. BARBARA A. (AA)(Alfred-Wegener-I~t~tt fuer Polarund Meeresforschung, Bremerhaven. Federal Reoublic Of Germanvl N00014-8l1C-0295; NObO14-83-C-0404 IEEE Transactions on Geoscience and Remote Sensing (ISSN 0196-2892), vol. 28, March 1990. p. 158-165. 900300 p. 8 refs 31 In: EN (English) p.1884
~~

j :

The possibility Of using SAR data to distinguish sea-ice regions with different atmospheric drag is explored. Both the amplitude of the radar return and statistics derived from SAR image data are examined. Roughness statistics data from several pack-ice areas are used in a backscatter model to predict the return from surfaces with measured drag coefficients. The results Suggest that the scattering coefficient for typical radar wavelengths is insensitive to the roughness elements responsible for the observed drag coefficient variations over Dack ice free of maior ridaes. For margin& ice zones, where ice concentration and floe &forma& contribute to atmospheric drag, a simple model for the atmospheric boundary layer is used to provide qualitative relationships between drag coefficient and regional ice properties (ice concentration, floe size distribution, floe edge densityj derivable from SAR data. Simple algorithms to produce maps of ice concentration and edge density are outlined and applied to 23.5-cm SAR digital image data.
I.E.

TYPE 1/4/91 Quest Accession Number : 90A30503 90A30503* NASA IAA Journal Article TsSue 12 An automated system for mosaicking spaceborne SAR imagery (AA)KWOK, RONALD; (ABICURLANDER, JOHN C.; (AC)PANG, SHIRLEY S. (ACI(JPL, Pasadena, ca) Set Propulsion Lab., California Inst. of Tech., Pasadena. (JJ574450) International Journal of Remote Sensing (ISSN 0143-1161). 01. 11, Feb. 1 9 9 0 , p. 209-223. 900200 p. 15 refs 9 In: EN (English) p.1827
~~ ~~~ ~~~ ~~~

TYPE 1/4/92 Quest Accession Number : 90A26668 90a26668* Nasa IRA ~ o u r n a larticle issue 10 An ice-motion trackina SVStem at the Alaska SAR facilitv (AAIKWOK, RONALD; *(Ai)CURLANDER, JOHN C.; (ACIPANG, SHIRLEY S . ; (ADIMCCONNELL, ROSS (AC](JPL, Pasadena, CA); (AD)(Veucel Corp., Boulder, CO] Jet Propulsion Lab., California Inst. of Tech., Pasadena. (JJ574450) IEEE Journal Of Oceanic Engineering (ISSN 0364-9059). ~ 0 1 . 15, Jan. 1990, p. 44-54. 900100 p. 11 refs 18 In: EN (English) p.0 An operational system for extracting ice-motion information from synthetic aperture radar (SARI imagery is being developed a5 part of the Alaska SAR Facility. This geophysical processing System (GPS] will derive ice-motion information by automated analysis of image Sequences acquired by radars on the European ERS-1, Japanese ERS-1, and Canadian RADARSAT remote sensing satellites. The algorithm consists Of a novel combination of feature-based and area-based techniques for the tracking of ice floes that undergo translation and rotation between imaging passes. The system performs automatic selection of the image pairs for input to the matching routines using an ice-motion estimator. It is desiqned to have a daily throughput of ten image pairs. A description is given of the GPS system, including an overview of the ice-motion-tracking algorithm, the System architecture, and the ice-motion products that will be available for distribution to geophysical data users I.E.

An automated system has been developed for mosaicking Spaceborne synthetic aperture radar ( S A R ) imagery. The System is capable Of producing multiframe mosaics for large-scale mapping by combining images in both the along-track direction and adjacent cross-track swaths from ascending and descending passes. The system requires no operator interaction and is capable of achieving high registration accuracy. The output product is a geocoded mosaic on a standard map grid such as UTM or polar stereographic. The procedure described in detail in this paper consists essentially of remapping the individual image frames into these standard grids, frame-to-frame image registration and radiometric smoothing Of the seams. These procedures are directly applicable to both the Magellan Venus Mapper and a scanning SAR design such as Radarsat, EOS S A R in addition to merging image frames from traditional SAR systems Such as SEASAT and SIR-B. With minor modifications, it may also be applied to spaceborne optical sensor data to g-ncrata large-scale mosaics efficiently and with a high degree of accuracy. The System has been tested with SEASAT, SIR-B and Landsat TM data. Examples presented in this paper include a 38-frame mosaic Of the Yukon River basin in central Alaska, a 33-frame mosaic of southern California and a three-frame terrain-corrected geocoded mosaic Of the Wind River basin in Wyoming. AUthor

TYPE 1/4/93 Quest Accession Number : 901325096 90NZ5096# NASA STAR Technical ReDOrt Issue 19 Concepts for high resolution space- based SAR/ISAR systems (AA)BOESSWETTER, CLAUS; (AB)WOLFRAMM, ARIBERT PAUL; [ACIPIKE, TIMOTHY KEVIN; (AD)HERMER, JEAN MICHEL (AD)(Thomson-CSF, Montrouge, France ) Messerschmitt-Boelkow-Bl~hm G.m.b.H., Ottobrunn (Germany, F.R.). ( MT620643) Unqernehmensgruppe Raumfahrt. MBB-UK-0057f89-PUB In its, Research and Development: Technical and Scientific Pbbliqations 1 9 8 9 ,261-270 ~ (SEE N90-25075 19-01) 890000 p. 10 Presente at AGARD Avionics Panel symposium on High Resolution Air and Hague, Netherlands, 8-12 May 1989 In: EN (English) Spaceborne f a d , , Avail: NTIS,HC Different beam pointing techniques, Such as monobeam, multibeam in azimuth/elevation are addressed for strip mapping SAR modes as well as for MTIIISAR modes. Basic requirements for selection of orbits are also discussed. The MTI/ISAR modes is intended to detect and classify from space clusters of fast moving small targets against the clutter background of the earth surface. High along-track resolution is achieved by using an antenna as Short as possible in the along-track direction, collecting a large number of return echoes and processing these, according to the regular SAR principle. But the antenna technology and the realtime processing technology necessary to handle the large amount of data generated by multimode resolution systems create difficulties. ESA

TYPE 1/4/94 Quest Accession Number : 90N24489 90N24489# NASA STAR Conference Paper ISSUe 1 8 Analysis and test interaction in the development of a hold down and

In ESA, -Fourth European Space Mechanisms and Tribology Symposium p 181-186 (SEE N90-24462 18-31) 900300 p. 6 In: EN (English) Avail: NTIS HC A14/MF A02 p.2539
The features of the hold-down and release mechanism Used to hold the folded synthetic aperture radar antenna onboard the ESA Remote Sensing Satellite (ERS-1) during launch are described. S i x hinged clamps are released in orbit via a system of Springs and cables connected to a pyrotechnical device. Improvements in the mechanism introduced during the early design stage on the basis of analytical investigations and component tests are highlighted. Special emphasis is given to the interaction between analysis and test results. Results from subsystem tests are considered. ESA

TYPE 1/4/95 Quest Accession Number : 90N24257 90N24257*# NASA STAR Technical Report Issue 1 8 Synthetic aperture radar imagery of airports and surrounding areas: Denver Stapleton International Airport / Final Report, 31 Aug. 1987 - 30 NOV. 1989 (AA)ONSTOTT, ROBERT G.; (AB)GINERIS, DENISE J. Environmental Research Inst. of Michigan, Ann Arbor. (E0356283) NASA-CR-4305; NAS 1.26:4305; DOT/FAA/DS-89/16 NAS1-18465 505-67-41-57 Washington 900700 p. 385 In: EN (English) Avail: NTIS HC A17/MF A02 p.2495 This is the third in a series of three reports which address the statistical description of ground clutter at an airport and in the surrounding area. These data are being utilized in a program to detect microbursts. Synthetic aperture radar ( S A R ) data were collected at the Denver Stapleton Airport using a set of parameters which closely match those which are anticipated to be utilized by an aircraft on approach to an airport. These data and the results of the clutter study are described. Scenes of 1 3 x 10 km were imaged at 9.38 GHz and HH-, W-,and HV-polarizations, and contain airport grounds and facilities (up to 14 percent), cultural areas (more than 50 percent), and rural areas (up to 6 percent). Incidence angles range from 40 to 84 deg. At the largest deoression anoles the distributed taroets. such as forest. fields. water. ani residential, rarely had mean scatk-iig coefficients greater than -10 dB. From 30 to 80 percent of an image had scattering coefficients less than -20 dB. About 1 to 1 0 percent of the scattering Coefficients exceeded 0 dB, and from 0 to 1 percent above 10 dB. In examining the average backscatter coefficients at large angles, the Clutter types Cluster according to the following groups: (1) terminals ( - 3 dB), (2) city and industrial (-7 dB), ( 3 ) warehouse (- 10 dB), (4) urban and residential (-14 dB1. and (51 m a s s (-24 dB1. Author

TYPE 1 / 4 / 9 6 Quest Accession Number : 90A24232 90A24232 NASA IAA Bookflonograph Issue 09 Synthetic aperture radar Stations with digital processing

(Russian book

Radiolokatsionnve stantsii 5 tsifrovvm sintezirovaniem aDerturv antennv (AA)ANTIPOV, GLADIMIR N. i (AB)GORIAINOV, VLADIMIR - T . i (AC)KULIN, ANATOLII N.; (AD)MANSUROV, VLADIMIR V.; (AE)OKHONSKII, ALEKSANDR G. MOSCOW, Izdatel'stvo Radio i Sviaz', 1988, 304 p. In Russian. 880000 p. 304 refs 77 In: RU (Russian) p.1297 The principles underlying SAR with digital processing are examined. Particular attention i s given to structural and logical Schemes of basic radar svstems. asoects of movino target indication, the digital simulation of rad& operation, and tra5ecto;y-signal simulation. ?t is noted that digital signal processing makes possible the optimal unification of SAR units, the automated design and development of SAR systems, and a reduction in SAR weight and size. B.J.

TYPE 1/4/97 Quest Accession Number : 90A23504 90A23504* NASA IAA Journal Article Issue 0 8 A Signal processing view of strip-mapping Synthetic aperture radar (AA)MUNSON, DAVID c . , JR.; (AB)VISENTIN, ROBERT L. (AB)(Illinois, University, Urbana) Illinois Univ., Urbana. (18647432) DAAL03-86-K-0111; JPL-957927; N00014-844-0149 IEEE Transactions on Acoustics, Speech. and Signal Processing (ISSN 0096-3518), sol. 37, Dec. 1989, p. 2131-2147. 891200 p. 17 ref5 49 In: EN (English) p.1136 The authors derive the fundamental strip-mapping SAR (synthetic aperture radar) imaging equations from first principles. They Show that the resolution mechanism relies on the geometry of the imaging situation rather than on the Doppler effect. Both the airborne and Spaceborne cases are considered. Range processing is discussed by presenting an analysis of pulse compression and formulating a mathematical model of the radar return signal. This formulation i s used to obtain the airborne SAR model. The authors study the resolution mechanism and derive the sianal ~rocessino relations needed to produce a high-resolution image. <TheyA introduce spotlight-mode SAR and briefly indicate how polar-format spotlight processing can be used in strip-mapping SAR. They discuss a number of current and future research directions in SAR imaging. I.E.

TYPE 1/4/98 Quest Accession Number : 90A23347 90A23347* NASA IAA Journal Article Issue 08 Opening and closing of sea ice leads - Digital measurements from Svnthetic aDertUre radar -(AA)FILY..M.: (AB)ROTHROCK, D. A. lAA)(GrenOble, Universite, Saint-Ma=tin-d'HeTeS, France); (AB)(Washington. Universitv. _ . Seattle) Grer.oble-1 nniv., saint Martin d'iieres (France). (G5963195) NAGW-412 Journal of Geophysical Research (ISSN 0148-0227), "01. 95, 3 8 1 1 . 15, 1990, p. 789-796. Research Supported hy the U.S. Navy and ESA. 900115 p. 8 refs 16 In: EN (English) p.1191

An algorithm that uses two Sequential SAR digital images to measure the opening and closing of sea ice leads is introduced. The methods used to analyze the SAR images is described, including the mesh of tie points and the classification of leads and ice. The results of the anaylsis are compared with independent manual measurements, showing that the cells covering each lead are best interpreted as opening or closing in a group, rather than individually. Also, consideration is given to an automated algorithm for grouping cells, the possibility of simplifying the method, and the relationship between the opening and closing measurements and the theory of their parametric relation to mean deformation. R.B.

TYPE 1/4/99 .~~~ , . Quest Accession Number : 901322826 90N22826*# NASA STAR Technical Report ISSUe I~6 spaceborne radar observations: a guide for Magcillan radar-image analysis l?.AIFORD. J. P . : (AB)BLOM, R. G.: (AC)CRISP, J. A,; (ADIELACHI, CHARLES: ,THEILIG, E. E . : (AH)WALL, [aij F k , T. G. : (AF)SAUNDERS, R. STEPHEN: (AG) s. 6.: (AI)YEWELL, S . B. Jet Propulsion Lab., California Inst. of Tech., Pasadena. (33574450) NASA-CR-184998: JPL-PUBL-89-41: NAS 1.26:184998 NAS~-918 844-20-00-30-02 891200 p. 132 original contains color illur;trations In: EN (English) a . .&rail: NTIS HC AO7/MF AOl: 25 functional color pages p.2249 Geologic analyses of spaceborne radar images of Earth are reviewed and summarized with respect to detecting, mapping, and interpreting impact craters, volcanic landforms, eolian and subsurface features, and tectonic landforms. Interpretations are illustrated mostly with Seasat synthetic aperture radar and Shuttle-imaging-Tadar images. Analogies are drawn for the potential interpretation of radar images of Venus, with emphasis On the effects of variation in Magellan look angle with Venusian latitude. In each landform category, differences in feature perception and interpretive capability are related to variations in imaging geometry, Spatial resolution, and wavelength of the imaging radar systems. Impact Craters and other radially symmetrical features may Show apparent bilateral symmetry parallel to the illumination Vector at low look angles. The styles of eruption and the emplacement of major and minor volcanic constructs can be interpreted from morphological features observed in images. Radar responses that are governed by small-scale surface roughness may serve to distinguish flow types, but do not provide unambiguous information. Imaging of Sand dunes is rigorously constrained by specific angular relations between the illumination vector and the orientation and angle of repose of the dune faces, but is independent of radar wavelength. with a single look angle, conditions that enable shallow subsurface imaging to occur do not provide the information necessary to determine whether the radar has recorded surface or subsurface features. The topographic linearity of many tectonic landforms is enhanced on images at regional and local scales, but the detection of structural detail is a strong function of illumination direction. Nontopographic tectonic lineaments may appear in response to contrasts in small-surface roughness or dielectric Constant. The breakpoint for rough surfaces will vary by about 25 percent through the Magellan viewing geometries from low to high Venusian latitudes. Examples of anomalies and system artifacts that can affect image interpretation are described. Author

TYPE 1/4/100 Quest Accession Number : 90A21604 90a21604# NASA IAA Journal article Issue 07 High-performance S A R processors for mission planning and control (AA)SCHOTTER, ROLAND; (AB)FRITSCH, BRUNO Dornier Post (ISSN 0012-5563). no. 3 , 1989, p. 13-15. 890000 p. 3 EN (English) p.925
SAR

In:

Pipeline-aTchitectuTe image processors have been devised for real-time operation whose operational modules perform FFT, correlation, Complex arithmetic, lowpass/bandpass filtering, and two-dimensional memory functions. All modules have been implemented in power-saving MOS technology to facilitate the use of real-time SAR processors in such limited ~ o w e r dissioation conditions as those Of mobile qround stations. Attention' ~ i igiven -to the ERS-1 'Quicklook' processor for a mobile Antarctic ground station: the Do-SAR experimental airborne SAR system, and the FLEX-SAR processor that is being implemented for the Do-SAR program. O.C.
~~~ ~

TYPE 1/4/101 Quest Accession Number : 90N21246 90~21246$ Nasa STAR conference ~ a o e rIssue 14 SAR image coding CODAGE D'IMAGES SAR (AA)TOURTIER, P. Thomson-CSF. Paris iFrancel. iTY609526) In AGARO, 'High Reiolution A i ; - and Spaceborne Radar 6 14-32) 891000 p. 6 I": FR (French) avail: NTIS HC aio, Nationals requests available only from AGaRD/Scientif Executive p.1964 The Synthetic Aperture Radar imagery causes a Very large flow rate, to the extent that the data flow is at a record level. The image coding technique reduces the flow rate so that the original quality is preserved. This permits the reduction of the transmission channel capacity and improves the flow rate. A different technique is presented for data flow compression. The technique performs best at low cosine transform and is described in detail. The results obtained by ThomSOn-CSF show that a compression rate of the magnitude of 4 or 5 is possible without visible image degradation. Transl. hy E.R.

m
N
4

TYPE 1/4/102 Quest Accession Number : 90N21245 90N21245# NASA STAR Conference Paper Issue 14 Image simulation of geometric targets for synthetic aperture radar SIMULATION D'IMAGES DE CIBLES GEOMETRIQUES POUR RADAR A OUVERTURE SYNTHETIOUE (AAINASR, J. M . Aerospatiale, Les M U r e a U X (France). (AG943352) Dept. Mesures. In AGARD, High Resolution Air- and Spaceborne Radar 7 p (SEE N90-21223 14-32) 891000 p. 7 In: FR (French) Avail: NTIS HC AlO/MF AO2: Non-NATO Nationals requests available only from AGARD/Scientific Publications Executive p.1964 A new technique for image simulation Which comes from a synthetic aperture radar is presented. The method is based on the embedding of an artificially simulated target in a real radar image captured by an operational antenna window on a satellite (SEASAT Or SIR-B). A L and C band was used for the capture. The tarqet dimensions studied were l a m e enough for use with long-waves provided the calculation techniques used with high frequencies were for a n equivalent area radar (SER). The calculation of SER allows the capture Of a raw signal received from the antennas. So that the possibility of simulation is low, Some restrictions are made. The results are sufficiently interesting enough to let the study of the behavior of a particular target become of use to civilians or the military, in the functional bounds of radar waves. Transl. by E.R

TYPE 1/4/103 Quest Accession Number : 90N21244 90~21244# NASA STAR conference paper sue 14 Concept and results of the DLR realtime SRR processor (AA)MOREIRA. ALBERT0 DeUtsche Forschungs- und VerSuChSanStalt fuer Luft- und Raumfahrt, (D0699060) Inst. for Radio Frequency Oberpfaffenhofen (Germany. F.R.). Technology. In AGARD. High Resolution Air- and Spaceborne Radar 6 p (SEE N90-21223 14-32) 891000 p. 6 In: EN (English) Avail: NTIS HC AIO/MF A02: NOWNATO Nationals requests available only from AGARD/Scientific Publications Executive p.1963 A real time azimuth processor was developed for the airborne E-SAR System (Experimental Synthetic Aperture Radar). The processor works with an unfocused compression method. This method greatly simplifies the data processing and is easily implemented by a moving average approach. A SAR image processed by a traditional unfocused processing method has a lower contrast. higher sidelobes, and worse resolution than in the focused case. A new algorithm was developed. so that a triangular amplitude weighting could be implemented into the unfocused processing method Without additional complications. Images processed in real time are presented. They show good contrast and strong suppression of the sidelobes. The processor hardware can be implemented with reduced Costs in small aircraft and is suitable for several applications Such as the detection of oil pollution Over the sea. Author

TYPE 1/4/104 Quest Ac&ssion Number : 90N21242 90N21242# NASA STAR Conference Paper Issue 14 Radar taraet imaae bv ISAR case studv (AAIMARIN?, S.: iABI$ARDINI, S.: (AC'IPRODI. F. Selenia S.P.A., Rome (Italy). (SF.408204) 'Dept. Of Radar. In AGARD. High Resolution Air- and spaceborne Radar 14 p (SEE N90-21223 14-32] 891000 D. 14 In: EN iEnalishl Avail: NTTS HC AlO I M~ F an?: . . . . . ,... ~ NO~-NATO Nationais requests aviilable only from AGARo/Scientific Publications Executive p.1963
~~

~~~

~~

Target imaging based on Inverse Synthetic Aperture Radar techniques are described. Theoretical and experimental results are presented. Theoretical topics illustrate the set of processing functions needed to Obtain the target image Starting from the radar echoes. Key processing steps include motion compensation and reconstruction of the reflectivity function. ~n experimental setup based on a currently available tracking radar, a data recorder and off-line processing facilities are illustrated. A high cross range resolution image of a MB-339 aircraft was obtained by processing recorded radar echoes from a Selenia X band tracking radar. Author

TYPE 1/4/105 Quest Accession Number : 90N21239 90N21239# NASA STAR Conference Paper Issue 14 A fast ISAR-imaging process and its inherent degrading effects on image quality (AA~AETNKE, K.-H.: (ABIROEDE, B. Deutsche Forschunqs- und Versuchsanstalt fuer Luft- und Raumfehrt, Inst. fuer Oberpfaffenhofen (Germany, F.R.). (00699060) Hochfrequenztechnik. In AGARD, High Resolution Air- and Spaceborne Radar 12 p (SEE "lo-21223 14-32) 891000 p. 12 In: EN (English) Avail: NTIS HC AlO/MF AOZ: Non-NATO Nationals requests available only from AGARD/Scientific Publications Executive p.1963 A method for a fast 2-0 inverse synthetic aperture radar (ISAR) imaging process is presented. A coherent short pulse radar is used to Sample amplitude and phase of the backscattered field from a continuouslv rotating object: This is being done while a narrow range gate is sweeping in range steps of 15 cm across the target plane at a typical speed of 150 m/s. Applying fast SAR principles, in an off-line process for each range cell, an acceptable good Cross range resolution can be obtained when processing angle intervals of less than 30 deg. The influence of analytical approximations as well as the effect of moving scattering centers through several range resolution cells fillring the precess interx:el can cause severe imase deqradations. T WO methods for Dartial and comolete compensation of the& effects under the aspect bf minimum lois in processing speed were developed and are presented.

TYPE 1/4/106 Quest Accession Number : 90N21238 90N2123811 NASA STAR Conference Paper Issue 14 Problems in ISAR processing with range resolution by stepped frequency bursts (AA)KRAEMER, GERD Forschunosinstitut fUer Hochfrequenzphysik, Werthhoven (Germany, F.R.). ~. (F1944607)In AGARD. High Resolution Air- and Spaceborne Radar 5 p (SEE N90-21223 14-32) 891000 p. 5 In: EN (English) Avail: NTLS HC AlO/MF A O 2 ; NO~-NATO Nationals requests available only from aGARD/Scientific Publications Executive p.1962 If a target image is reconstructed from an ISAR (Inverse Synthetic aperture Radar) measurement by immediate application of the Discrete Fourier Transform, the image becomes blurred with increasing distance from its center. It i s shown that with an ISAR sensor applying stepped frequency bursts, samples of the 2-0 Fourier Transform Of a 2-0 scatterer density are measured and how a target image can be reconstructed. Author

TYPE 1/4/107 Quest Accession Number : 90N21236 90N21236# NASA STAR Conference Paper ISSUe 14 a solution for real time motion compensation for SAR without using inertial navigation systems (AA)MOREIRA, JOAO R. Deutsche Forschungs- und Versuchsanstalt fuer Luft- und Raumfahrt, wesseling (Germany, F.R.). (00705482) Inst. for Radio Frequency Technology. In AGARD, High Resolution Air- and Spaceborne Radar 6 p (SEE 30-21223 14-32) 891000 p. 6 In: EN (English) Avail: NTIS HC AlO/MF A02; Non-NATO Nationals requests available Only from AGARD/scientific Publications Executive p.1962 A new solution is given for real time motion compensation. The main idea is to extract all the necessary motions of the aircraft from the radar backscatter sianal usina a new radar confiauration and new methods for evaluating the- azimuth -spectra of the radar signal. Hence an inertial navigation system becomes unnecessary for many applications. The motion compensation parameters for real time motion error correction are the range delay, the range dependent phase shift and the pulse repetition freouencv. The motions of the aircraft to be extracted are the displacement in line of sight direction, the aircrafts yaw and drift angle and forward velocity. Results show that a three look image with an azimuth resolution of 3m in the L-band usina a small aircraft is achievable and t h e implementation of this method in real time using an array processor is feasible Author
~~ ~~~~~~~~~ ~ ~ ~ ~

TYPE 1/4/108 Quest Accession Number : 90N21235 90N21235# NASA STAR Conference Paper ISSUe 14 a motion compensation Study for the PHARUS project (AA)OTTEN, M. P. G. organisatie voor Toeyepast Natuurwetenschappelijk Onderzoek, The Hague (Netherlands). (01465661) Radar and Communications Div. In AGARD. High Resolution Air- and Spaceborne Radar 12 p ( S E E N90-21223 14-32) 891000 p. 12 In: EN (English) Avail: NTIS HC AlO/MF AO2; Non-NATO Nationals requests available Only from AGARD/Scientific Publications Executive p.1962 In the PHARUS project, a polarimetric C band SAR is being developed, which will be preceded by a nonpolarimetric test System called PHARS. a motion compensation study is also part of preparatory studies for the final PHARUS design. A SAR data simulator was developed as a tool for the study of the effects of aircraft motion on the SAR image. From the SAR mapping geometry, a terrain description, the radar parameters, and detailed trajectory and attitude data of a non-maneuvering aircraft, the simulator generates raw data with a given range resolution. This can be processed, by azimuth compression, into the SAR image. a secondary purpose of the simulation is to determine the impact of several design parameter choices, and to provide well defined test input for S A R processing software. The results of test runs with real flight data Were verified theoretically, and have Shown the need for motion compensation. It was also shown that a major advantage of simulation, in that it can take many factors into account at the Same time, including for instance the SAR processing method, which is hard to do theoretically. Author

TYPE 1/4/109 , , Quest Accession Number : 90N21234 90~21234# Nasa STAR conference paper Issue 14 Concept for a spaceborne Synthetic Aperture Radar (SAR) sensor based on active phased array technology (AA)BRUNNER, A.; (AB)LANGER, E.: (AC)OETTL, N.; (AD)ZELLER, K. N. (AD)(Deutsche Forschungs- und Versuchsanstalt fuer Luft- UDd Raumfahrt, Oberpfaffenhofen, Germany, F.R. ) F.R.). (SK032012) Siemens A.G.. Munich (Germany,
Telecommunication/semiconductors.

In AGARD, High Resolution Air- and spaceborne Radar 10 p (SEE N90-21223 14-32) 891000 p. 10 In: EN (English) Avail: NTIS HC AlO/MF AO2; Non-NATO Nationals requests available Only from AG?.RD/Scientific Publications Executive p.1962
F O ~ surveillance with spaceborne remote sensing systems, quite often a Spatial resolution of 1 m or less is requested. A SAR concept is presented for a low flying satellite. Assuming a peak power of 5 kW and using active phased array technology, a Swath width of about 30 km at an off nadir angle of 35 deg is considered to be reasonable. A wide swath width combined with a high resolution can only be achieved if a fixed antenna beam is used for transmitting which illuminates the whole swath width, while a very narrow antenna beam scans the Swath in the manner as the reflected pulse travels from the near range to the far range a c m s s the Swath width. For the active antenna system. a high efficiency of the transmit/receive modulus, low losses in the feeding network and doubly polarized radiating elements with high polarization purity are considered of Utmost importance. An antenna based on the slotted waveguide principle is described. The technology of the GaAs based modules with special respect to space requirement resulting in an economic solution of the power generation below 3 w per module is described. author

TYPE 1/4/110 Quest Accession Number : 90N21233 90N21233# NASA STAR Conference Paper Issue 1 4 Real-time adaptive radiometric correction for imaging radars systems (AA)MOREIRA, JOAO R.; (ABIPOETZSCH, WINFRIED Deutscne FOrSChUngS- und Versuchsanstalt fuer Luft- und Raumfahrt, Wesseling (Germany, F.R.). (00705482) Inst. for Radio Frequency Technology. In AGARD, High Resolution Air- and Spaceborne Radar 6 p (SEE N90-21223 14-32) 891000 p. 6 In: EN (English) Avail: NTIS HC AlO/MF AO2; Non-NATO Nationals requests available only from AGARD/Scientific Publications Executive p.1962

TYPE 1/4/111 Quest Accession Number : 90N21229 90N21229# NASA STAR Conference Paper Issue 1 4 ROSAR (Helicopter-Rotor based Synthetic Aperture Radar) (AA)KLAUSING, HELMUT; (ABIKALTSCHMIDT, HORST: (AC)KEYDEL, WOLFGANG (AC)(Deutsche Forschungs- und Versuchsanstalt fuer Luft- und Raumfahrt, Wesseling, Germany, F.R. ) Me55eT5Chmitt-Hael)iow-Blohm G.m.b.H., Ottobrunn (Germany, F . R . ) . [ MT6206431 In AGARD, High Resolution Air- and Spaceborne Radar 12 p (SEE N90-21223 14-32] 891000 p. 12 In: EN (English) Avail: NTIS HC AlO/MF A02; Non-NATO Nationals requests available only from AGARo/scientific Publications Executive ~.1961 ROSAR is a synthetic aperture radar concept based on rotating antennas of a helicopter for pilot sight target detection and target localization with hiqh resolution. The ROSAR COnCeDt has DOtential benefits for civil and military helicopterborne imaging- appli&tions, if the antennas are mounted at the tips of the rotor blades. The concept has two main potential benefits, the imaging field of view is 360 deg and there is no need for a forward velocity of the carrier platform. AS opposed to SAR systems based on linear movement of the antenna, ROSAR imaging is based on synthetic aperture of a circular shape. Thus, the image formation process requires a polar format processing architecture. The ROSAR principle is also applicable for other radar mapping systems with rotating antennas, not only for helicopters. Author

a new solution is given of a real time radiometric image correction that a l s o minimizes the quantization and saturation noise introduced by the process of analog-to-digital conversion of raw data of coherent and noncoherent imaging radar systems. The implementation of this procedure was successfully performed with the experimental SAR System (E-SAR) of the DLR . Author

TYPE 1/4/112 Quest accession Number : 90N21223 90N21223# NASA STAR Conference Proceedings Issue 14 High Resolution Air- and Spaceborne Radar Advisory Group for Aerospace Research and Development, Neuilly-sur-seine (France). (AD4554581 Avionics Panel. AGARD-CP-459; ISBN-92-835-0530-1 891000 p. 224 symposium held in The Hague, Netherlands, 8-12 May 1989 In ENGLISH and FRENCH In: AA (Mixed) Avail: NTIS HC AIO/MF A02i Non-NATO Nationals requests available only from AGARD/Scientific Publications Executive p.1960 Imaging techniques are important sources of information in military operations. They may serve for purposes such as target detection and location, reconnaissance, classification and identification of fixed or moving objects as well as for orientation over unknown terrain. Despite considerable advances in electro-optical imaging systems the radar sensor has become an attractive alternative for several reasons: large range performance, penetration Of weather, smoke, dust and foilage, day and night operation. On the other hand high resolution radar techniques such as synthetic aperture radar (SARI and inverse Synthetic aperture radar [ISAR) promise geometrical resolution Of about 1 m and less. For individual titles, see N90-21224 through N90-21246.

TYPE 1 / 4 / 1 1 3 Quest Accession Number : 90N18371 90N18371*# NASA STAR Technical Report IsSue 11 Synthetic aperture radar imagery of airports and surrounding areas: Archived S A R data / Final Reoort. 31 AUCI. 1987 - 30 NOV. 1989 (AAIONSTOTT, ROBERT G. ; (AB)bINERIS, DEGISE J. Environmental Research Inst. of Michigan, Ann Arbor. (E03562831 NASA-CR-4275; NAS 1.26:4275; DOT/FAA/DS-89/14 NASI-18465 505-67-41-57 NASA Washington 900200 p. 214 In: EN (English) Avail: NTIS HC AlO/MF A02 p.1453 The statistical description of ground clutter at an airport and in the surrounding area is addressed. These data are being utilized in a program to detect microbursts. Synthetic awrture radar I s m ) data were acquired from the ERIM SAR data- archive -and were examined for utility to this program. Eight digital scattering coefficient images were created of five airDorts. These data are described along with the results of the clutter stuby. These scenes were imaged at 9.38-GHz and HH- and W-polarizations and -contained airport grounds and facilities, indUstria1,~iesidential. fields, forest, and water. Incidence angles ranged from 12 to 72 deg. Even at the smallest incidence anales. the distributed taraets such as forest. fields, water, and residential' rarely had mean scafteriny coefficient;

Which produced the largest cross sections were largely confined to the airport grounds and areas highly industrialized. The largest cross sections were produced by observing large buildings surrounded by Smooth surf aces. Author

TYPE 1/4/114 Quest Accession Number : 90A17853 90A17853* NASA IAA Journal Article Issue 05 Phase calibration of polarimetric radar images (=)SHEEN, DAN R.: (ABIKASISCHKE, ERIC S . ; (AC)PREEMAN, ANTHONY (AB)(Michigan, Environmental Research Institute, Ann Arbor): (AC)(JPL, Pasadena, CAI Environmental Research Inst. of Michigan, Ann Arbor. (E03562831 N00014-87-C-07262: NAGW-1101 (Remote sensing Society, IEEE, URSI, et al., International Geoscience and Remote Sensing Symposium /IGARSS ' 8 8 / on Remote Sensing: Moving Towards the 21st Century, Edinburgh, Scotland, Sept. 12-16, 1988) IEEE Transactions on Geoscience and Remote Sensing IISSN 0136-28921. ,~~~~ . . VOl. 27. NOV. 1989. . .0. 719-731. 891100 0. 13 refs 14 In: EN (English) p.663 The problem of phase calibration between polarization channels of an imaging radar is studied. The causes of various types of phase errors due to the radar system architecture and system imperfections are examined. A simple model is introduced to explain the spatial variation in phase error as being due to a displacement between the phase centers of the vertical and horizontal antennas. It is also shown that channel leakage can cause a spatial variation in phase error. Phase calibration using both point and distributed ground targets is discussed and a method for calibrating phase using only distributed target is verified, subject to certain constraints. Experimental measurements using the NADC/ERIM P-3 synthetic-aperture radar ( S A R ) System and NASA/JPL DC-8 SAR, which operates at C-, L-, and P-bands, are presented. Both of these Systems are multifrequency, polarimetric, airborne, SAR systems. I.E.

TYPE 1/4/115 Quest Accession Number : 90A17852 90A17852 NASA IAA Journal Article Issue 05 Multitemuoral and dual-uolarization observations of aqricultural vegetation'covers by X-band SiR images (AA)LE TOAN, THUY; (AB)MOUGIN, ERIC: (AC)LOPES, ARMAND; (A0)LAUR. HENRI d'Etude Spatiale des Rayonnements, Toulouse, France); (Ac)(centre (ADIIESA. Eurooean SDace Research Institute, Frascati, Italy) ' ( i & " i s e n k g Society, IEEE, URSI, et al., Internatianal Geoscience and Remote Sensing symposium /IGARSS ' 8 8 / on Remote sensing: Moving Tnwirdc t h e 71s+ C a n t i i r v . Edinburoh. Scotland. Seot. 12-16. 1988) IEEE NO". 1989, p. 709-718.
891100

p. 10

refs 18- In: EN (English)

p.684

A study is presented Of synthetic-aperture (SARI images which aims to demonstrate the capability of radar to identify vegetation types and to determine the vegetation canopy parameters. The investigations, performed on bitemporal and dual-polarization (HH and W ) X-band S A R images of an agricultural scene, included all the steps involved from S A R images to applications: (1) radar backscatter values were retrieved from SAR images, using external references targets: (2) the polarization and temporal responses of the vegetation covers were analyzed, highlighting the particular behavior of flooded rice fields; (3) the observations of rice fields were interpreted by a theoretical model. During an extended period of. rice plant growth, the model, in agreement with the observations, suggests a possible use of both HH and W images for rice field mapping and monitoring. An example of rice field mapping with an algorithm based on the above results is also presented. I.E.

TYPE 1/4/116 Ouest Accession Number : 90A17851 90A17851 NASA IAA Journal Article ISSUe 05 Estimation of the SAR system transfer function through processor defocus ILAIVACHON. PARIS W.: R. KEITH . IABIRANEY. . . ~---, (AB)(Canada Centre for Remote Sensing, Ottawa) (Remote sensing society, IEEE, URSI, et al., International Geoscience and Remote sensing symposium /IGARss '88/ on Remote sensing: Moving Towards the 21st Centurv. _ . Edinburgh. Scotland. Seut. 12-16, 1988) IEEE Transactions on Geoscience and Remote'sensing ( ISSN-0196-28321, vol. 27, NOV. 1 9 8 9 , p. 702-708. 891100 p. 7 refs 12 In: EN (English) p.651
~~~~~~~

~~~~~~

~~~~

~~~~

TYPE 1/4/117 Quest Accession Number : 90A17846 Qm17R4fi Nasa IAA Journal Article Issue 05 Comulex S A R imaaerv and sueckle filtering for wave imaging (ABjMACKLIN, S . TREVOR (AA')COROEY, RALPH (AB)(GEC Research, Ltd., Marcani Research Centre, Great Baddow, England) IR-mnte senqino , Sacietv. IEEE. URSI. et al.. ~nternationalGeoscience and Remote sensing svmuosium /iGARss' ' 8 8 / bn Remote Sensing: Moving Edinburgh, Scotland, Sept. 12-16, 1988) IEEE Towards the 21st -Centu;y, Transactions on Geoscience and Remote sensing Sensing (ISSN 0136-2892). VO1. 27. N o v ~ 1989, 19RX n. NOV. p. 66fi-fi73. 666-673. Research suuuorted supported bv by ESA and Royal AerOSUaCe Aerospace Establishment. 891100 p. 8 refs 13 In: EN (English) p.694
~~~~~~

A.;
~~~~~~

~.

It is generally accepted that in order to derive wave directional spectra from synthetic-aperture radar ( S A R ) image data, the system transfer function (STF) must be removed from the raw SAR image spectrum. The STF, very nearly equivalent to the magnitude of the Fourier transform of the N-look perfect-focus impulse response, can be estimated from actual SAR image data as the magnitude of the Fourier transform of the autocorrelation function of the Speckle in a low-contrast scene. The authors Outline and demonstrate a novel approach for estimating the STF based on the observation that system defocus, to the first order, does not impact speckle statistics. Therefore, a speckle pattern suitable for STF estimation is produced from typical SAR data, including ocean waves, by operating the processor out of focus. Processor defocus does not %blur' the speckle, but it does blur the wave image, thus reducing the image correlation function, and hence the image spectrum to be essentially that of the STF desired. I.E.

A method is described to predict the wavenumber dependence of the speckle component in spectra of Synthetic aperture radar intensity images. Filtering of this component is an important Step in recovering waveheight spectra for synthetic-aperture radar (SARI images of the ocean, and an effective means of doing so is required for the 'wave mode' of the European satellite ERS-I. The method uses the correlation function of the corresponding complex images and has been teated using a variety of airborne and spaceborne imagery obtained over both land and sea. Examples are shown of both Successful and unsuccessful applications of the method. The successes show a great improvement in speckle filtering over previous techniques, while the failures can uossibly be explained in terms of artifacts of particular sAR processors. I.E.

TYPE 1/4/118 Quest Accession Number : 90N17744 90N17744ff NASA STAR Conference Paper Issue 10 Power conditioning for active array SAR antennas (AA)SCHAEPER, W. (00272085) Dornier System G.m.b.H., Friedrichshafen ( G e r m a n y , F.R.). In ESA. European Space Power, Volume 1 p 457-461 (SEE N90-17678 10-20) 890800 p. 5 In: EN (English) Avail: Nl'IS HC Al9/MF A03 p.1338 A trade off between different overall concepts rather than circuit details, in designing active array SAR (synthetic aperture radar) antenna Power conditioning and distribution Subsystems, is presented. A hierarchically structured system employing ac distribution is shown to be favorable. Active antennas are shown to be failure tolerant to a certain extent. Special attention is paid to particular reliability aspects. Breadboard hardware is presented. A discussion of rurthcr improvement of pulse energy storage is included. ESA

TYPE 1/4/119 Quest Accession Number : 90N16112 90N16112# NASA STAR Technical ReOOrt Issue OR Necessity and benefit of the X-SAR Space Shuttle exper~ment (AA)KEYDEL, WOLFGANG; (AB)OETTL, HERWIG (AB)(Deutsche ForSChungs- und VersuChSan5talt fuer Luft- und Raumfahrt, Oberofaffenhofen. Germanv .. F.R. i Euiopean Space'Agency, Paris (France). (E6854803) ESA-TT-1164; DFVLR-MITT-88-29; ETN-90-96192 Transl. into ENGLISH of Notwendigkeit und Nutzen des X - S A R - S p a C e - S h u t t l e - E x p e r i m e n t s 891000 p. 38 Oriqinal language document was announced as N88-246R8 T n~ : N . .F ~ (Englishj Avail:. HC A O ~ / M F A O ~ :original ~ e r m a nversion available from DFVLR, VB-PL-DO, Postfach 90 60 58, 5000 Cologne, Federal Republic of Germany. 26.50 deutsche marks p.1057

NTIS

The X-band ( 9 . 6 G I I Z ) Synthetic Aperture Radar (X-SAR) is studied. It i s the first space qualified SAR in X-band. The X-SAR is to be flown on several Shuttle missions as Dart of the O.S. Radarlab nrnnrnm . for . . Parth observation. The examples shown clearly indicate that the X-band is an extremely valuable and important frequency range for areas of remote sensing. X-band measurements supply valuable additions to measurements in the L- and C-bands. It can be applied to the classification of cultivated land surfaces and of ice and snow and for finding pollution in rivers, lakes and on the seas. ESA
~~~~~~ ~~~~~

TYPE 1/4/120 Quest Accession Number : 90N11218 90Nll218# NASA STAR Technical Report Issue 02 Simulation of synthetic aperture radar 1: Feature density and accuracy requirements / Final Report, Jan. 1986 - an. 1988 (AA)CP.ANE, PETER M.; (AB)BELL, HERBERT H.; (AC)XALINYAX, ROBERT G.; (AD)DOOLEY, JOHN; (AE)HUBBARO, DAVID C. (AE)(Dayton Univ., OH.) Air Force Human R e s o U l ~ e SLab., Brooks AFB. TX. (AI169645) AD-A211019; AFHRL-TR-88-42 890700 p. 37 In: EN (English) Avail: NTIS p.213 HC A03/HF A01 Simulations of ground mapping radar are generated from Defense Mapping Agency Digital Feature Analysis Data (DFAD) Products Which Were developed to sipport real beam ground mapping radar.-Criteria for inclusion of-an object within a given level of DFAD are radar significance, height, and size. In aeneral. l e v e l 1 includes abiects loom and l a m e r and level 2 includes objects '30m and larger. A prot;itype DFAD incorpo;ating features as small as 10m (level X) was developed to support the higher resolution Synthetic Aperture Radar (SAR); however, the minimum data base rrquirements for simulating SAR have not been determined. The present reoort describes three studies conducted to determine these requirements. In- the first study, 8-1B and F-15E radar operators rated SAR Siimulations produced from level 2 DFAD (30m) and level X (lam), plus two experimental data bases with 15m and 20m cauture criteria. Eiahtv-two Dercent of SAR-experienced subjects found th; 15m data acceptable. Simulations prodiiied f r m levsl 2 depict& a r e z s of high rsflector deiisit'y' as a single feature with uniform brightness and were acceptable to only 18 percent of the Subiects. A Stud" of the Radar Scone Interuretation 1RSIl cues used bv 8-1B oifensive S V S & n S officers C O S b S ) founi that roads and rivers wer;? critical cues, whereas individual Structures were used only in the immediate vicinity of the aimpoint. A third study compared B-1B os0 oerformance on a naviaation uodate task usina 10m and 15m feature data and kenerically enhanced ;om (level 2) and loom [level 1) GRA

TYPE 1/4/121 Quest Accession Number : 90A10187 90A10187 NASA IAA Conference Paper Issue 0 1 Speckle statistics in synthetic aperture radar (SAR) imagery with correlated looks (AA)APRIL, G . V.; (AB)HARVEY, E. R. (AB)(Universite Laval, Quebec, Canada) IN: statistical Optics: Proceedings of the Meeting, Sa" oiego, CA, A U ~ . 18, 19, 1988 (A90-10186 01-35). Bellingham, WA, Society of Photo-Optical Instrumentation Engineers, 1988, p. 2-7. 8 8 0 0 0 0 p. 6 refs i o I": EN (English) p.26 The high resolution in Synthetic aperture radar (SAR) systems i s reduced by the Speckle appearing in the images. This speckle noise is generated by the coherent processing of radar signals and exists in all types of coherent imaging systems. The Statistical properties of speckle noise in a four-look seasat SAR image are Studied to show evidence of correlation between underlying looks resulting in correlated values for adjacent pixels in the final image. Concentrating on homogeneous areas of the image, experimental determination of the SNR and Of the probability density Of the recorded intensity i s performed. It is shown that the intensity fluctuations due to speckle noise in homogeneous areas are not gamma distributed. A model incorporating the empirical correlation coefficient is extracted from experimental data. Author

_ .

'TYPE 1/4/126 Quest Accession Number : 89A39617 89A39617* NASA IAA Journal Article Issue 16 Geometric accuracy in airborne SA8 images (AA)BLACKNELL, D . : (ABIQUEGAN, S.: (ACIWARD, 1. A.; (AD)FREEMAN, A.; (AE)FINLEY, I. P. (AC)(GEC Research, Ltd., Marconi Research Centre, Great Baddow, England) i (AD)(california Institute of Technology, Jet Propulsion Laboratory, Pasadena: GEC Research. 1,td.. Marconi Research Centre. Great Baddow. England) : (AE)(Royal Signals and Radar Establishment, Malvekn, England) GEC-Marcmi Electronics Ltd., Chelmsford (England). (GB135511) IEEE Transactions on aerospace and Electronic Systems (ISSN 0018-9251). "01. 25, March 1989, p. 241-258. Research supported by the Royal signals and Radar Establishment. 890300 p. 18 refs 11 In: EN (English) p. 2483 Uncorrected across-track motions of a synthetic awrture radar (SARI platform can cause both a severe loss of azimuthal positioning accuracy in, and defocusing of, the resultant SAR image. It is shown how the results of an autofocus procedure can be incorporated in the azimuth processing to produce a fully focused image that is geometrically accurate in azimuth. Range positioning accuracy is a150 discussed, leading to a comprehensive treatment of all aspects of geometric accuracy. The system considered is an X-band SAR.
1 . 7 3

TYPE 1/4/127 Quest Accession Number : 89A39609 89A39609 NASA IAA Journal Article Issue 16 Estimating the Doppler centroid Of SAR data (AA)MADSEN, S. NORVANG (AA)(Danmarks Tekniske Hojskole, Lyngby, Denmark) I E E E Transactions On Aerospace and Electronic Systems (ISSN 0018-9251). vol. 25, March 1 9 8 9 , p. 134-140. 890300 p. 7 refs 1 3 In: EN (English) 0.2482 After reviewing frequency-domain techniques for estimating the ~ o p p l e r centroid of synthetic-aperture radar (SAR) data, the author describes a time-domain method and highlights its advantages. In particular, a nonlinear time-domain algorithm called the sign-Doppler estimator (SDE) is shown to have attractive properties. An evaluation based on an existing SEASAT processor is reported. The timc-domain algorithms are shown to be extremely efficient with resDect to reauirements on calculations and memory, -and hence they are - w e l l suited-to real-time systems where the Doppler estimation is based on raw SAR data. For Offline processors where the Doppler estimation is performed on processed data, which removes the problem of partial coverage of bright targets, the DeltaE estimator and the CDE (correlation Doppler estimator) algorithm give similar performance. However, for nonhomogeneous scenes it is found that the nonlinear SDE algorithm, Which estimates the Dopoler-shift on the basis of data signs alone, gives superior performance. I.E.

TYPE 1/4/128 Quest Accession Number : 89A38341 89.438341 NASA IAA Conference PaDer ISSUe 16 Preparatory microwave remote sensinq activities

in

View

of

the

IN: International symposium on Space Technology and science, lbth, Sapporo, Japan, May 22-27, 1988, Proceedings. Volume 2 (A89-38031 1 6 - 1 2 ) . Tokyo, AGNE Publishing, Inc., 1988, p. 2271-2276. 880000 p. 6 refs 7 In: EN (English) p.2455 Current developments regarding airborne SAR-sen~ors in DFVLR are described as well a5 radiometric calibration activities. The characteristics of the L-band and c-band SAR are presented as well as those Of airborne X-band SAR. It is Concluded that a hiqh radiometric resolution and a calibration in all frequency ranges u & d as well as interchannel stability of all sensors is necessary for the operational interpretation of SAR images. K.K.

. ..

_.

TYPE 1/4/129 Quest Accession Number : 89.435334 89A35334# NASA IAA Journal Article Issue 14 stochastic processes in microwave remote sensing stochastische,prozesse in der Mikrowellenfernerkundung (AA)KEYDEL, W. (AA)(DFVLR, Institut fuer Hochfrequenztechnik, Oberpfaffenhofen, Federal Republic Of Germany) (URSI and Informationstechnische Gesellschaft, Gemeinsame Tagung, Kleinheubach, Federal Republic of Germany, Oct. 3-7, 1988) Kleinheubacher Berichte (ISSN 0343-5725). VO1. 32, 1989, p. 151-151h. In German. 8 9 0 0 0 0 p. 9 refs E In: GM (German) p.2144 Stochastic processes and their significance for the efficiency of systems and remote sensing procedures in the microwave radiometry and radar are discussed. Fluctuation problems in remote sensing, image effects, and specxle are addressed, and the influence of system components and eigennoise in SAR are considered. The geometry and radiometric resolution power, measurement accuracy, and tne calibration of microwave remote sensing systems are examined. C.D.

TYPE 1/4/130 Quest Accession Number : 89A31944 89~31944 NASA IAA Journal Article Issue 12 (synthetic x-SAR specification, design, and performance modeling aperture radar for shuttle mission) (AA)MILLER, DAVID (AA](Dornier system GmbH, Friedrichshafen, Federal Republic Of Germany) iEEE Transactions on Geoscience and Remote Sensing (ISSN 0196-2892). v01. 27, March 1989, p. 170-176. 890300 p. 7 refs 5 in: EN (English) p.1800 The detailed design and development phase of the X-SAR 9.6-GHz vertical polarization segment of the Shuttle Radar Lab (SRL) mission, Started in spring 1987, is discussed. The design incorporates a planar array antenna based on metallized carbon-fiber-reinforced plastic technology, a traveling-wave-tube amplifier, and a dual-channel receiver using surface acoustic wave devices. several radar parameters are controllable from ground to achieve high performance over a wide range of measurement geometries. Definitions of image performance parameters and modeling algorithms developed in Support of instrument specification and performance monitoring during hardware development are presented. The algorithms have been implemented as a software tool to perform sensitivity analysis and generate interesting graphical results to support top-level instrument verification.
I.E.

TYPE 1/4/131 Quest Accession Number : 89A29428 89A29428 NASA IAA Journal Article Issue 11 Airborne MTI via diqital filterinq (AA)ENDER, J.; (ABIKLEMM, R. (AB)(Forschungsgesellschaft fuer angewandte Naturwissenschaften, Forschungsinstitut fuer Funk und Mathematik, Wachtberg-Werthhoven, Federal Republic Of Germany1 IEE Proceedings, Part F: Radar and Signal Processing (ISSN 0143-70701, vol. 136, pt. F, no. 1, Feb. 1989, p. 2 2 - 2 8 . 890200 p. 7 refs 9 In: EN (English) p.1594 A simple clutter suppression technique for airborne moving target indicators (MTIs) is proposed which exploits the special properties Of a linear equispaced array antenna aligned in the flight direction. In the present method, equidistant pulses are transmitted, and the temporal and spatial samples are equivalent. The method is found to provide better target detection than airborne MTI techniques based on oversampling. It is suggested that the inverse clutter filter can be employed a5 an output for SAR imaging with improved SNR. R.R.

0 -

TYPE 1/4/132 Quest Accession Number : 89N28942 89N28942# NASA STAR Technical Report ISSUe 23 The Archimedes 2.3 Experiment on oil slick dection over the North Sea, Avril 1988: Measurement Results ohtained by the E-SAR system of the Aerospace Research Establishment (AA)HORN, RALF; (ABIMOREIRA, ALBERT0 Deutsche Forschungs- und Versuchsanstalt fuer Luft- und Raumfahrt, Obervfaffenhofen (West Germany). (DO6990601 Abteilung Hochfrequenzsysteme DFVLR-MITT-89-08; ISSN-0176-7739; ETN-89-95299 890200 p. 4 6 In: EN (English) Avail: NTIS HC A03/MF AOl; DFVLR, VB-PL-DO, Postfach 90 60 58, 5000 Cologne, Fed. Republic of Germany, 18.50 OM p.3319

TYPE 1/4/133 Quest Accession Number : 89N28911 89N28911# NASA STAR Conference Paper Issue 23 Real time SAR processing techniques (AA)SCHOTTER, ROLAND Joint Publications Research Service, Arlinqton, VA. (519573941 In its JPRS Report: science and Teihnology. USSR: Space. 16th International congress of the international Society for Photogrammetry and Remote Senaino. Volume 2 0 7 5 - 8 2 (SEE N89-28903 23-431 Transl. into ~~~~~, ENGLISH from International Archives of Photogrammetry and Remote Sensing (Kyoto, Japan), V. 27, pt. B9, 1988 p 184-190 890131 p. 8 In: EN (English) Avail: NTIS HC AO9/MF A01 p.3314
~~~~~ ~

The results of the Archimedes 2a experiment on Oil slick detection over the North Sea obtained by the experimental airborne SAR System E-SAR (synthetic aperture radar) are presented. This system is part of the research program at the Institute for Radio Frequency Technology of the German Aerospace Research Establishment. During the experiment carried Out on 21 April 1988 the System was operated in L-band at 1.29 GHz. The report discusses briefly Some problems with oil slick detection by SAR. It gives an overview of the E-SAR system configuration and system performance. SAR data and image processing procedures are described as well. SAR images of an oil slick of 5 cubic meters of crude oil and corresponding image analysis results are presented. ESA

Spaceborne synthetic aperture radar (SAR) Systems provide raw radar data information at high data rates of 10 MWOrdS per second. Real time SAR more than 1 giga-operations processors, therefore, must perform (multiplications, additions) per second in order to generate images from the raw data. Since conventional computer Systems are not able to cope with these requirements oornier has developed real time SAR processors on the basis of a modular pipeline concept. The processing pipeline is built up to standardized hardware modules which are required cor digital signal processing of two-dimensional data. These module$ show identical electrical and mechanical interfaces so that each hardware module can be used in any place of the pipelines. The basic principles are presented for the real time modular pipeline concept and its processing rate requirements for real time SAR processing applications. The implementation of Some of the most important modules like Fast Fourier Transformation, correlation, interpolation, and data memory is described. Finally, a short outlook on further applications of the pipeline processing concept is given. Author

TYPE 1/4/134 Quest Accession Number : 89N28890 89N28890# NASA STAR Conference Paoer Issue 2 ? EXAS: Experiment a n Autonomous SAR erocessor calibration (AA)RUNGE, H.; (AB)POPELLA, A,: (ACjNOACK, W. Joint Publications Research service, Arlington, Va. (~1957394) In its JPRS Reoort: Science and Technoloov. U S S R : S ~~. n . ~. r 16 t .r ~ . . . .h . International Congress of the International Society for Photogrammetry and Remote Sensing, Volume 1 p 132-140 (SEE N89-28876 23-43) Transl. into ENGLISH from International Archives Of Photogrammetry and Remote sensing (Kyoto, Japan), V. 27, 1988 p 369-377 890130 P. 9 In: EN (English) Avail: NTIS HC A12/MF A 0 1 p.3311

~~~~~~

~~

_ 1

~~~~~

TYPE 1/4/135 Quest Accession Number : 89N28882 89N28882ff NASA STAR Conference Pane= Issue 21 KRAS: A Danish high resolution airborne SAR (-)MADSEN, SOREN NORVANG; (AB)CHRISTENSEN, ERIK LINTZ; (AC)SKOU, NIELS Joint Publications Research service, Arlington, VA. (51957394) In its JPRS Report: science and Technology. USSR: Space. 16th International Congress Of the International Society for Photogrammetry and Remote Sensing, Volume 1 p 48-57 (SEE N89-28876 23-43) ~ r a n ~ 1into . ENGLISH from International Archives of Photogrammetry and Remote sensing (Kyoto, Japan), v . 27, 1988 p 90-97 890130 p. 10 In: EN (English) Avail: NTIS HC A12/MF A01 p.3309 A C-band high resolution airborne Synthetic aperture radar (SAR) is presently being Constructed. The main purpose of the project, which i s called KRAS, is to develop the knowledge base required to build advanced coherent radars. The design rationale is presented. The design of the radar is based on digital technology to the largest possible degree. This results in a very flexible radar system, with most of the system parameters being Software controlled. Variable waveforms of bandwidths larger than 100 MHz and durations up to 20 microsec can be generated. Calibration of the System was also given much consideration, and design principles usually applied in radiometers were implemented. The significant flexibility and the calibration is of major importance since the system is intended f o r applications ranging from medium resolution wide swath mapping, i.e., sea ice mapping or oil pollution surveillance, to high resolution narrow Swath mapping f o r cartography o r reconnaissance. Author

Future synthetic aperture radar (SAR) missions like the SIR-C with the German X-band radar will acquire multifrequency and multipolarization data with various incidence angles. In order to exploit the missions full scientific potential DFVLR is going to calibrate both its Intelligent SAR Processor (ISAR) and the final image products. The idea of the EXAS proposal i s to use the extra data gained from an independent Doppler measurement. These data will be gathered at a geolocated receiver in order to tune and finally calibrate the ISAR processor. The receiver will be adjusted to the sensor carrier frequency, the signals will be mixed down to baseband, be digitalized and transferred to the ISAR hardware system. This can be performed in parallel for three frequencies with two different polarizations. Immediately after the fly-over the data will be analyzed. The azimuth antenna pattern will be computed and fed back into the processing chain. The replica Of the chirps radiated by the sensor will be recorded by the ground receiver. Therefore, the exact range reference function is known and can be used f o r the processing. The calibrated processor will produce images with a very accurate absolute pixel location. In order to verify this, a cluster of geolocated point reflectors shall be positioned over the Swath and the SAR image will be a cartographic map. The experiment plan and the compared with contributions of the three institutes are described. Author

TYPE 1/4/136 Quest Accession Number : 89A2684.5 89A26846 NASA IAA Journal Article Issue 10 A simulation far spaceborne SAR imagery Of a distributed, moving scene (AAIVACHON. PARIS W.: (AB)RANEY. R. KEITH; (ACIEMERY, WILLIAM J. (AA)(can?.de. Centre for Remote Sensing, Ottawa); (AB)(Radarsat Project Office, Ottawa. Canada); (AC)(Colorado. University, Boulder) IEEE Transactions on Geoscience and Remote sensing (ISSN 0196-2892). VOl. 27, Jan. 1 9 8 9 . p. 67-78. 890100 p. 12 refs 37 In: EN (English) p.1510
A computer simulation that is designed to represent aspects of spaceborne synthetic-aperture radar ( S A R ) imaqery of the ocean surface is presented. The simulation is unique in that-a scatterer density (per resolution cell) is explicitly included, thus allowing the incorporation of various scatterins natures. from ourelv saecular to ~ u r e l vdiffuse. T h e ~~ simulation may be -applied t o ocekn su;fa& wave imaging since velocity bunching and Scene coherence times are also included. Certain assumptions inherent in the velocity bunching formulation limit the applicability of the simulation in its present form to soaceborne SAR svstems onlv. TWO experiments based on this simulation are considered: ( i ) the effect of varying the target density; and ( 2 ) the effect of the mean scene coherence time in the -i -m i l n i n n clf e O C B Z ~ sr?ell nystom. In each case, the ; i m i i : t i n ..-3---x Outputs are compared with actual SEASAT SAR imagery. On the basis of certain statistics derived f r o m the simulated scenes, it is Shown that specular statistics are quantitatively correct for scenes that may appear diffuse in hard copy form. This observation Suggests that quantitative norms be used (rather than intuitive oDinion or armearanre) for . . . investigating ocean scattering statistics, for example. I.E.
~~~~~~~~~~ ~ ~~~~

TYPE 1/4/137 Quest Accession Number : 89N24688 89N24688# NASA STAR Technical Report Issue 18 Necessity and benefit of the X-SAR space shuttle experiment (AA)KEYDEL, WOLFGANG; (AB)OETTL, HERWIG DeUtSche ForsChllna5- und Versuchsanstalt fuer Luft- und Raumfahrt. Oberpfaffenhofen (Wekt Germany). (00699060) DFVLR-MITT-88-29: ISSN-0176-7736; ETN-89-94375 881000 p. 41 Original contains color illustrations In GERMAN; ENGLISH summary In: AA (Mixed) Avail: NTIS HC AO?/MF A01: DFVLR. VB-PL-DO. PDStfach 90 60 58. 5000 Cologne, Federal Republic of Germany. 26.50 deutsche marks p.2582
~ ~~~

~.

The X, L, and C-band synthetic aperture radars (SAR) planned to fly in 1992 on space shuttle missions are investigated for their effectiveness in Earth surface vegetation classification and recognition. The scattering of electromagnetic waves by twigs, leaves, branches, and trunks permits vegetation growth length estimation. Earth surface roughness is analyzed by. Rayleigh criterion. The SAR sensors are well adapted to Earth surface observation independently of daylight and atmospheric conditions. ESA

TYPE 1/4/138 Quest Accession Number : 89A22586 89~22586 NASA IAA ~ o u r n a lArticle ISSW 07 Validation Of a svnthetic aoerture radar ocean wave imaaina theorv hv the Shuttle Imaging Ra,kar-B expekiment over the North S e a (AA)BRUENING, CLAUS: (AB)AI.PERS, WERNER; (AC)ZAMBRESKY, LIANA F.: (AD)TILLEY, DAVID G. (AA)(European Centre f o r Medium Range Weather Forecasts, Reading, England): (AB)(Bremen, Universitaet, Federal Republic of Germany): (AC)(European Centre for Medium Range Weather Forecasts, Reading, England: GKss-Forschungszentrum Geesthacht GmbH, Federal Republic of Germany); (AD)(Johns Hopkins University, Laurel, MU) BMFT-01-QS-86174: N00014-83-G-0126 Journal of Geophysical Research (ISSN 0148-0227). 01. 93, uec. 15, 1988, p. 1540~-15425. 881215 p. 23 refs 26 In: EN (English) p.1029

-~

~1

SAR image intensity spectra measured during the SIR-B mission (October 6 and October 8 , 1984) over the North Sea were compared with an ocean wave spectra hindcast by n third-generation wave prediction model (WAMODEL). It was found that, while the hindcast ocean wave spectra had only single Desks, most of the measured SAR imaqe spectra Of October 6 showed double beaks. It is Shown that the double peaks-are generated by the SAR imaging mechanism, when the SAR MTF consisting of the sum of the complex real aperture radar (RAR) MTF and the velocity bunching MTF has a strong minimum near the range direction, by Which the wave spectrum is cut into two. The azimuth angle at which this minimum occurs depends strongly o n the phase of the RAR MTF. I.S.

TYPE 1/4/139 Quest Accession Number : 89N18953 89N18953# NASA STAR Conference Paper Issue 11 The impacts of yuantisation noise on the ERS-1 synthetic aperture radar performance (AAIRICHARDS,8. E.: (AB)LANCASHIRE, D. c. Marconi Space Systems Ltd., Portsmouth (England). (MF879009) In ESA. Proceedings Of the 1988 International Geoscience and Remote Sensing Symposium (IGARSS) 8 8 on Remote Sensing: Moving Towards the 21st Century, Volume 2 p 1159-1160 (SEE N89-18836 1 1 - 4 3 ) 880800 p. 3 In: EN (English) Avail: NTIS HC A99/MF AOl; ESA Publications Div. ESTEC, Noordwijk, Netherlands, $120 US Or 250 Dutch quilders 0.1552 The contribution of digitizing errors to system resolution is considered. It is Often described as an additional quantization noise power given by delta 2/12 where delta is the digitizer step size. In a real system, this is an oversimplification and if the less obvious effects of digitizing are not included, then System performance can be seriously degraded. A qualitative description is given of possible effects and the tradeoff necessary between relevant system parameters. The quantizing principles as applied to the ERS-I SAR are presented and the treatment of system imperfections described. Numerical examples for an ideal case Of a perfect receiver and for a more realistic receiver are given. ESA

TYPE 1/4/140 Quest Accession Number : 89N18949 89N18949# NASA STAR Conference Paper Issue 11 Change detection in AGRISAR images (A&)QUEGAN, S . ; (AB)YANASSE, C.; (AC)BLACK, S.; (AD)DANSON, M. Sheffield univ. (England). (51380652) I n ESA, Proceedings of the 1988 International Geoscience and Remote Sensing symposium (IGARSS) 8 8 on Remote Sensing: Moving Towards the 215t Century, Volume 2 p 1139-1140 (SEE N89-18836 11-43) 880800 p. 2 In: EN (English) Avail: NTIS HC A99/MF AOl; ESA Publications Div. ESTEC, Noordwijk, Netherlands, $120 US or 250 Dutch guilders p.1552 Synthetic aperture radar ( S A R ) data at four dates spanning the growing season over a UK Site was used to derive multitemporal crop radar cross section signatures. This involves measuring and correcting variations due to the SAR System within single images, and calibrating images gathered at different times. Calibration based on use of point targets, fields of particular crop types, and global measures assuming statistical homogeneity are evaluated. NO basis for multitemporal calibration is apparent. ESA

TYPE 1/4/14] , , Quest Accession Number : 89N18948 89N189486 NASA STAR Conference Paper ISSUB 11 Extraction of agricultural plant parameters from multitemporal Thematic MBDDBI .. lTMi and X-SAR data (AA)MAusER, w.: (ABIRIEG,A. Freiburg Univ. (Germany, F.R.). (F6817510) Inst. for Physical Geooraohv and HYdroloOV. BMFT-01-QS-86090; BMFT-01-QS-87033 In ESA, Proceedings Of the 1988 International Geoscience and Remote Sensing Symposium (IGARSS) 88 on Remote Sensing: Moving Towards the 2lst century, Volume 2 p 1133-1137 (SEE N89-18836 11-43) 880800 p. 5 In: EN (English) Avail: NTIS HC A99/MF A01: ESA Publications Div. ESTEC, Noordwijk, Netherlands, $120 US O r 250 Dutch guilders p.1551
I

TYPE l/4/142

..

__

Quest Accession Number : 89N18925 89N189256 NASA STAR Conference Paper Issue 11 Taking a broader view: Radarsat adds ScanSAR to its operations (AA)I.USCOMBE, A. P . Spar Aerospace Ltd., Ste-Anne-de-Bellevue (Quebec). (SV029037) In ESA, Proceedinqs of the 1988 International Geoscience and Remote sensing symposium (IGAKSS) 8 8 on Remote Sensing: Moving Towards the 215t Century, Volume 2 p 1027-1032 (SEE N89-18836 11-43) Sponsored in part by 880800 13. 6 In: EN (Enalishi Avail: NTIS HC the Canadian qovernment A99/MF AOli E<A Publications Div. ESTEC, Noordwijk; Netherlands, $120 U S or 2 5 0 Dutch guilders p.1548 ScanSAR operations were added to Radarsat SAR CaDabilities to enable very wide Swaths to be imaged. The principle of ScaiSAR is Outlined and thc critcria used in defining the operations are explained. The major imolications of ScanSAR f o r the SAR svstem. in the instrument on-board the saiellite and in the processor on the ground, are identified. The definition of 2 standard forms Of ScanSAR operation, providing coverage respectively of swaths of over 300 and 500 km, is described. These swath widths compare with the maximum of 150 km previously available with Radarsat using conventional SAR imaging. The ScanSAR operations are not restricted to these two standard forms, however; the SAR can also image with a variety of other combinations of beams on command from the ground. ESA

Within the AGRISAR86 campaign, 4 S A R images were produced of an agricultural area using the VARAN-S X-band SAR. Three TM-scenes were analyzed. Ground truth^ was gathered on an area Of 48 sqkm and the field boundaries were digitized to produce images of the measured plant parameters. After qeometric reaistration. reqressions were calculated between sensor data-and measured-biomass, plant-height, and water content of different plant species. Land use classifications were carried out. Results show strong correlations between plant parameters and the ratio between bands 4 and 5 of TM. For the SAR data a separation between cereals and corn is possible using a multitemporal approach. ESA

TYPE 1/4/143 Quest Ac&ession Number : 89N18924 89N189246 NASA STAR Conference Paper Issue 11 E-SAR: The exoerimental airborne L/C-band SAR SYatem of DFVLR (AA)HORN, R. Deutsche Forschungs- und Versuchsanstalt Suer Luft- und Raumfahrt, Wesseling (Germany, F.R.). (00705482) Inst. for Radiofrequency Technology. In ESA. Proceedings of the 1988 International Geoscience and Remote Sensing symposium (IGARSS) 88 on Remote Sensing: Moving Towards the 21st Century, Volume 2 p 1025-1026 (SEE N89-18836 11-43) 880800 p. 2 In: EN (English) Avail: NTIS HC A99/MF AOli ESA Publications Uiv. ESTEC, Noordwijk, Netherlands, $120 US or 250 Dutch guilders p.1548 The E-SAR system was developed to Study the SAR method and its problems, such as motion error correction and overall system calibration. The sensor is desiqned to operate on board DO 2 2 8 aircraft in either L- or C-band. The system features stripline active array anteniras, built in test equipment for system calibration, real time motion error correction, and quicklook data processing. ESA

TYPE 1/4/144 Quest Accession Number : 89N18923 89N189236 NASA STAR Conference Paper Issue 11 X-SAR: A new spaceborne Synthetic aperture radar (AA)VELTEN, E. H. Dornier-Werke G.m.b.H., Friedrichshafen (Germany, F.R.). (D0425275) In ESA, Proceedings of the 1988 International Geoscience and Remote sensing symposium (IGARSS) a8 on Remote Sensing: Moving Towards the 21st century, volume 2 p 1021-1024 (SEE N89-18836 11-43) 880800 p. 4 I: EN (English) Avail: NTIS HC A99/MF A01: ESA Publications Div. ESTEC, Noordwijk, Netherlands, $120 US or 250 Dutch guilders p.1547 The X-synthetic aperture radar ( S A R ) instrument design is described, including mechanical and electrical layout, operation modes, and interfaces to Shuttle and SIR-C, as well as major design parameters of the subsystems. The configuration of the ground segment is illustrated. The x-SAR design is determined by the selection of the resonant Slotted waveguide CRFP antenna working at a center frequency of 9.6 GHz and the high power amplifier using a 3 kW traveling wave tube. The radio frequency electronics comprises the frequency and chirp generation, up- and down-conversion, and filtering. The digital instrument control and data handling electronics provides high speed A/D conversion with selectable bandwidth, buffering, and conversi& to serial format f o r onboard storage on tape or direct transmission to ground. The X-SAR includes an onboard calibration system with two different loops. ESA

TYPE 1/4/145 Quest Accession Number : 89N18885 89N18885# NASA STAR Conference Paper Issue 1 1 Enuineering calibration Of the ERS-1 active microwave instrumentation in orbif (AA)ATTEMA, E. European Space Agency. European Space Research and Technology Center, ESTEC, Noordwijk (Netherlands). (E6889478) In its Proceedings of the 1988 International Geoscience and Remote sensing Symposium (IGARSS) ' 8 8 on Remote Sensing: Moving Towards the 21st Century, Volume 2 p 859-862 (SEE N89-18836 11-43) 880800 p. 4 In: EN (English) Avail: NTIS HC A99/MF AOl; ESA Publications Div. ESTEC, p.1541 Noordwijk, Netherlands, $120 US or 250 Dutch guilders The ESA baseline plan for radiometrical calibration of the output of the synthetic aperture radar and the scatterometer that constitute a Sensor known as Active Microwave Instrument (AMI) Onboard the ERS-1 satellite, is outlined. On the basis of a nonlinear, time-variant System model for the AMI, a composite Strategy is described consisting a prelaunch instrument characterization and postlaunch instrument monitoring, using internal stimuli, as well as external reference targets. ESA

TYPE 1/4/146 Quest Accession Number : 89N18851 89N18851*# NASA STAR Conference Paper ISSUe 1 1 The Alaska S A R processor (AA)CAP.ANDE, R. E.; (AB)CHARNY, 8. Jet Prooulsion Lab.. California Inst. of Tech.. Pasadena. (535744501 In ESA. Proceedinas of the 1988 International Geoscience and Remote Sensing Symposium (I~ARSS)' 8 8 on Remote Sensing: Moving Towards the 21st Century, Volume 2 p 695-6536 (SEE N89-18836 11-43) 880800 p. 4 In: EN lPnolish1 Avail: NTIS HC A99/MF A01: ESA Publications Div. ESTEC. ~, . . . . . ~ ~ , Noordwijk, Netherlands, $120 US or 250 Dutch guilders p.1536
~~~ ~~~ ~~

~~

~~~~~

~~~~

~~~

~~

The Alaska SAR processor was designed to process over 200 100 km x 100 km (seasat like) frames per day from the raw SAR data, at a ground resolution Of 30 m X 30 m from ERS-1. J-ERS-1. and Radarsat. The near real time processor is a set of custom hardware modules operating in a pipelined architecture, controlled by a general purpose computer. Input to the processor is provided from a high density digital cassette recording of the raw data stream as received by the ground station. A two pass processing is performed. During the first pass clutter-lock and auto-focus measurements are made. The second pass uses the results to accomplish final image formation which is recorded on a high density digital cassette. The processing algorithm uses fast correlation techniques for range and azimuth compression. Radiometric compensation, interpolation and deskewing is also performed by the processor. The standard product of the ASP is a high resolution four-look image, with a low resolution (100 to 200 m) many look image provided simultaneously. ESA

TYPE 1/4/147 Quest Accession Number : 89N18850 89N188506 NASA STAR Conference Paper Issue 11 ~ e w architecture for a real-time SAR processor (AA)AFSMBEPOLA, 8 . GEC-Marconi Electronics Ltd., Chelmsford (England). (GB135511) In ESA. Proceedinas Of the 1988 International Geoscience and Remote Sensing Symposium (IGARSS) ' 8 8 on Remote Sensing: Moving Towards the 21st Century, Volume 2 p 691-694 (SEE N89-18836 11-43) 8 8 0 8 0 0 p. 4 In: EN (English) Avail: NTIS HC A99/MF AOl; ESA Publications Div. ESTEC, Noordwijk, Netherlands, $120 US or 250 Dutch guilders p.1536 A processor architecture for real-time SAR azimuth processing is presented. It consists of a linear or circular array of identical processing modules. Hardware design is simplified by having a regular array of modules with nearest neiuhbor connectivitv. Architecture is expandable to meet a variety of swafh width, resolution, and throughput requirements. There is no explicit corner-turning. Input and Output are in range line order. Memory requirements are minimized. ESA

TYPE 1 / 4 / 1 4 8 Quest Accession Number : 89N18716 891318716# NASA STAR Conference Paper Issue 11 Multitemooral and dual Dolarization of auricultural C ~ O .D Sby . X-band SAR images (AA)LETOAN, T.; (AB)LAUR, H. Centre d'Etude Spatiale des Rayonnements, TOUloUSe (France). (CK523228) In ESA. Proceedinas of the 1988 International Geoscience and Remote sensing symposium (IGARSS) ' 8 8 on Remote Sensing: Moving Towards the 21st Century, Volume 3 p 1291-1294 (SEE N89-18704 11-43) 880800 p. 4 In: EN (English) Avail: NTIS HC A99/MF AOl; ESA Publications Division, ESTEC, p.1513 Noordwijk, Netherlands, $120 US or 250 Dutch guilders X-band SAR images on agricultural crops were analyzed, emphasizing the differences between two parallel polarizations HH and vv, Observed on cover tYDes at two dates. The most strikina feature is the sinoular behavior of flooded ricefields, which differ; from the early to fully growth stage. The behavior, explained by backscattering models oh vegetation canopy with highly reflecting undelying surface, Suggests the use of the polarization ratio between HH and VV for ricefield monitoring. ESA

..

TYPE 1/4/149 Quest Accession Number : 89N18710 89~1871on NASA STAR conference paper I~~~~ 11 Adaptive SDeCkle filterinq for SAR imaaes (AAiL0PES;A.i (AB)TOUZI,-R. Centre d'Etude Spatiale des Rayonnements, Toulou5e (France). (CK523228) In ESA. Proceedinas Of the 1988 International Geoscienrr and Remotcentury, voiume 3 p 1263-1266 (SEE ~89-18704 11-43) 886800 p . 4 I": EN (English) Avail: NTIS HC A99/M AOl; ESA Publications Division, ESTEC, Noordwijk, Netherlands, $120 US or 250 Dutch guilders p.1512 A generalized adaptive filter is developed Which can be adapted to preserve any kind of information characterized by the appropriate index. Filters are tested on a 4 look SAR 580 imaae. The best known adaotive filtets are optimized Such that they average well homogeneous areas and preserve edge and textural information (with less noise smoothing) at the Same time. However, such filters do not smooth well the noise in textured areas and much work must be done to develop filters which, by taking into account all the properties of the speckle, reduce it as well within homogeneous areas as in the textured ones without 1055 of information. ESA

TYPE 1 / 4 / 1 5 0 Quest Accession Number : 89A16980 89A16980 NASA IAA Journal Article I s s u e 04 Comparisons of Simulated and actual synthetic aperture radar gravity wave images (AA)HARGER, ROBERT 0.; (AB)KORMAN, CAN E (AB)(Maryland, University, College Park) Journal of Geophysical Research (ISSN 0148-0227). "01. 93, NO". 1 5 , 1988. D. 13867-13882. NaVV-SuUDOrted research. 8 8 1 1 1 5 D. 16 refs 23 . .. In: EN IEnglish) p.509 A series of SAR images obtained with an L band system during the Tower Ocean Wave and Radar Dependence experiment have been compared with simulations of an SA8 ocean imaging model based on two-scale hydrodynamic and electromagnetic scattering models. The best focus parameter is estimated using B subimage cross-correlation technique. Results show a magnitude increase with an increase in magnitude of the angle between the dominant long wave and the SAR a x e s , and illustrate the independence of the altitude and the range-to-velocity ratio. R.R.

c z

TYPE 1/4i151 Quest Acbession Number : 89A15915 89A15915* NASA IAA Journal Article Issue 04 Radar oolarimetrv - Analvsis tools and aoolications DIA E L.:. (ABIFARR, TOM G.: (ACIVAN ZYL, JAKOB J.; (AD)ZEBKER, HOWARD A. (AD)(California Institute of Technology, jet Propulsion Laboratory, Pasadena) Jet Propulsion Lab., California Inst. of Tech., Pasadena. (JJ57445O) IEEE Transactions on Geoscience and Remote Sensing (ISSN 0196-2892). "01. 26, NOV. 1988, p . 774-789. 881100 p . 16 refs 22 In: EN (English)

TYPE 1/4/15%

(AA)EVANS,

..

Quest Accession Number : 89A15914 89A15914 NASA IAA 30urnal Artlcle Issue 04 A statistical and geometrical edge detector for S A R images (AA)TOUZI, RIDHA; (AB)LOPES. ARMAND; (AC)BOUSQUET, PIERRE (AC)(Centre d'Etude Spatiale des Rayonnements, Toulouse, France) IEEE Transactions on Geoscience and Remote Sensing (ISSN 0196-2892), Vol. 26, NOV. 1988, p. 764-773. 881100 p. 10 refs 1 1 In: EN (English) p.528 A constant-false-alarm-rate (CFAR) edge detector based on the ratio between pixel values is described. The probability distribution of the image obtained by applying the edge detector is derived. Hence, the decision threshold can be theoretically determined for a given probability Of false alarm as a function of the number of looks of the image under study and the size of the processing neighborhood. For a better and finer detection, the edge detector operates along the four usual directions over windows of increasing sizes. A test performed, for a given direction, on a radar image of an agricultural scene shows good agreement with the theoretical study. The operator is compared with the CFAR edge detectors Suitable for radar images. I.E.

p.561

The authors have developed several techniques to analyze polarimetric radar data from the NASA/JPL airborne SAR for earth science applications. The techniques determine the heterogeneity of scatterers with subregions, optimize the return power from these areas, and identify probable scattering mechanisms for each pixel in a radar image. These techniques are applied to the discrimination and characterization of geologic surfaces and vegetation cover, and it is found that their utility varies depending on the terrain type. It is Concluded that there are several classes of problems amenable to single-frequency polarimetric data analysis, including characterization of Surface roughness and vegetation structure, and estimation of vegetation density. Polarimetric radar remote sensing can thus be a useful tool for monitoring a set of earth science parameters. I.E.

TYPE 1/4/15] Quest Accession Number : 891115913 89A15913* NASA IAA Journal Article IsSue 04

0 'TYPE 1/4/154 Quest Accession Number : 89N13033 89N13033# NASA STAR Conference PaDer Issue 04 Phase versus orbital velocity in SAK wave imaging: Paradox lost (AA)RANEY, R. K.: (ABIVACHON, P. W. Canada centre for Remote Sensing, Ottawa (ontario). (CE390486) RAUARSAT Project Office. In ESA. Proceedings Of the 1988 International Geoscii.nce and Remote sensing Symposium (IGARSS 1988) on Remote Sensing: Moving Towards the 21St Century, Volume 1 p 405-406 (SEE N89-12936 04-42) 880800 p. 2 In: EN (English) Avail: NTIS HC A99/M E03; ESA Publications Div., ESTEC, p.511 Noordwijk, Netherlands, 120 us dollars or 250 Dutch quilders The focus paradox in ocean wave SAR imaging from the air is reconciled. Improved wave imagery from an airborne SAR is possible by compensating individual looks (in a multilook data set) for wave movement prior to look summation. By observing the direction of wave motion between looks, the omnipresent 180 deg ambiguity (in wave direction estimation through spectral analysis) may be resolved using only the SAR data from o n e pass of the sensor. (There are known methods for resolution of the directional ambiguity for a single made sea using two opposed passes). Approximation OE the required image shift by focus adjustment i s not recommended because the azimuth impulse response is degraded in the process, the method by definition is tuned to only one wave component, and the resulting image shift is in the azimuthal direction only and thus not necessarily in the direction Of wave propagation. For directional spectral calculations, Fourier transformation of individual looks by magnitude summation leads to better results than the normal method of Fourier transformation Of the look summed wave image. These results do not depend on invocation of any particular wave imaging mechanism. ESA

illustrate the implementation of the procedure, two function. TO calibrated SAR images (X-band, 3.2-cm wavelength) are presented, along with the radar cross-section measurements of specific scenes within each image. The sources of error within the SAR image calibration procedure are identified. I.E.

TYPE 1/4/155 Quest Accession Number : 89N13032 89N13032*# NASA STAR Conference Paper Issue 04 Estimating aircraft SAR response characteristics and approximating ocean wave spectra in the Labrador Sea (AAITILLEY, D. G. Johns Hopkins Univ., Laurel, Md. (JS767253) Applied Physics Lab. In ESA, Proceedings of the 1988 International Geoscience and Remote Sensing symposium (IGARSS 1988) on Remote Sensing: Moving Towards the 21st Century, Volume 1 p 399-402 (SEE N89-12936 04-42) sponsored by NASA and the ONR 880800 p. 4 In: EN (English) Avail: NTIS HC A99/MF E O ] : ESA Publications Div., ESTEC, Noordwijk, Netherlands, 120 US dollars or 250 Dutch guilders p.511 The data processing methods employed to compute estimates of two-dimensional wave height-variance spectra from the ocean imagery obtained in the Labrador Sea by a C-band airhorne SAR system are described. The SAR Spectra are compared for high and low altitude geometries with large and small elevation angles. A Surface contour radar and a radar ocean wave spectrometer aboard an aircraft are used to verify the surface wave spectrum. ESA

TYPE 1/4/156 Quest Accession Number : 893313029 89~11029# NASA STAR conference Paper I S S U ~04 ComDlex sAR imaaerv and sneckle filterina for ERS-1 wave mode (AAiCORUEY, R. i.i-(ABlMAbKLIN, J. T. . GEC-Marconi Electronics Ltd., chelmsford (England). (GB135511) Tn ESA. Proceedinas O E the 1988 International Geoscience and Remote Sensina Svm~osium (IGARSS 1988) on Remote sensina: Movins Towards the 21st .. century, volume i p I ~ ~ - ~ ~ o ' ~89-12936 (sEE 04-42] ss0800 p. 4 In: EN (English) Avail: NTIS HC A99/MF E03; ESA Publications uiv., ESTEC, Noordwijk, Netherlands, 120 US dollars or 250 Dutch guilders p.510
~~~ ~

A method to predict the wavenumber dependence of the speckle component in spectra of SAR intensity images is described. Filtering of this component is an important step in recovering wave height spectra from SAR images of the ocean, and an effective means of doing so is required for the Wave mode of ERS-1. The method USeS the correlation function Of the corresponding complex images and was tested using airborne and spaceborne imagery obtained over land and sea. Examples of successful and unsuCCeSSfu1 applications of the method are shown. The successes show a great improvement in speckle filtering over previous techniques, while the failures can be explained in terms of artefacts of an individual S A R processor or too coarse a digitization of complex pixel amplitudes. ESA

TYPE 1/4/157 Quest Accession Number : 89N13021 89N13021# NASA STAR Conference Paper Issue 04 equations r for develooina Formulation of the ~ r o ~ e .~ standards in coherent dual polarisation SAR- imaging(AAIBOERNER, W . - M . : (AB)KOSTINSKI, A. 8 . Illinois Univ.. chicaao. lIB5254001 Communications Lab In ESA, Prociedings- of 'the 1988 International Geoscience and Remote Sensing Symposium (IGARSS 1988) on Remote Sensing: Moving Towards the 21St Century, Volume 1 p 351-353 (SEE N89..12936 04-42) 880800 p. 3 In: EN (English) Avail: NTIS HC A99/MF EO?: ESA Publications Div., ESTEC, p.509 Noordwijk, Netherlands, 120 US dollars or 250 Dutch guilders Crucial inconsistencies in the basic equations of radar polarimetry which a r e common in the literature were detected. The formulations of the polarization state definitions given in the IEEE/ANSI Standards 149-1979 are in error. These and other inconsistencies and conceptual errors are discussed. The correct formulae for the proposed revision of the polarimetric standards toqether with a well-defined and consistent Procedure for measuring target scattering matrices in monostatic and bistatic arrangements are given. The proposed procedure can be applied to an arbitrary measurement process in any general elliptical polarization basis. ESA

TYPE 1/4/158 Quest Accession Namber : 89N13020 89N13020*# NASA STAR Conference Paper Issue 0 4 The NASA/JPL multifrequency, multipolarisation airborne SAR System IAAIHELD. D. N.: IABIBROWN. W. E.; IACIFREEMAN. A,; IADIKLEIN. J. D.: (AE)ZEBKER; H. A,: (Af)SATO, T:: (AGIMILLER, T.; (AHINGUYEN; Q . : (AI)LOU,
Y.

Jet Propulsion Lab., California Inst. of Tech., Pasadena. (JJ574450) In ESA, Proceedings of the 1 9 8 8 International Geoscience and Remote sensing Symposium (IGARSS 1988) on Remote Sensing: Moving Towards the 215t Century, Volume 1 p 345-349 ( S E E N89-12936 04-42) 880800 p. 5 In: EN (English) ESA Publications Div., ESTEC, Avail: NTIS HC A99/MF EO); p.509 Noordwijk. Netherlands, 120 US dollars or 250 Dutch guilders
were

Polarimetric synthetic aperture radars, operating at L-, C- and P-band, dfsigncd to replace and upgrade a system destroyed in a n aircraft accident. Ground and flight tests were conducted, and the radar was flown over a calibration site in a sequence of experiments designed to calibrate the system. The radar also took part in science campaigns. ESA

/ : / I 5 9 TYPE : Quest Accession Number : 89N13018 89N13018*# NASA STAR Conference PaDBr Issue 04 Calibration of multipolarisation ima'ging radar (AAIFREEMAN, A.; (ABIWERNER, C.; (ACISHEN, YUHSHEN Jet Propulsion Lab., California Inst. Of Tech., Pasadena. (55574450) In ESA, Proceedings of the 1988 International Geoscience and Remote Sensing Symposium (IGARSS 1 9 8 8 ) on Remote Sensing: Moving Towards the 215t Century, Volume 1 p 335-339 (SEE N89-12936 04-42) 880800 p. 5 In: EN (English) Avail: NTIS HC A99/MF E03; ESA Publications Div., ESTEC, p.509 Noordwijk, Netherlands, 120 US dollars or 2 5 0 Dutch guilders
An experiment was designed to calibrate airborne imaging radar (AIR) images, in terms Of relative and absolute backscatter, and relative phase between Dalarization channels. The alibration uses measurements made the radar system itself (internal calibration) and using within ground-based corner reflectors and transponders (external calibration). The techniques developed for the AIR calibration campaign will form the basis of the calibration aporoach for SIR-C. ESA

'TYPE 1/4/160 Quest Accession3umber : 891113015 89N13015# NASA STAR Conference Paoer Issue 04 Phase calibration of polarimetric S h (AA)SHEEN, D. R.; (AB)KASISCHKE, E. S.; (ACISHUCHMAN, R. A. Environmental Research Inst. Of Michigan, Ann Arbor. (E0356283) Radar science Lab. In ESA, Proceedings Of the 1988 International Geoscience and Remote Sensing Symposium (IGARSS 1988) on Remote Sensing: Moving Towards the 21st Century, Volume 1 p 323-326 (SEE N89-12936 04-42) 880800 p. 4 In: EN IEnqlish) Avail: NTIS HC A99/MF E03: ESA Publications Div.. ERTFC. ~~~~. Nooedwijk, Netherlands, 120 us doilars or.250 Dutch guild&p.568 Techniques to measure the phase characteristics of a polarimetric imaging SAR were developed. Techniques to calibrate the phase portion of the radar signature from an airborne, polarimetric SAR system, targets (dihedral and trihedral corner reflectors) with known characteristics were deployed, and several passes of polarimetric data collected. The techniques used to reduce and analyze these data are presented. The radar-measured scattering matrix is compared to the theoretical scattering matrix. The results Of this calibration are discussed in the context of the rms phase errors achieved for the S A R System utilized in the experiment. ESA

TYPE 1/4/161 Quest Accession Number : 89N13014 89N13014*# NASA STAR Conference Paper Issue 0 4 Desian considerations for advanced multi-Dolarisation SAR (=)KLEIN, J. D.; (A6)FREEMAN. A. Jet Propulsion Lab., California Inst. of Tech., Pasadena. (JJ574450) In ESA, Proceedings of the 1988 International Geoscience and Remote sensina SvmDosium lIGARSS 19881 on Remote Sensinq: MoVinQ Towards the 21st Century, volume i p ~ I ~ - ~ ~ I ' (~89-1293s sEE 04-42) 880800 p. 5 I " : EN (English) Avail: NTIS HC A99/MF E03; ESA Publications Div., ESTEC, p.508 Noordwijk, Netherlands, 120 US dollars or 250 Dutch guilders
~

~.
~

TYPE 1/4/162 Ouest Accession Number : 89N12971 89N12971# NASA STAR Conference Paper Issue 04 SAR-seen multimode waves in ice: Evidence of imaging nonlinearities (AA)RANEY, R. K.; (AB)VACHON, P. W. Canada Centre for Remote Sensing, Ottawa (Ontario). (CE390486) RADARSAT Project Office. In ESA, Proceedings of the 1988 International Geoscience and Remote sensing Symposium (IGARSS 1988) on Remote Sensing: Moving Towards the 21st Century, volume 1 p 141-144 (SEE N89-12936 04-42] 880800 p. 4 In: EN (English) Avail: NTIS HC A99/MF E03; ESA Publications oiv., ESTEC, p.501 Noordwijk, Netherlands, 120 US dollars or 250 Dutch guilders
~~~~~

System design and verification of an airborne imaging radar (AIR) system Issues Of importance to multipolarization Systems (e-g., mutual coherence and amplitude balance between channels) are emphasized. Methods of detecting and correcting channel imbalances are discussed, and AIR test results are presented. ESA

are discussed.

A two dimensional wave field analytic model based on the synthetic aperture radar (SAR) Velocity bunching mechanism that is extended to the multi-modal case and includes the effects of wave component translation between looks is discussed. Using this model, directional spectra results are presented for a bimodal s e a . AS significant wave height is increased, the image spectra evolve from the correct bimodal farm through suppression of the correct modes to creation of a new and dominant Spectral artifact propagating at approximately 90 deg to the true wave direction. The simulated wave images compare favorably to actual imagery of waves in ice from the LIMEX/LEWEX 87 EXPERIMENT using similar radar, viewing geometry, and Wave parameters. It is Concluded that coherence time limitation is beneficial, as it expands the effective linear domain of the SAR imaging process. ESA

TYPE 1/4/163 Quest Accession Number : 89N12956 89N12956*# NASA STAR Conference Paper Issue 04 The effects of undersampling on multipolarization SAR images (=)FREEMAN, A,; (AB)DUBDIS, P. C.; (AC)KLEIN, J. D. Jet Propulsion Lab., California Inst. of Tech., Pasadena. (55574450) In ESA. Proceedings of the 1988 International Geoscience and Remote sensing Symposium (IGARSS 1988) On Remgte Sensing: Moving Towards the Z1St Centurv. volume 1 D 75-78 (SEE N89-12936 04-421 880800 p. 4 In: EN (EngliGh) Avail: HC ~ 9 9 1 E03; ~ ~ESA Publications biv., ESTEC, Noordwijk, Netherlands, 120 US dollars or 250 Dutch guilders p.498

TYPE 1/4/164 Quest accession Number : 89N12954 89N12954# NASA STAR Conference Faper Issue 04 of the oolarimetric matched Filter in hiQh resolutj-on the cnnrcnt .~~~~~ radar imaaina: An alternative for speckle radiation
~~~~

.~

~~~~

NTIS

sensina Svmoosium (IGARSS 19881 on Remote sensinq: Movinq Towards the 21-+ Century, volume i p 69-72 '(SEE ~89-1293604z42) asasoo p. 4 ~ n : EN (English) Avail: NTIS HC A99/MF E03; ESA Publications Div., ESTEC, Noordwijk, Netherlands, 120 US dollars or 250 Dutch guilders p.498
j-_

_ _

~~

The scattering matrix and Stokes matrix formats for multipolarization synthetic aperture radar (SARI images are introduced. The effects of converting to the Stokes matrix format, without first doubling the sampling rate to allow for the conversion from a complex to an intensity format is quantified and discussed. It is shown,that for most applications the Stokes matrix format is acceptable, slnce errors introduced by undersampling tend to average Out when large numbers of pixels are averageu. FOT those applications requiring analysis of single image pixels, the scattering matrix format is recommended. ESA

The potential of an exclusively polarimetric image filtering approach, i.e., filtering which takes full advantage Of the POL-SAF matrix data provided on a pixel-by-pixel basis and which complements the existing scalar Speckle reduction techniques, was assessed. The three stage polarimetric optimization procedure search for optimal polarizations on a pixel-by-pixel basis is combined with a subsequent statistical analysis of polarization eigenvectors (versus surface category), and the digital adjustment of polarimetric variables. Results are encouraging. In order to improve the efficiency of the polarimetric filtering method, it must be combined with other image processing and statistical communication theory techniques. Filter efficiency depends critically on peak sharpness of the histograms, and multilook incoherent averaging, block discretization, and nonuniform quantization can be used to imwove peakedness. ESA

TYPE 1/4/165 Quest Accession Number : 89N12953 89N12953# NASA STAR Conference Paper Issue 04 Statistical properties of phase diffe difference between cence betwee" two orthogonally-polarised SAR signals (AA)EOM, H. J.; (AB)BOERNER, W . - M . Illinois Univ., Chicago. (18525400) Dept. of Electrical Engineering and COmDUter Science. In EiA, Proceedinqs of the 1988 International Geoscience and Remote sensing symposium (IGARSS 1 9 8 8 1 on Remote sensing: Moving Towards the 21st Century, Volume 1 p 6 5 - 6 6 (SEE N89-12936 04-42) 880800 p. 2 s n : EN (Enqlishl Avail: NTIS HC A99lMF E03: SA Publications Div.. ESTFC. NooGdwijk, Netherlands, 120 US doilars or 250 Dutch guilders p.498
I

TYPE 1/4/1GG Quest Accession Number : 89A12173 09A12173 NASA S A A Journal Article Issue 0 7 Synthetic aperture radar imaging Of ocean wavcs from a n airborne platform Focus and tracking issues (AA)RANEY, R. K.; (AB)VACHON, P . W. (AA)(Radarset Project Office, Ottawa. Canadal: fABlICanada Centre f o r Remote sensing, ottaia) . ..

Journal

of

Geophysical

1988, p. 12475-12486.

881015

Research p. 12

(ISSN 0148-0227). "01. 9 3 , OCt. 1 5 , refs 19 In: EN (English) p.167

The theoretical behavior of two linearly polarized (VV and HH polarization states) radar backscattcred fields is examined Statistically. The coherency between two polarized signals is studied in terms of their statistical behavior on phase difference between two signals. The probability density function corresponding to the phase difference between two siqnals is derived and studied in terms O f the dearee Of oolarization. and the variance of each polarized signal. The. phase iifference is considered with respect to the polarimetric SAR data collected over rugged terrain and ocean. It is found that the phase coherency between two copolarized channels (VV and HH polarizations) strongly depends upon the degree of terrain roughness structure. The degree of polarization is also found to be closely related to terrain/oceanic surface roughness and anisotropy. ESA

This paper addresses the aspects of focus and tracking in the process of SAR imaging of ocean waves from an airborne platform. It is demonstrated that there is a direct relationship between focus and wave phase velocity, through purely noncoherent consequences of the SAR response to the translating reflectivity density envelope of the wave field. It is also Shown that the orbital velocity affects the phase of the received signal, leading to velocity bunching, and is scaled by the ratio of sensor altitude to sensor velocity. It is suggested that better performance can be obtained by compensating individual looks for wave movement before look summation, while using nominal perfect focus. 1,s.

TYPE 1/4/167 Quest Accession Number : 89A10938 89A10930*# NASA IAA Conference Paper Issue 01 SAR imaqe data compression for an on-line archive system (AAICHANG, C. Y.: (AB)KWOK, R.; (AC)CURLANDER, J. C. (AC)(california Institute of Technology, Jet Propulsion Laboratory,

TYPE 1/4/168 Quest Accession Number : 89N10315 89N10315# NASA STAR Conference Paper I s s u e 01 The use of the complex correlation function in the recovery of ocean wave spectra from SAR images (AAICORDEY, R. A,: (AB)MACKLIN, J. T. D llll_..", s.=lilnnal . Marconi CO. Ltd., Great BaddoW (England). (MF831696) Jet Propulsion Lab., California Inst. of Tech., Pasadena. (JJ574450) ESA-G878/87-HGE-I(SC) In ESA. Proceedings of the 4th International IN: International symposium on Remote sensing of Environment, ~ 1 s t .A"" colloquium on Spectral Signatures in Remote Sensing p 63-67 (SEE ~89-10305 Arbor, MS. OCt. 2G-30, 1987, Proceedinris. Volume 1 fA89-10926 0 1 - 4 3 1 . Ann 880400 p. 5 In: EN (English) Avail: NTIS HC A23/MF AOl; ESA ~~~~~. 01-43) Arbor, MI, Environmentel Research SnstiGute of Michigan, 1987. p. 171-182. Publications Division, ESTEC, Noordwijk, Netherlands 80 Dutch guilders 870000 p. I2 refs 10 In: EN (English) p.92 P.58
~~

This paper summarizes the investigation of SAR image data compression for an on-line archive data distribution system. This system is planned for the ground processing system of Alaska SAR Facility (ASF) and Shuttle Imaging Radar (SIR-c). The objective of the SAR image data compression is to enable the data archive system to provide the remote users a large data base with good image quality, short response time, low transfer cost, and minimal decoding complexity. The requirements and limitations of the on-line archive data distribution system are presented. The effects of SAR image data characteristics on data compression are addressed. The users' survey results suggest that compression ratios between 1 o : l and 20:l appear suitable. Based on the algorithm evaluation results, the two-level tree-searched vector quantization technique has been recommended as the SAR image data compression algorithm for the on-line archive data distribution system. Author

A method to predict the wavenumber dependence of the specKle component in spectra of synthetic-aperture radar intensity images was tested using data from VARAN-S and Sensat systems. The method uses tile correlation function of the corresponding complex images and assumes that pixel statistics are Gaussian. It is expected that such a technique will be of use in the routine recovery of ocean wave height spectra from SAR imagery. Results from the Agrisar campaign with VARAN-S over land and $88 are good, with Speckle spectra being well matched by their predicted forms. Ocean spectra from Seasat are, however, poorly matched in their dependence on azimuth wavenumber. This is not tholight ti. be c s ~ c o dby any sca -urface o r propagation effect hut rather to be an artifact of signal processing. ESA

TYPE 1/4/169 Quest Accession Number : 88A51680 88A53680* NASA IAA Journal Article ISSUe 21 Spatial compression of Seasat SAR imagery (AA)CHANG, C. Y.: (AB)KWOK, RONALD; (AC)CURLANDER, JOHN C. fACifCalifornia Institute of Technology, Jet Propulsion Laboratory, Pa;ad&a) Jet Propulsion Lab., California Inst. of Tech., Pasadena. (55574450) (IEEE, URSI, NASA, et a l . , I G A R S S '87 - International Geoscience and Remote Sensing Symposium, University of Michigan, Ann Arbor, May 18-21, 1987) IEEE Transactions on Geoscience and Remote Sensing (ISSN 0 1 9 6 - 2 8 9 2 ) , "01. 26, Sept. 1988, p. 671-685. 880900 p. 11 refs 25 I n : EN (English) p.1812 The results of a Study of techniques for spatial compression Of synthetic-aperture-radar (SAR) imagery are summarized. Emphasis is on image-data volume reduction for archive and online storage applications while oreservins the imaae resolution and radiometric fidelity. A quantitative analysis of various techniques, including vector quantization (VQ) and adaptive discrete cosine transform (ADCT), is presented. Various f i_ lrtn r a such as comoression ratio. . . . . . . .~alaorithm comoleritv. .. and imaae . quality . are considered in determininq the optimal alqorithm. The compression system requirements are established ?or electronic access of an online archive system based on the results of a survey of the science community. The various algorithms are presented and their results evaluated considering the effects of' speckle noise and the wide dynamic range inherent in SAR imagery. The conclusion is that although the ADCT produces the best signal-to-distortion-noise ratio for a given compression ratio, the two-level tree-searched VQ technique is preferred due to its simplicity Of decoding and near-optimal performance. I.E.
~~~~~ ~ ~

TYPE 1 / 4 / 1 7 0 Q u e s t Accession Number : 88A52625 88A52625* NASA IAA Journal Article I s s u e 2 1 Satellite radar interferometry - Two-dimensional phase unwrapping (AA)GOI.OSTEIN, RICHARD M.: (AB)ZEBKER, IIOWARD A.: (AC)WERNER, CHARLES L. (AC)(CaliCornia Institute of Technology, Jet Propulsion Laboratory, Pasadena) Jet Propulsion Lab., California Inst. of Tech., P a s a d e n a . (555744501 Radio Science (ISSN 0048-6604). "01. 23, July-AUg. 1988, p. 713-720. 880800 p. 8 refs 8 In: EN (English) p.3775 Interferometric synthetic aperture radar observations provide a means for obtaining high-resolution digital topographic naps from measurements of amplitude and phase of two complex radar images. The phase Of the radar echoes may only be measured modulo 2 p i ; however, the whole phase at each point in the image is needed to obtain elevations. An approach to 'unwrapping' th? 2 pi ambiguities i n the two-dimensional data set is presented. It 1s found that noise and geometrical radar layover corrupt measurements locally, and these local e r r o r s c a n propagate to form global phase errors that affect the entire image. It is Shown that the l o c a l errors, or residues, c a n be readily identified and avoided in the global phase estimation. A rectified digital topographic map derived from the unwrapped phase values is presented. Author

~~

~~~~~

TYPE 1/4/171 , , Quest Accession Number : 88A46766 88A46766* NASA IAA Journal Article ISSUe 19 M l l l t i s m a n r classification of sedimentary rocks (AAIEVANS, DIANE (AA)(California Institute of Technology, Jet Propulsion Laboratory, Pasadena1 Jet Propulsion Lab., California Inst. of Tech., Pasadena. (55574450) Remote Sensing of Environment (ISSN 0034-4257). "01. 25, July 19813, p. 129-144. 880700 p. 19 refs 29 In: EN (English1 P.3242
~~~~ ~~ ~ ~~ ~

TYPE 1/4/172 Quest Accession Number : REA44651 88~44651 NASA IAA Journal Article Issue 18 An effect of coherent scattering in spaceborne and airborne SAR images (AA)RANEY, R. K.; (AB)GRAY, A. L.; (AC)PRlNCZ, J. G. (AA)(Radarsat Project Office, Ottawa, Canada): (AC)(Canada Centre for Remote Sensing, Ottawa) International Journal of Remote Sensing (ISSN 0143-1161). VOl. 9, May p.3091 1988, p. 1039-1049. 880500 p. 11 refs 10 In: EN (En9lish)
~ ~~

A comparison is made between linear discriminant analysis and supervised classification results based on signatures from the Landsat TM, the Thermal Infrared Multispectral Scanner (TIMS), and airborne SAR. 81one and combined into extended spectral signatures for seven sedimentary rock units exposed on the margin of the Wind River Basin, Wyoming. Results from a linear discriminant analysis showed that training-area classification accuracies baaed on the multisensor data were improved an average of 15 percent Over TM a l o n e , 24 percent Over TIMS alone, and 4 6 percent Over SAR alone, with similar improvement resulting when supervised multisensor classification maps were compared to supervised, individual sensor classification maps. When training area signatures were used to map spectrally simil& materials in an adjacent area, the average classification accuracy improved 19 percent using the multisensor data over TM alone, 2 percent over TIMS a l o n e , and 1 1 percent over SAR alone. It is Concluded that certain sedimentary lithologies may be accurately mapped using a single sensor, but classification of a variety of rock types can be improved using multisensor data sets that are sensitive to different characteristics such as mineralogy and surface roughness. Author

Radar imagery obtained by the Shuttle Imaging Radar-B (SIR-B) is compared to high resolution aircraft imagery of the Same urban and agricultural areas close to the city of Montreal, Canada. It is clear that the SIR-B radar is more sensitive than the aircraft radar to reflections from extended, along-track radar targets. The effect is evident for both urban and agricultural areas in which the street or field orientation is near parallel to the radar azimuth direction. It is likely that such reflectivity enhancement is due to coherent combination of the scattered field from uDpropriate scattering centers. The paper considers the observed phenomena and probes potential scattering models to explain the results. Author

c:
P

'TYPE 1/4/17] Quest Accession Number : 88A42786 883142786 NASA IAA Conference Paper Issue 17 A basis for SAR oceanography: Theory and experiment (AA)HARGER, ROBERT 0 . ; (AB)KORMAN, CAN (AB)(Maryland, University, Colleqe Park) IN: IEEE National Radar Conference, 3rd. Ann Arbor, MI, Apr. 20, 21, 1988, Proceedings (A88-42751 17-32). New Yark, Institute of Electrical and Electronics Engineers, ~ n c . ,1 9 8 8 , p. 200-203. 880000 p. 4 refs 18 I " : EN (English) p.2904
A comparison Of synthetic-aperture radar (SAR) sea images generated by a simulation program implementing a two-scale theory and by the Jet Propulsion Laboratory L-band SAR flow in the TOWARD oceanographic experiment is discussed. A subimage cross-correlation technique estimates the best image focus. The sea Surface imagery is composed Of twoscales, a long-wave ensemble and a short-wave ensemble, and both gravity waves. The hydrodynamic and electromagnetic scattering appropriate two-scale approximate models are combined with a model for SAR imaging of a time-variant, extended scene. The resulting model is complicated and generally requires a simulation, which generates the complex, high-resolution SAR image to reveal its full nature. The two-scale theory's predictions agree well with the actual measurements, including the discriminating prediction that the best focus correction is proportional to the long wave's phase velocity.

'TYPE 1/4/174 Quest Accession Number : 88A42771 88A42'771 NASA IAA Conference Paper Issue 17 A statistical model for prediction of precision and accuracies of radar Scatterinq coefficient measurements derived from SAR data (AAIKASISCHKE, ERIC s.; (ABIFOWLER,GARY w.; (ACIWACKERMAN, CHRISTOPHER C. ( A C ) (1,lichigan. Environmental Research Institute, Ann Arbor) IN: IEEE National Radar Conference, 3rd, Ann Arbor, MI, Apr. 2 0 , 21, 1188, Proceedings (A88-42751 17-12). New York, Institute of Electrical and Electronics Engineers, Inc., 1 9 8 8 , p. 111-117. 880000 p. 7 refs 11 In: EN (English) p.2827

..

A model is presented to estimate a relative error-bound associated with radiometric calibration of the scattering coefficient derived from synthetic-ilpcrture radar (SAR) data. This error bound is Sescd on a Statistical c o e t f i c i e n t - o f - v a r i a t i o n error model. The error model w a s exercised parametrically to determine what factors most significantly influence radiometric errors from SAR systems. It was found that errors in the measurement of the antenna elevation gain-pattern resulted in the most dramatic increases in the statistical error bound. The results from this analysis can be utilized in both the design of SAR systems and measurements programs needed to calibrate SAR systems to minimize the error bounds associated with radiometric calibration. I.E.

I.E.

TYPE 1/4/175 Quest Accession Number : 88A42755 88A42755 NASA IAA Conference Paper Issue 17 CCRS C/x-airborne Synthetic aperture radar: An R and D tool for the ERS-1 time frame (AA)LIVINGSTONE, C. E.; (AB)GRAY, A. L.; (ACIHAWKINS, R. K.; (A0)OLSEN.
R.
~~

R.

(AD)(Canada Centre for Remote sensing, Ottawa) IN: IEEE National Radar Conference, 3rd. Ann Arbor, MI, Apr. 2 0 , 21, 1988, Proceedings (A88-42751 1 7 - 3 2 ) . New York, Institute of Electrical and Electronics Engineers, Inc., 1 9 8 8 , p. 15-21. 880000 p. 7 In: EN (English) p.2826 The airborne synthetic-aperture radar (SARl system developed for the Canada Centre for Remote sensing i s discussed. The SAR consists Of two radars at C- and X-bands. Each radar incorporates dual-channel receivers and dual-polarized antennas; a high quality, seven-look, real-time processor; a sensitivity time control for range-dependent gain control; a motion compensation system for antenna steering in azimuth and elevation; and baseband I- and Q-signal phase rotation. The system features a high-power transmitter with a low-power backup and can map to either side of the aircraft, at high or low resolution, at incidence angles which in high resolution span Of 0 to 80 deg. The radar operating parameters, data products, key specifications and the motion-compensation scheme used are given. Properties Of the real-time imagery are discussed and examples of C-baild rAR dsta in .-. L l l l F E u p z L a L A > , y III""S> aLe p ' C > r r r L r u . I.E.
LllF
^_._.

TYPE 1/4/176 Quest Accession Number : 88A42753 88A42753* NASA I A A Conference Paper Issue 17 Preliminary results from the NASA/JPL multifrequency, multipolarization Synthetic aperture radar (AA)HELD, DANIEL N.: (ABIBROWN, WALTER E.; (AC)MILLER, TIMOTHY w. (AC)(California Institute of Technology, Jet Propulsion Laboratory, Pasadena) Jet Prbpulsion Lab., California Inst. of Tech., Pasadena. (SJ5744501 IN: IEEE National Radar Conference, 3rd. Ann Arbor. M1,'Apr. 20,'21, 1988, Proceedings (A88-42751 17-32). New York, Institute of Electrical and 880000 p. 2 In: EN Electronics Engineers, Inc., 1988, p. 7 , 8. (English) p.2826 A brief description is given of the three-frequency polarimetric synthetic-aperture radar built and tested at the Jet Propulsion Laboratory. The radar has the capability to simultaneously acquire fully polarimetric data at P-, L-, and C-bands from a DC-8 airborne platform. The radar has an instantaneous data rate of approximately 2.1 Gb/S and has selectable record rates between 80-240 Mb/s. The system has a wide dynamic range featuring 8-b analog-to-digital converters and full digital floating-point processing. The processing is accomplished offline on a minicomputer system assisted by an array processor. Sample images are presented. I.E.

L:--

-.2..

L.

TYPE 1/4/177 Quest Accession Number : 88A37288 8821372881 NASA IAA Journal Article ISSUe 15 SAR swath-widening techniques (AAIWALTER, WINFRED: (ABIBRAUN, HANS MARTIN Dornier-Post (English Edition) (ISSN 0012-55631, no. 1, 1988, p. 24-27. 880000 p . 4 In: EN (English) p.2442 The possible ways of expanding the swath illuminated by SAR antennas used in space-based imaging encompass antenna lengthening, the positioning of two parallel beams, and the use of electronic beam steering in a phase-controlled active array. Attention is presently given to the comparative results of a study that considered these alternatives' mechanical, thermal, structural, and electrical requirements and characteristics. The passive antennas are found to be mechanically only insignificantly different from the active phased array antenna, in such matters as panel arrangement, launcher stowage, mechanical and thermal stability, and pointing accuracy. O.C.

TYPE 1/4/178 Ouest Accession Number : 88A27831 88A27831 NASA IAA Journal Article Issue 10 Simulation of bit-quantization influence on SAR images (AA)WOLFRX-M, A. P . : (AB)PIKE, T. K. (AB)(DFVLR, Institut fuer Hochfrequenztechnik, Oberpfaffenhofen, Federal Republic of Germany) (COSFAR, WMO, URSI, et al., Plenary Meeting, 26th, symposium 3, Workshop v, and Topical Meeting A2 on Remote Sensing from space, Toulouse, France, June 30-July 11, 1986) Advances in Space Research (ISSN 0273-11771, vol. 7, no. 11. 1987, D. 285-288. 870000 p . 4 refs 10 In: EN (English) p. 1546

The influence of two-bit and four-bit quantization schemes on the ocean wave spectra obtained in the wave imaging mode of the first European Remote Sensing Satellite ERS-1 is analyzed. The SAR images utilized were obtained through simulation using a static ocean-wave radar model and a comprehensive Software SAR system simulation model. The results indicate that spectra produced by the four-bit quantization are not significantly degraded from the optimum, but that the two-bit quantization requires Some gain adjustment for optimal spectral reproduction. The conclusions are supported by images and spectral plots covering the various options simulated. C.D.

TYPE 1/4/179 , , Quest Accession Number : 88A27829 88A27829 NASA IAA Journal Article ISSUe 10 DFVLAR'S intelligent SAR-processor - ISAR (AAINOACK, W.: (AB)RUNGE. IAAINOACK. (ABIRUNGE, H. (AB)(DFVLR, iABj (DFVLR, Obeipf8ffenhbfen. Oberpfaffenhofen, Federal Republic of Germany) (COSPAR, WMO, URSI, et al., Plenary Meeting, 26th, Symposium 3, Workshop A2.on Remote sensing Sensing from space, Space, T O oU ~~ lO oU uS ~~ e France, , , V, and Topical Meeting A2,on June 30-July 11, 1986) Advances in space Research (ISSN 0273-11771, vol. 7, no. 11, 1987, p. 273-279. 870000 p. 7 refs 8 In: EN (English) p. 1514 The fact that future SAR sensors like ERS-1 and X-SAR will be operational systems requires a processor system design which is significantly different from existing SAR COrrelatorS. Future systems require highest throughput and reliability. In addition, more attention must be paid to the user community needs in terms of various product levels and adequate production and organization schemes. This paper presents the design of the ISAR system which is identified by a distributed processor architecture using a high Speed array processor, enhanced by a two-dimensional accessible memory. a front-end processor and a knowledge engineering workstation. An expert system will support a human system operator for the mass production of SAR images and the detection and correction Of system malfunctions. As a result the System will be accessible and comprehensive for both experts and operators. Author

TYPE 1/4/180 Quest Accession Number : 88A27827 88A27827 NASA IAA Journal Article Issue 10 Radar calibration techniques including propagation effects (AAIHARTL, P H . ; (ABIKEYDEL, W.; (AC)KIETZM&NN, H.; (ADIHEEL, F. (AA)(Stuttgart, universitaet, Federal Republic of Germany); (AC)(DFVLR, Institut fuer Hochfrequenztechnik, Wessling, Federal Republic of Germany) (COSPAR, WMO, URSI, et al.; Plenary Meeting, 26th, Symposium 3 , Workshop V. and Topical Meeting A2 on Remote Sensing from space, To~louse, France, June 30-July 11, 1986) Advances in space Research (ISSN 0273-11771, vol. 7, no. 11, 1987, p. 259-268. 870000 p. 10 refs 13 In: EN (English) p.1514 This paper outlines the extent and difficulties associated with absolute calibration procedures of SAR systems. The calibration principles of SAR systems are reviewed, and the calibration concepts of several SAR systems are described. Accuracy considerations are addressed. C.D.

TYPE 1/4/181 Quest Accession Number : 88N26541 88N26541*# NASA STAR ISSUe 20 Data volume reduction for imaging radar polarimetry / Patent application (AAIZEBKER. HOWARD A . ; (ABjHELD, DANIEL N.: (ACIVANZYL, JAKOB J.;

(AE)(Jet Pripulsion'Lab., California Inst. bf Tech., Pasadena.) National Aeronautics and Space Administration. Pasadena Office, Calif. (ND894694) NASA-CASE-NPO-17184-1-CU: NAS 1.71:NPO-17184-1-CU; US-PATENT-APPL-SN-195225 NAS7-918 880405 p. 26 In: EN (English) Avail: NTIS HC A03/MF A01 p.2788
TWO alternative methods are presented for digital reduction of Synthetic aperture multipolarized radar data using scattering matrices, or using Stokes matrices, of four consecutive along-track pixels to produce averaged data for generating a synthetic polarization image. NASA

TYPE 1/4/182 Quest Accession Number : 88A23547 88~23547 Nasa I J O U ~ article ~ ~ I I S S U ~08 Interpretation of Seasat radar-altimeter data over sea ice using near-simultaneous SAR imagery (AA)ULANDER, LARS M. H. (M)(Chalmers Tekniska Hogscola, Goteborg, Sweden) ESA-6617/85/F/FL/(SC) International Journal of Remote Sensing (ISSN 0143-1161), "01. 8 , NOV. 1987, p. 1679-1686. Research supported by the Swedish Board for Space Activities. 871100 p. 8 refs 20 I": EN (English) p.0 The backscatter properties of Seasat altimeter data in the Beaufort Sea on October 3 , 1978 show distinct z o n e s , Which are interpreted in terms of geophysical characteristics. An overlapping and near-simultaneous synthetic-aperture radar image shows regions of open water, new ice, and multi-year sea ice which correspond to the different zones. It is found that the altimeter signal is sensitive to the ocean-ice boundary and that it indicates the ice type. The pulse-echo waveforms also Suggest that several scattering components are present in the returned power over sea ice. Author

TYPE 1/4/103 Quest Accession Number : 88N23357 88N23357 NASA STAR Technical Report Issue 16 Investisation of the imaqinq . . of ocean surface waves usinu a svnthetic aperture ;aaar UNTERSUCHUNG DER ABBILDUNG VON OZEANOBERFLAECHENWELLEN DURCH EIN SYNTHETIC APERTURE RADAR (AAIBRUENING, CLAUS Max-Planck-Inst. fuer Meteorologie, Hamburg (West Germany). (MN457760) SER-A-WISS-ABHANDL-84; ETN-88-91470 870000 p. 98 In: GM (German) Avail: Fachinformationszentrum Karlsruhe. . 7514 E ~ ~ e n S t e i n - L e 0 0 0 1 d s h ~ f ~ " 2, Fed. Republic Of Germany, 20 DM p.2236

TYPE 1/4/184 Quest Accession Number : 88A15372 88~15372 Nasa IM ;ourna: Article issue 04 Spectral properties of homogeneous and nonhomogeneous radar images (-)MADSEN, SOHEN NORVANG (AA)(Danmarks Tekniske Hojskole, Lyngby, Denmark) IEEE Transactions on aerospace and Electronic Systems ( r s s ~ 0018-9251), VOl. RES-23, July 1987, p. 583-588. 870700 p. 6 refs 13 In: EN (English) p.0 On the basis of a two-dimensional, nonstationary white noise model for the complex radar backscatter, the spectral properties of a one-look synthetic-aperture radar (SARI system is derived. It is shown that the power spectrum Of the complex SAR image is Scene independent. It is also shown that the spectrum of the intensitv imaue is in aeneral related to the radar Scene ipectrum by, a linear-inte&al equation, a Fredholmjs integral equation Of the third kind. Under simplifying assumptions, a closed-form equation giving the radar Scene spectrum as a function of the SAR image spectrum can be derived. Author

_.
~

~~~

The imaging Of long, wind-induced Ocean waves by a SAR using a two-dimensional Monte Carlo simulation model was investigated. It is shown that the nonlinear distortions in the SAR-imaging are essentially due to phase modulation and Doppler broadening of the SAR-signal by the subscale orbital velocities of backscattered SAR-Signal. The pronounced azimuthal decay Of the Spectral energy density in the saR-variance Spectrum is due to the Doppler-broadening-induced reduction of the azimuthal resolution, and limits the signal-to-noise ratio of the signal. The validity of the sAR-simulation model was demonstrated by Comparison with SAR-variance spectra taken by the Shuttle Imaging Radar. ESA

TYPE 1/4/185 Quest Accession Number : 88N15284 88N15284*# NASA STAR Technical Report Issue 07 SAR [Synthetic ADerture Radar). Earth observing system. VOlUme 2F: . . Instrument panel report National Aeronautics and Space Administration. Goddard Space blight Center, Greenbelt, Md. (NC999967) NASA-TM-89701; NAS 1.15:89701 870000 p. 260 Original document contains color illustrations In: EN (English) Avail: NTIS HC A12/MF A01 p.871 The scientific and engineering requirements for the Earth Observing System (EOS) imaging radar are provided. The radar is based on Shuttle Imaging Radar-C (SIR-C), and Would include three frequencies: 1.25 GHZ, 5.3 GHZ, and 9 . 6 GHZ; selectable polarizations for both transmit and receive channels; and selectable incidence angles from 15 to 5 5 deg. There would be three main viewing modes: a local high-resolution mode with typically 25 m resolution and 50 km swath width; a regional mapping mode with 100 m resolution and up to 200 km swath width; and a global mapping mode with typically 500 m resolution and up to 700 km s w a t h width. The last mode allows global covecage in three days. The EOS SAR will be the first orbital imaging radar to provide multifrequency, multipolarization, multiple incidence angle observations of the entire Earth. Combined with Canadian and Japanese satellites, continuous radar observation capability will be possible. Major applications in the areas of glaciology, hydrology, vegetation science, oceanography, geology, and data and information systems are described. J.P.B.

REPORT DOCUMENTATION PAGE

~~

1. Recipient's Reference

4 . Security Classific

of Document

AGARD-LS-182
5 . Originator

ISBN 92-835-0683-9

UNCLASSIFIE
~~

Advisory Group for Acrospacc Research and Deve opmcnt 1 I & North Atlantic Trcaty Organization 7 ruc Anccllc, 92200 Neuilly sur Scinc, Francc FUNDAMENTALS AND SPECIAL PROBLEMS 1 18 OF SYNTHETIC APERTURE RADAR ( S A R )

6 . Title

-~

7.presented0n

5th-6th Octobcr 1992 in Rad Neuenahr, Germany, 8th-9th Octobcr 1992 in Gebze-Kocaeli (ncar Istanbul), Turkey and 26th-27th October 1992 in Ottawa, Canada Various

8. Author(s)/Editor(s)

-~

August 1992
-~

IO. Aulhor's/Editor's Address

Various
I

198 This document is distributed in accordance with AGARD policies and rcgulations, which are outlincd on the back covers of all AGARII publications.

12. Distribution

13. Keywords/Descriptors

Airborne radar Algorithms Digital techniques Inverse synthetic aperture radar Polarization (waves)
14. Abstract
1 '

Remote sensing Simulation Spaceborne equipment Synthetic aperture antennas Synthetic aperture radar

lThc Lccturc Series killkwcrthc field of airborne and spaceborne SAR with respect to its . 'technical realisation in order to convey the participants' ideas and know-how on SAR, on its capabilities and on the technology necessary for the successful construction and application of airborne and spaceborne SAR systems.
'

,., , ,
,'

,~

.The basic principles of SAR will be.cxplaincd and SARwill be'comparcd to airborne and spaccbornc radar with real aperture.

Thc influcncc of the antenna parametFrs on specification and capabilities of SAR and the advantages, necessities and limits will,bbe considered.
I

Digital SAR processing is indispensable for SAR. Theories and special algorithms will be'given' &ng with basic processor configurations and different processing techniques on a hardware an software basis. The simulation of SAR-systcms as well as SAR-products 'willalso bela topiobf the Lecture Serif A presentation of the present state of the art, giving examples of presently planned and realised airborne and spaceborne SAR with its foreseen applications bill~'conclude.the Lecture Series This Lecture Series, sponsored by the Avionics Panel of AGARD, has been implementcd by thr Consultant and Exchange Programme.

3.

AGARD Lectureseries 182 Advisory Group for Aerospace Research and Development, NATO FUNDAMENTALS AND SPECIAL PROBLEMS O F SYNTHETIC APERTURE RADAR (SAR) Published August 1992 198 pages The Lecture Series will cover the field of airborne and spaceborne SAR with respect to its technical realisation in order to convey the participants ideas and know-how on SAR, on its capabilities and on the technology necessary for the successful construction and application of airborne and spaceborne SAR systems. The basic principles of SAR will be explained and SAR will be compared to airborne and spaceborne radar with real aperture. PT.0. AGARD Lecture Series 182 Advisory Group for Aerospace Research and Development, NATO FUNDAMENTALS AND SPEClAL PROBLEMS O F SYNTHETIC APERTURE RADAR (SAR) Published August 1Y92 198 pages The Lecture Series will cover the field of airborne anc spaceborne SAR with respect to its technical realisation ir order to convey the participants ideas and know-how or SAR, on its capabilities and on thc technology necessar) for the successful construction and application of airbornf and spaceborne SAR systems. The basic principles of SAR will be explained and SAF will be compared to airborne and spaceborne radar wit1 real aperture.

AGARD-LS-182

Airbornc radar Algorithms Digital techniques Inverse synthetic aperture radar Polarization (waves) Remote sensing Simulation Spacebornc equipmcnt Synthetic aperture antennas Synthetic aperture radar

AGARD Lecture Series 182 Advisory Group for Aerospace Research and Development, NATO FUNDAMENTALS AND SPECIAL PROBLEMS OF SYNTHETIC APERTURE RADAR (SAR) Published August 1992 198 pages The Lecture Series will cover the field of airborne and spaceborne SAR with respect to its technical realisation in order to convey the participants ideas and know-how on SAR, on its capabilities and on the technology necessary for the successful construction and application of airborne and spaceborne SAR systems. The basic principles of SAR will be explained and SAR will be compared to airborne and spaceborne radar with real aperture. P.T.O.

AGARD-LS-I82

kirborne radar 9lgorithms ligital techniques nverse synthetic aperture radar Polarization (wavcs) Remote sensing Simulation Spaceborne equipment Synthetic aperture antennas Synthetic aperture radar

AGARD-LS-IXZ

Airborne radar Algorithms Digital techniques Inverse synthctic aperture radar Polarization (waves) Remote sensing Simulation Spaceborne equipment Synthetic aperture antennas Synthetic aperture radar

AGARD Lecture Series 182 Advisory Group for Aerospace Research and Development, NATO FUNDAMENTALS AND SPECIAL PROBLEMS OF SYNTHETIC APERTURE RADAR (SAR) Published August 1992 198 pages The Lecture Serics will cover the field of airborne and spaceborne SAR with respect to its technical realisation in order to convey the participants ideas and know-how on SAR, on its capabilities and on the technology necessary for the succcssful construction and application of airborne and spaceborne SAR systems. The basic principles of SAR will be explained and SAR will be compared to airborne and spaceborne radar with real aperture.

AGARD-LS-182

Airborne radar Algorithms Digital techniques Inverse synthetic aperture radar Polarization (waves) Remote sensing Simulation Spaceborne equipment Synthetic aperture antennas Synthetic aperture radar

PT.0

PT.0

The influence of the antenna parameters on specification and capabilities of SAR and the advantages, necessities and limits will be considered. Digital SAR processing is indispensable for SAR. Theories and special algorithms will be :iven along with basic processor configurations and different processing techniques o n a iardware and software basis. The simulation of SAR-systems as well as SAR-products will also bc a topic o f the Lecture Series. A presentation of the present statc of the art. giving cramples of presently planncd md realised airborne and spaceborne SAR with its foreseen applications will conclude the Lecture Series. This Lecture Series, sponsored by the Avionics Panel of AGARD and the Consultant and Exchange Programme of AGARD, presented on 5th-6th October I992 in Bad Neuenahr, Germany, 8th-9th October 1992 in Gebze-Kocaeli (near Istanbul), Turkey and 26th-27th October 1992 in Ottawa, Canada.

The influence of the antenna parameters on specification and capabilities of SAK and the advantages, necessities and limits will be considered, Digital SAR processing is indispensablc for SAR. Theories and special algorithms will be given along with basic processor configurations and different processing techniques on a hardware and software basis. The simulation of SAR-systems as well as SAK-products will also be a topic of thcLecturc Series. A prcscntation of the present statc of the art, giving examples of presently planned and realised airborne and spaceborne SAR with its foreseen applications will conclude the Lecture Series. This Lecture Series, sponsored by the Avionics Panel of AGARD and the Consultant and Exchange Programme of AGARD, presented on 5th-6th October 1992 in Bad Neuenahr, Germany, 8th-9th October 1992 in Gebze-Kocaeli (near Istanbul), Turkey and 26th-27th October 1992 in Ottawa, Canada.

ISBN 92-835-0683-9 The influence of the antenna parameters on specification and capabilities of SAR and the advantages, necessities and limits will be considered. Digital SAR processing is indispensable for SAR. Theories and special algorithms will be given along with basic proccssor configurations and different processing techniques on a hardware and software basis. The simulation of SAR-systems as well as SAR-products will also be a topic of the Lecture Series. A prcscntation of thc prescnt state of the art. giving examples of presently pkanned and realised airborne and spaceborne SAR with its foreseen applications will conclude the Lecture Series. This Lecture Series, sponsored by the Avionics Panel of AGARD and the Consultant anc Exchange Programme of AGARD, presented on 5th-6th October 1992 in Bac Neuenahr, Germany, 8th-9th Octobcr I992 in Gebze-Kocaeli (near Istanbul), Turke) and 26th-27th October 1992 in Ottawa, Canada.

ISBN 92-835-0683-9 The influence of the antenna parameters on specification and capabilities of SAR and the advantages, necessities and limits will he considered, Digital SAR processing is indispensable for SAR. Theories and special algorithms will be given along with basic processor configurations and different processing techniques on a hardware and software basis. Thc simulation of SAR-systems as well as SAR-products will also be a topic of the Lecture Serics. A presentation of the present state o f the art, giving examples of prcscntly planned and realised airborne and spaceborne SAR with its foreseen applications will conclude the Lecture Series. This Lecture Scries. sponsored by the Avinnics Panel of AGARD and the Consultant and Exchange Programme of AGARD, presented on 5th-6th October 1992 in Bad Neuenahr. Germany, Xth-gth October I992 in Gebzc-Kocaeli (near Istanbul), Turkey and 26th-27th October 1992 in Ottawa, Canada.

ISBN 92-835-0683-9

ISBN 92-X15-06X1-Y

L " m
NATO

-4-

OTAN

7R U E ANCELLE .92200 NEUILLY-SUR-SEINE

FRANCE
Telephone (1)47.38.57.00 ' Telex 610 176 TBlPcopie (1)47.38.57.99

DIFFUSION DES PUBLJCATIONS AGARD NON CLASSIFIEES

L'AGARD ne ditient pas de stocks de ses publications, dans un but de distribution genirale a I'adresse ci-dessus. La diffusion initiale des publications de I'AGARD est effectuie aupris des pays memhres de cette organisation par I'intermediaire des Centres Nationaux de Distribution suivants. A I'exception des Etats-Unis, ces centres disposent parfois d'exemplairesadditionnels; dans les cas contraire,on peut se procurer ces exemplaires sous forme de microfiches ou de microcopies aupris des Agences de Vente dont la liste suite. ALLEMAGNE Fachinformationszentrum, Karlsruhe 0-75 14 Eggenstein-Leopoldshafen2 BELGIQUE Coordonnateur AGARD-VSL Etat-Major de la Force Airienne Quanier Reine Elisaheth Rue d'Evere, 1140 BNxelles CENTRES DE DIFFUSION NATIONAUX ISLANDE Director of Aviation c/o Flugrad Reykjavik ITALIE Aeronautica Militare Ufficiodel Delegato Nazionale alI'AGARD Aeroporto Pratica di Mare 00040 Pomezia (Roma) LUXEMBOURG VoirBelgique NORVEGE Norwegian Defence Research Establishment Attn: Bihlioteket P.O. Box 25 N-2007 Kjeller PAYS-BAS Netherlands Delegation to AGARD National Aerosoace Lahoratorv NLR Kluyverweg 1 2629 HS Delft PORTUGAL Portuguese National Coordinator to AGARD Gahinete de Estudos e Programas CLAFA Base de Alfragide Alfraeide 2700h"ora ROYAUME UNI Defence Research Information Centre Kentigern House 65 Brown Street Glasgow G2 8EX

CANADA Directeur du Service des Renseignements Scientifiques Ministire de la Defense Nationale Ottawa, Ontario K1A OK2
~~~~ ~~~~~

DANEMARK Danish Defence Research Board Ved Idraetsparken 4 ESPAGNE INTA AGARD Publications) Pintor osales 34 28008 Madrid

ETATS-UNIS National Aeronautics and Space Administration Langley Research Center M/S 180 Hampton, Virginia 23665 FRANCE O.N.E.R.A. (Direction) 29, Avenue de la Division Leclerc 92322 Chitillon Cedex

GRECE Hellenic Air Force Air War College TURQUIE Scientific and Technical Library Milli Savunma Ba$kanl@ MSB Dekelia Air Force Base ARGE Daire Baskanhgi ( RG ) Dekelia,AthensTGA 1010 Ankara LE CENTRE NATIONAL DE DISTRIBUTION DES ETATS-UNIS (NASA) NE DETIENT PAS DE STOCKS DES PUBLICATIONS AGARD ET LES DEMANDES DEXEMPLAIRES DOIVENT ETRE ADRESSEES DIRECTEMENT AU SERVICE NATIONAL TECHNIQUE DE L'INFORMATION (NTIS) DONT L'ADRESSE SUIT.

a 2

AGENCES DE VENTE ESA/Information Retrieval Service The British Library European Space Agency Document Supply Division $,i%xt Royal Road 10, rue Mario Nikis Boston Spa, Wetherby Springfield, Virginia 22161 75015 Paris West Yorkshire LS23 7BQ France Royaume Uni Etats-Unis Lesdemandesdemicrofichesoudephotocopies dedocuments AGARD (y comprisles demandesfaitesaupresduNT1S)doiventcomporter la dinomination AGARD, ainsi que lenumiro de sine de I'AGARD (par exemple AGARD-AG-315 Des informations analogues, telles que le titreet ladatedepuhlication sontsouhaitables.Veuillernoterqu il yalieude spicifier AGARD-2-nnn et AGARD-AR-nnnlors de la commandederapports AGARD et desrapports consultatifs AGARDrespectivement.Des riferences bibliographiquescompl~tes ainsi que des resumts des publications AGARD figurent dans les joumaux suivants: Scientifique and Technical Aerospace Reports (STAR) Government Reports Announcements and Index (GRA&I) puhlie par la NASA Scientific and Technical puhlie par le National Technical Information Serwce Information Division Springfield NASA Headquarters NIT) Virginia 22161 Washington D.C. 205 6 Etats-Unis Etats-Unis (accessible igalement en mode interactif dans la base de donnees bibliographiques en ligne du NTIS, et sur CD-ROM) National Technical Information Service

Impriml par Specialised Printing Services Limited 40 Chigwell Lane, Loughron, Essex IGlO 3Tz

NATO

OTAN

7 RUE ANCELLE. 92200 NEUILLY-SUR-SEINE

FRANCE
Telephone (1)47.38.57.00 Telex 610 176 Telefax (1)47.38.57.99

DISTRIBUTION OF UNCLASSIFIED AGARD PUBLICATIONS

AGARD does NOT hold stocks of AGARD publications at the above address for general distribution. Initial distribution of AGARD publications is made to AGARD Member Nations through the fallowing National Distribution Centres. Further copies itre sometimes available from these Centres (except in the United States), but if not may he purchased in Microfiche or Photocopy form from the Sales Agencies listed below. NATIONAL IXSTRIBUTION CENTRES BELGIUM LUXEMBOURG See Belgium Coordonnateur AGARD - VSL Eta-Majorde la Force Airienne NETHERLANDS Quartier Reine Elisabeth Netherlands Delegation to A G A W Rue d'Evcre. 1140 Bruxelles National Aerospace Laboratory, NLR Kluyvenveg I CANADA . .2629 HS Delft Director Scientific Inionnation Services Dept of National Defence NORWAY Ottawa. OntarioKIA OK2 Norwegian Defence Research Establishment Attn: Biblioteket DENMARK P.O. Box 25 Danish Defence Research Board N-2007 Kjeller Vcd ldractsparken 4 2100 Copenhagen 0 PORTUGAL Portuguese National Coordinator to AGARD FRANCE Gabinete de Estudos e Programas O.N.E.R.A. (Direction) CLAFA 29 Avenue de la Division Leclerc Base de Alfragide 92322 Chitillon Cedex Alfragide 2700 Amadora GERMANY Fachiniormationszentrum SPAIN Karlsruhc INTA AGARD Publications) D-75 14 Eggenstein-Leopoldshafen2 Pintor osales 34 28008 Madrid GREECE Hellenic Air Force TURKEY Air War College Milli Savunma Ba$kanliii (MSB) Scientific and Technical Library ARGE Daire Bagkanldi . . .(ARGE) Dekelia Air Force Base Ankara Dekelia,AthensTGA 1010 UNITED KINGDOM ICELAND Defence Research Information Centre Director of Aviation Kentigern House c/o Flugrad 65 Brown Street Reykjavik Glasgow G2 8EX
~

UNITED STATES ITALY National Aeronautics and Space Administration (NASA) Aeronautica Militare Lanelev Research Center Ufficio del Deleeato Nazionale all'AGARD M/S 1$0 Aeroporto Pratca di Mare Hampton, Virginia 23665 00040 Pomezia (Roma) THE UNITED STATES NATIONAL DISTRIBUTION CENTRE (NASA) DOES NOT HOLD STOCKS OF AGARD PUBLICATIONS, AND APPLICATIONS FOR COPIES SHOULD BE MADE DIRECT TO T H E NATIONAL TECHNICAL INFORMATION SERVICE (NTIS) AT THE ADDRESS BEL'OW. SALES AGENCIES National Technical Information Service (NTIS) 5285 Port Royal Road Springfield, Virginia 22 16 I United States ESA/lnformation Retrieval Service European Space Agency 10, rue Mario Nikis 75015 Paris France The British Library Document Supply Centre Boston Spa, Wetherby West Yorkshire LS23 7BQ United Kingdm

Requests for microfiches or photocopies of AGARD documents (including requests to NTIS) should include the word'AG/\RD'and the AGARD serial number (for example AGARD-AG-3 15 Collateral information such as title and publication date is desirable. Note that bihlioeraohical AGARD Reportsand Advisory Reportsshould bespecifledasACARD-R-nnnandAGARD-AR-nnn,respectively.Full '. references and abstracts of AGARD publications are eiven in the fdn.-:-Scientific andTrchnical A------' " . . . . published by NA: Information Divir NASA Headquar Washington D.C. United Statcs

.. . ..

.......*

JcI

v c e s umiled

40 Chigwei Lane, Luughtun, Essex IG103TZ