Professional Documents
Culture Documents
The Delay/Doppler Radar Altimeter: R. Keith Raney
The Delay/Doppler Radar Altimeter: R. Keith Raney
5, SEPTEMBER 1998
Abstract— The key innovation in the delay/Doppler radar al- Pulse-limited operation necessarily implies that conven-
timeter is delay compensation, analogous to range curvature cor- tional altimeters are relatively wasteful of radiated power. For
rection in a burst-mode synthetic aperture radar (SAR). Follow- example, the footprint diameter of GEOSAT over a quasiflat
ing delay compensation, height estimates are sorted by Doppler
frequency, and integrated in parallel. More equivalent looks are surface is less than 1/10 of the antenna pattern within the
accumulated than in a conventional altimeter. The relatively half-power width. Hence, most of the radiated power falls
small along-track footprint size is a constant of the system, outside of the pulse-limited area and cannot be used for
typically on the order of 250 m for a Ku-band altimeter. The height estimation. Other disadvantages of conventional radar
flat-surface response is an impulse rather than the more familiar
step function produced by conventional satellite radar altimeters. altimeters include footprint dilation over rougher terrain, and
The radar equation for the delay/Doppler radar altimeter has the tendency of the footprint location to hop from one elevated
an h05=2 (c )1=2 dependence on height h and compressed pulse region to another rather than to trace out the elevation profile
length , which is more efficient than the corresponding h03 c without negative influence from the topography. Footprint
factor for a pulse-limited altimeter. The radiometric response dilation leads to less than optimal estimation of surface height
obtained by the new approach would be 10 dB stronger than
that of the TOPEX/Poseidon altimeter, for example, if the same and roughness.
hardware were used in the delay/Doppler altimeter mode. This Doppler beam-sharpened (DBS) altimeters have been pro-
new technique leads to a smaller instrument that requires less posed as a means of reducing the along-track footprint size of a
power, yet performs better than a conventional radar altimeter. radar altimeter. The performance of DBS altimeters has proven
The concept represents a new generation of altimeter for earth
to be very disappointing. The main reason is that relatively
observation, with particular suitability for coastal ocean regions
and polar ice sheets as well as open oceans. few “looks” are available, virtually eliminating most of the
incoherent averaging that is essential to precision altimetric
Index Terms—Doppler beam sharpening, radar altimeter, syn-
thetic aperture radar.
measurements. The Johns Hopkins University Applied Physics
Laboratory (JHU/APL) delay/Doppler approach is fundamen-
tally different; there is far more averaging in the delay/Doppler
I. INTRODUCTION approach than is possible from a DBS altimeter.
The principal objectives of the delay/Doppler altimeter [12],
T HE PRINCIPAL objective of a satellite radar altimeter
is to measure the height of reflecting facets scanned by
passage of the instrument overhead. Geophysical elevation
[13] are to operate more efficiently and more effectively. The
first objective is met by compensating for systematic range
is derived from the record of radar height measurements, delay errors; thus, the entire (beam-limited) along-track signal
corrected with respect to precise orbit knowledge and path history contributes to height measurement rather than only
delays. Height precision is set by the radar pulse length and the much smaller pulse-limited area. Stated another way, the
by the amount of averaging available for each estimate. Height delay/Doppler altimeter uses much more of the instrument’s
is defined in the context of this paper as the minimum range radiated energy than does a conventional beam-limited altime-
between the radar and scatterers that lie along the ground track ter. The second objective is met by using Doppler selectivity
of the satellite. to reduce the width of the postprocessing along-track footprint;
By definition, a conventional satellite altimeter [Fig. 1(a) this minimizes unwanted terrain dependency of the footprint
and (b)] uses the echo delays from within the pulse-limited size and position.
footprint to estimate minimum radar range [8]. Outside of the The delay/Doppler altimeter uses pulse compression in the
pulse-limited footprint, each scatterer’s echo appears at rela- range dimension, just as is customary for incoherent radar
tively greater delay. The (compressed) pulse length determines altimeters [5]. The range signal is a long linearly frequency
the diameter of the pulse-limited footprint associated with a modulated (FM) pulse, which is multiplied by a delayed replica
quasiflat surface [2]. For a typical radar altimeter, such as FM pulse immediately upon reception, and low-pass filtered.
GEOSAT, the pulse-limited footprint is on the order of 2 km The delay is chosen to match the expected range to the mean
in diameter [6], expanding to many kilometers as large-scale reflecting surface, the so-called track point. This “deramp”
surface roughness increases. strategy transforms “range” into a continuous wave (CW)
signal whose frequency is proportional to height, relative to
the track point [7], [15], [16]. A conventional altimeter and the
Manuscript received April 1, 1997; revised October 27, 1997. delay/Doppler altimeter both complete range compression by
The author is with The Johns Hopkins University Applied Physics Labora-
tory, Laurel, MD 20723-6099 USA (e-mail: keith.raney@jhuapl.edu). application of an inverse Fourier transform (IFFT) to convert
Publisher Item Identifier S 0196-2892(98)04976-6. the CW signals into height. These are summed to produce
0196–2892/98$10.00 1998 IEEE
RANEY: DELAY/DOPPLER RADAR ALTIMETER 1579
(a) (c)
(b) (d)
Fig. 1. Comparison of a conventional pulse-limited radar altimeter’s (a) illumination geometry side view and (b) footprint plan view, to a delay/Doppler
altimeter’s (c) illumination geometry side view and (d) footprint plan view. The conventional radar altimeter measurement space is inherently 1-D, whereas
that of the delay/Doppler altimeter is 2-D.
Fig. 3. Each individual range history is hyperbolic, and for a given altitude,
its shape is known very well. Delay offset r can be calculated readily if
the scatterer’s (along-track) position is known. Unfortunately, the problem is (3)
multivalued in the range time/along-track position domain and, hence, admits
no unique solution.
Using the same numbers as before, the difference in range
delay for a 1-km change in reference height , say from 800
objective of range delay compensation. The desired compen-
to 801 km, is only 45 1/800 cm. In similar fashion,
sation cannot be applied directly, as the range delays in the
it may be shown that the change in relative delay due to the
2-D signal domain are multivalued. Following the along-track
orbital factor is much less than 1 mm. It follows that
FFT’s, however, data in the resulting range (delay)/along-
an acceptable range delay estimate can be derived with only
track frequency (Doppler) domain are single-valued, as derived
an approximate knowledge of the scatterer’s height, and this
in Section III. Range delay compensation can be applied in
value may be held constant over the width of the range gate
this domain. Section IV introduces the new 2-D footprint
and over the duration of the azimuth signal.
of the delay/Doppler approach and offers comments on its
Delay compensation implies that the extra delay is removed
salient features. Section V describes the flat surface response
from the signal over its history. Given that this unwanted delay
function of the delay/Doppler altimeter and contrasts it with
is known, and that the signal locus is known, the extra delay
that from a conventional radar altimeter. Section VI derives the
could be compensated completely by subtracting an equivalent
new radar equation applicable to the delay/Doppler altimeter,
amount from the indicated range at all points on the along-
including its substantial gain advantage when compared to
track range history. This would lead to the range locus having
a conventional pulse-limited altimeter. Section VII outlines
a constant effective height at all observation opportunities.
one method of range gate and pulse timing for a practical
Unfortunately, it is not possible to apply the delay com-
spacecraft instrument. The paper closes with comments and
pensation scheme directly to the received data because to do
conclusions.
so would require knowledge of the along-track position of
each scatterer. Even more frustrating, the signal loci from more
II. DELAY COMPENSATION
than one scatterer are present in each echo ensemble. Echoes
The key to delay/Doppler processing is suggested by the are received simultaneously from all individual scatterers
range histories sketched in Fig. 3. The curves describe the within the illuminated beam width of the antenna, which spans
relative delay suffered by the signal loci at all points away from the observation space of the system. Three representative range
minimal radar range. The objective of delay compensation is to histories are shown in Fig. 3. Clearly, at any position in the
eliminate the unwanted extra delay, compared to the minimum data field, the delay compensation problem is multiple valued,
range measurement, suffered by the round-trip signal at all for which a single delay compensation is impossible in the
other along-track altimeter positions. After compensation, the range/along-track domain.
range indicated for each scatterer over its entire illumination
history is equal to its minimum range. The relative delay
III. DELAY/DOPPLER DOMAIN
for the th scatterer, given the altimeter along-track position
, and the scatterer’s along-track position and (minimal) Transformation from the along-track signal domain to the
height , is along-track frequency (Doppler) domain reduces delay com-
pensation to a single-valued problem. Recall that the Fourier
(1) shift theorem [17] states that application of a Fourier transform
to a function with position shift leads to the Fourier
transform of the function, multiplied by a CW phase term
where the orbital factor is given by
whose frequency is proportional to the shift. Let the along-
(2) track signal history for each scatterer be . Let be
the nadir position of the altimeter and be the along-track
RANEY: DELAY/DOPPLER RADAR ALTIMETER 1581
is proportional to (minimum) delay relative to the range will add Doppler shift to the signals, which if not compensated
track point, and frequency in the along-track direction is will add unwanted along-track shift to all data positions. To
proportional to the scatterer’s along-track position relative to neutralize the effects of vertical velocity, offsetting Doppler
the zero-Doppler position. shifts can be applied to assure registration of the height
The remaining data processing is carried out in parallel, estimates accumulated over the sequence of processed bursts
consisting of a range IFFT at each Doppler frequency bin, that span the along-track length of the antenna illumination
detection, and assignment of the height estimates to their pattern.
respective along-track positions. The process is repeated over The along-track velocity ( 6.5 km s ) of a conventional
subsequent blocks of data, from which many looks are accu- altimeter causes the effective footprint for a multipulse wave-
mulated at each along-track position. As the altimeter passes form to be elongated along track. Detected returns from
over each scatterer, the corresponding height estimates move many pulses are averaged together to build each multilook
in sequence from the highest Doppler filter to each lower waveform. Such signal summations typically extend over 0.1 s,
frequency filter until the scatterer is out of sight. Thus, during which time the antenna illumination pattern progresses
the final height estimate at each along-track position is the in the along-track direction by a distance ( 650 m) that is
average (incoherent sum, normalized) of estimates from all comparable to the radius of the pulse-limited circle. As a
Doppler filters. If the Doppler filters are designed to span the result, the effective footprint for a conventional altimeter, after
along-track antenna beamwidth, all data along track contribute multilook summation, is a set of elliptical annuli, elongated
to the height estimates. The along-track impulse response along track, rather than the circular single-pulse footprints nor-
is determined by the Doppler filter weightings. Along-track mally cited in the literature. In contrast, multilook processing
impulse position is determined by the zero-Doppler position in the delay/Doppler altimeter does not cause spreading of the
for each burst of data. along-track footprint. Instead, Doppler signal analysis not only
reduces the length of the effective footprint along track, but
also synchronizes the relative location of each height estimate
IV. FOOTPRINT with the forward motion of the instrument, thus eliminating
Refer again to the altimeter footprints sketched in Fig. 1. along-track elongation of the footprint.
Fig. 1(a) shows the side view of a pulse radiating from It follows that the customary concept of “flat surface re-
a conventional altimeter toward a quasiflat surface. As the sponse” applies only to the delay time dimension for a
pulse propagates, its radially symmetric intersections with the delay/Doppler altimeter. This means that the inherent de-
surface define a sequence of spreading annuli as in Fig. 1(b) lay/elevation ambiguity that plagues pulse-limited altimeters
[8]. It can be shown that the annuli each have equivalent areas, is reduced from two spatial dimensions to only one dimen-
which give rise to a relatively constant backscatter level after sion. The cross-track ambiguity that remains is suggested in
the initial response. Fig. 1(d), which shows that at any given Doppler frequency,
As introduced above, the delay/Doppler altimeter [Fig. 1(c)] there are two possible sources for reflections having a given
uses coherent processing over a block of received returns (relative) time delay. These arise from either side of the mini-
to estimate the Doppler FM imposed on the signals by the mum delay locus, which nominally is the subsatellite track. Of
forward motion of the altimeter. Doppler analysis of the data course, the point of first reflection (at zero relative delay time)
allows their relative along-track positions to be estimated may be to one side of the subsatellite track, as would be true in
relative to the position of the altimeter. It follows that the general when there is a nonzero cross-track terrain slope. The
along-track dimension of the signal data and the cross-track cross-track ambiguity and the delay/elevation ambiguity both
(range or time-delay) dimension are separable. In contrast to may be at least partially resolved through application of other
the response of a conventional altimeter, which has only one means, such as the monopulse phase sensing technique [4].
independent variable (time delay), the delay/Doppler altimeter
response has two independent variables: along-track position
( ) and cross-track position (functionally related to time-
delay ). After delay/Doppler processing, these two variables V. FLAT SURFACE RESPONSE
describe an orthonormal data grid, as shown in Fig. 1(d). With The average flat surface response function describes the
this data space in mind, delay/Doppler processing may be time-delay history of reflections observed by an altimeter
interpreted as an operation that flattens the radiating field in over a quasiflat horizontal surface [2]. (In this work, quasiflat
the along-track direction. In this transformed data space, the means that the observed reflectivity per area increment is
resolved ( ) cells have constant along-track length, but their independent of incidence over the restricted set of angles
cross-track width decreases as the square root of delay time. illuminated by the radar and there is no perceptible variation
The true nadir point is contained in the plane orthogonal in terrain elevation within the illuminated region.) To accom-
to the satellite ground track for which the Doppler shift is modate the delay/Doppler altimeter, the concept of flat surface
zero. The along-track location of this plane is independent of response must be extended to two independent variables, one
satellite attitude as well as terrain slope. The delay estimates at corresponding to relative delay time and one corresponding
all Doppler frequencies can be located along track with respect to the along-track position of the altimeter. The shape of the
to nadir. Ideally, the along-track nadir location is equivalent delay-time response is determined primarily by the resolved
to along-track spacecraft position. Vertical spacecraft velocity scattering area in the cross-track direction. The shape of
RANEY: DELAY/DOPPLER RADAR ALTIMETER 1583
(10)
Fig. 8. Given comparable hardware (the same average transmitted power, the received during the 100- s interval between transmitted
same antenna gains, etc.), the response of a delay/Doppler altimeter can be pulses. The timing between bursts is controlled so that the
more than 10 dB stronger than the response of a conventional pulse-limited
altimeter. receive intervals are synchronized with the arriving echoes.
These parameters were chosen to make the best use of the
available time, under the constraints of minimizing transmitted
the delay/Doppler altimeter over the pulse-limited altimeter is peak power and maximizing the number of looks.
given simply by the ratio of the equivalent areas over which The approach described in more detail in the following
the signals are integrated. It is evident that the delay/Doppler paragraphs has a superficial similarity to TOPEX, in that the
altimeter makes better use of transmitted power, primarily burst length corresponds to the round-trip range delay time. In
because more of its integration contributes directly to energy contrast to the TOPEX model, however, for the delay/Doppler
at the estimated height of each scatterer. altimeter, the interburst interval is used for echo reception, not
the interpulse periods. As will be seen, this supports a robust
C. Example range gate timing scheme yet maintains an efficient duty cycle.
The magnitude of the potential gain improvement can be This burst sequence strategy has other features that work well
illustrated through an example. Consider an altimeter that is with the delay/Doppler data processing requirements.
similar to TOPEX/Poseidon [7] in most regards, having the The suggested burst mode plan is shown in Fig. 9. Burst
same average radiated power, same beamwidth, same pulse length can be constant, chosen to be a little less than the
compression and range resolution, etc. The only difference minimum round-trip delay time from the altimeter to the
is that the altimeter is operated in the delay/Doppler mode. surface. The PRF must be higher than the Doppler bandwidth
The radiometric performance of the delay/Doppler altimeter of the echo ensemble, yet it must be low enough that the pulse
compared to that of TOPEX/Poseidon is given by the ratio length is less than the pulse period (1/PRF). Once chosen, the
of their two (flat surface) response functions, scaled by their PRF can be constant. The number of pulses transmitted in each
relative powers (18). The result, plotted in Fig. 8, shows that burst can be a constant. The burst period should be variable,
the radiometric response of the delay/Doppler altimeter is but a variation of only 5% would be more than adequate to
more than ten times stronger at the desired altitude mark than accommodate the expected span of ranges to be presented
the response of a conventional radar altimeter, which is an to a system onboard a satellite in circular orbit. The burst
advantage of more than 10 dB. period is constrained to map the position of all height estimates
onto along-track locations that are congruent with those of the
VII. ONBOARD ALONG-TRACK PROCESSING previous burst. The processing is completed for each along-
track position by summation of the detected height estimates
The delay/Doppler altimeter requires pulse-to-pulse signal from the entire Doppler spectrum as the altimeter passes over.
coherence to support the along-track fast Fourier transform
(FFT). Coherence implies that the pulse repetition frequency
(PRF) must be higher than that used on previous radar altime- A. Design Constraints
ters, such as TOPEX. This section elaborates on consequences This particular burst mode strategy, combined with the
of the high PRF requirement and suggests a design approach desire for pulse-to-pulse coherence, leads to design rules and
that leads to a satisfactory solution for the along-track timing performance constraints for the delay/Doppler altimeter, which
and data sequencing issues. are explored in this section. It turns out that election of the
At each of its two frequencies, the TOPEX altimeter [7] uses interburst interval as the echo reception window, rather than
a burst sequence of 38 pulses. Each pulse has about 100- s the interpulse period, constrains the design to a relatively
length and 220- s pulse repetition period (4500-Hz PRF). The narrow set of options. Fortunately, these options should be
burst length corresponds to the round-trip pulse propagation acceptable for many applications. There are tradeoffs that are
time from the altimeter to the surface. Reflected signals are considered further as they arise in the discussion.
1586 IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, VOL. 36, NO. 5, SEPTEMBER 1998
(21)
Thus, the range history at every Doppler frequency can be which in the present case is approximately 150 looks at 800-
assigned to its corresponding along-track position as each km altitude. This would expand to about 200 looks at the
block is processed. TOPEX altitude of 1334 km. Note that these looks are for each
The upper bound on pulse length is set by the PRF cell whose dwell time is about 35 ms. In order to compare 200
according to looks to the TOPEX example, the number of looks needs to be
increased in proportion to its 53-ms track interval [7]. Thus,
(24) 200 (53/35) 300, which is larger than the 228 samples
that TOPEX averages in an equivalent time period.
which for 13.333 kHz is about 75 s. In the burst mode
described here, there is no need for an interpulse period larger
than zero. The mode allows the pulses to be concatenated to D. The Need for Focusing
completely fill the burst length, becoming in effect a sawtooth- The foregoing has been predicated on a simple isometry
modulated CW signal. The PRF also sets an upper limit on the between Doppler frequency and along-track spatial position.
unambiguous range gate length This equivalence is valid for an along-track resolution that is
comparable to or larger than the first Fresnel zone [1]. In SAR
(25) parlance, this zone is known as the unfocused SAR resolution
where for 13.333 kHz is about 11.25 km, which is more than [3]. Using the classic quarter wavelength criterion, it may be
adequate. shown that the radius of the first Fresnel zone is
track resolution below about 200 m, and the radar wavelength [4] J. R. Jensen, “Angle measurement with a phase monopulse radar
is not a parameter of choice. altimeter,” IEEE Trans. Antennas Propagat., to be published.
[5] J. L. MacArthur, C. C. Kilgus, C. A. Twigg, and P. V. K. Brown,
“Evolution of the satellite radar altimeter,” Johns Hopkins APL Tech.
VIII. CONCLUSIONS Dig., vol. 10, no. 4, pp. 405–413, 1989.
[6] J. L. MacArthur, P. C. Marth, Jr., and J. G. Wall, “The GEOSAT radar
The delay/Doppler altimeter represents a new class of altimeter,” Johns Hopkins APL Tech. Dig., vol. 8, no. 2, pp. 176–181,
1987.
radar altimeter. The onboard algorithm compensates for the [7] P. C. Marth, J. R. Jensen, C. C. Kilgus, J. A. Perschy, J. L. MacArthur,
along-track component of the incrementally larger range delay D. W. Hancock, G. S. Hayne, C. L. Purdy, L. C. Rossi, and C. J.
suffered by all scatterers when they are not within the beam- Koblinsky, “Prelaunch performance of the NASA altimeter for the
TOPEX/Poseidon project,” IEEE Trans. Geosci. Remote Sensing, vol.
limited (minimum range) footprint. Implementation depends 31, pp. 315–332, Mar. 1993.
on access to the delay/Doppler signal domain. In turn, this im- [8] R. K. Moore and C. S. Williams, Jr., “Radar terrain return at near-vertical
plies that the pulse-to-pulse data sequence must be coherent, so incidence,” Proc. IRE, Feb. 1957, vol. 45, pp. 228–238.
[9] E. M. Purcell, “The radar equation,” in Radar System Engineering, L.
that a Fourier transform can be used to expose the along-track N. Ridenour, Ed. New York: McGraw-Hill, 1947, ch. 2.
(Doppler) frequencies. In this domain, the delay curvatures [10] R. K. Raney, “Doppler properties of radars in circular orbits,” Int. J.
are known and can be fully compensated by a phase multiply. Remote Sensing, vol. 7, no. 9, 1986, pp. 1153–1162.
[11] , “Considerations for SAR image quantification unique to orbital
The technique is analogous to range curvature correction used systems,” IEEE Trans. Geosci. Remote Sensing, vol. 29, pp. 754–760,
in SAR processing. Data rates admit real-time implementation Sept. 1991.
[12] , “Delay compensated Doppler radar altimeter,” United States
with compact onboard hardware. Patent 5 736 957, April 7, 1998 (international patents pending).
An example has been worked to show that substantially [13] , “A delay/Doppler radar for ice sheet monitoring,” in Proc., Int.
sharpened along-track resolution can be achieved with a simul- Geosci. Remote Sensing Symp. IGARSS’95, Florence, Italy, July, 1995,
pp. 862–864.
taneous increase in the number of looks, at least in comparison [14] R. K. Raney, H. Runge, R. Bamler, I. G. Cumming, and F. H. Wong,
with the best available pulse-limited altimeter. Indeed, it can “Precision SAR processing using chirp scaling,” IEEE Trans. Geosci.
be shown that the number of statistically independent looks Remote Sensing, vol. 32, pp. 786–799, July 1994.
[15] J. L. Walker, “Range-Doppler imaging of rotating objects,” IEEE Trans.
per along-track distance is always larger for the delay/Doppler Aerosp. Electron. Syst., vol. AES-16, pp. 23–52, Jan. 1980.
radar than for a conventional radar altimeter. These benefits are [16] E. J. Walsh, “Pulse-to-pulse correlation in satellite radar altimeters,”
Radio Sci., vol. 17, pp. 786–800, July/Aug. 1982.
a direct consequence of integration over the full along-track [17] P. M. Woodward, Probability and Information Theory, with Applications
beamwidth, a feature unique to the delay/Doppler approach. to Radar. London, U.K.: Pergamon, 1953.
Direct benefits of the delay/Doppler delay compensation in-
clude: 1) increased efficiency of height estimation, 2) increased
geometric stability of the height estimation footprint, and 3)
increased averaging for each elevation estimate, leading to
reduced estimation variance. These technical benefits lead
R. Keith Raney (S’61–M’62–SM’80–F’91) was
to an ocean altimeter that provides improved performance with the Canada Centre for Remote Sensing, Ot-
while requiring fewer onboard resources. Furthermore, the tawa, Ont., from 1976 to 1994, where he helped to
peaked impulse response waveform and the stability of the initiate the Radarsat mission and was responsible
for the conceptual design of the SAR system. He
instrument’s footprint should help to extend radar altimetry to contributed to the conceptual design of NASA’s
more demanding applications, such as polar ice sheets. Magellan, the European Space Agency’s ERS-1
SAR and scatterometer, and the Shuttle Imaging
Radar SIR-C. He was scientific authority for the
ACKNOWLEDGMENT world’s first digital processor for satellite SAR data
(Seasat, 1978) and contributed to the design of the
The author has benefitted from numerous animated dis- X-SAR digital processor at DLR, Germany. He is currently a member of the
cussions with colleagues. The comments, suggestions, and Principal Professional Staff, Space Department, The Johns Hopkins University
creative questions from R. Jensen have been especially helpful, Applied Physics Laboratory, Laurel, MD, and is the Assistant Supervisor of
the Ocean Remote Sensing Group. He is the principal inventor for the United
and they are gratefully acknowledged. States patent on the chirp scaling SAR processing algorithm and has a patent
on the delay/Doppler radar altimeter. These and other contributions in remote-
REFERENCES sensing systems, theory, and applications are documented in more than 300
professional publications.
[1] P. Beckman and A. Spizzichino, The Scattering of Electromagnetic Dr. Raney has been an Associate Editor for the International Journal of
Waves from Rough Surfaces. London, U.K.: Pergamon, 1963. Remote Sensing since its inception in 1980 and Associate Editor (radar) for
[2] G. S. Brown, “The average impulse response of a rough surface and its the IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING since 1982. He
applications,” IEEE Trans. Antennas Propagat., vol. AP-25, pp. 67–74, is a past President of the Geoscience and Remote Sensing Society (1988 and
Jan. 1977. 1989) and a Fellow of the Electromagnetics Academy. He is a recipient of the
[3] L. J. Cutrona and G. O. Hall, “A comparison of techniques for achieving Outstanding Achievement Award and the Outstanding Service Award of the
fine Azimuth resolution,” IRE Trans. Military Electron., vol. MIL-6, pp. IEEE Geoscience and Remote Sensing Society, a Group Achievement Award
119–133, 1962. for the Magellan Radar Science Team, as well as other awards.