6S User Guide Version 2, July 1997

55
APPENDIX III : DESCRIPTION OF THE SUBROUTINES
Contents
Description of the differents subroutines used to compute the geometrical conditions:
SUBROUTINE POSGE 61
SUBROUTINE POSGW 62
SUBROUTINE POSLAN 63
SUBROUTINE POSMTO 64
SUBROUTINE POSNOA 70
SUBROUTINE POSSOL 74
SUBROUTINE POSSPO 78
Description of the subroutines used to compute the different atmospheric functions:
SUBROUTINE ABSTRA 81
SUBROUTINE AEROSO 88
SUBROUTINE ATMREF 94
SUBROUTINE CHAND 96
SUBROUTINE CSALBR 97
SUBROUTINE DISCOM 98
SUBROUTINE DISCRE 99
SUBROUTINE ENVIRO 100
SUBROUTINE GAUSS 105
SUBROUTINE INTERP 106
SUBROUTINE ISO 107
SUBROUTINE KERNEL 108
SUBROUTINE MIE (AND EXSCPHASE) 109
6S User Guide Version 2, July 1997
56
SUBROUTINE ODA550 128
SUBROUTINE ODRAYL 130
SUBROUTINE OS 132
SUBROUTINE SCATRA 135
SUBROUTINE TRUNCA 137
Description of the subroutines used for BRDF ground
SUBROUTINE HAPKALBE 139
SUBROUTINE IAPIALBE 140
SUBROUTINE MINNALBE 140
SUBROUTINE OCEALBE (AND GLITALBE) 140
SUBROUTINE RAHMALBE 140
SUBROUTINE ROUJALBE 140
SUBROUTINE VERSALBE 140
SUBROUTINE WALTALBE 140
SUBROUTINE BRDFGRID 141
SUBROUTINE HAPKBRDF 142
SUBROUTINE IAPIBRDF 144
SUBROUTINE MINNBRDF 148
SUBROUTINE OCEABRDF (AND OCEATOOLS) 149
SUBROUTINE RAHMBRDF 159
SUBROUTINE ROUJBRDF 161
SUBROUTINE VERSBRDF 163
SUBROUTINE WALTBRDF 167
SUBROUTINE AKBRDF 168
Description of the subroutines used to update atmospheric profile (airplane, elevated target):
SUBROUTINE PRESPLANE 175
SUBROUTINE PRESSURE 176
6S User Guide Version 2, July 1997
57
Description of the different subroutines used to read the data:
SUBROUTINE SOLIRR 179
SUBROUTINE VARSOL 180
SUBROUTINE AVHRR 181
SUBROUTINE GOES 185
SUBROUTINE HRV 186
SUBROUTINE METEO 188
SUBROUTINE MSS 189
SUBROUTINE TM 190
SUBROUTINE MODIS 191
SUBROUTINE POLDER 193
SUBROUTINE CLEARW 195
SUBROUTINE LAKEW 196
SUBROUTINE SAND 197
SUBROUTINE VEGETA 198
SUBROUTINE DICA 1 TO 6 199
SUBROUTINE METH 1 TO 6 200
SUBROUTINE MOCA 1 TO 6 201
SUBROUTINE NIOX 1 TO 6 202
SUBROUTINE OXYG 3 TO 6 203
SUBROUTINE OZON 1 204
SUBROUTINE WAVA 1 TO 6 205
SUBROUTINE DUST 206
SUBROUTINE OCEA 206
SUBROUTINE SOOT 206
SUBROUTINE WATE 206
SUBROUTINE BBM 206
SUBROUTINE BDM 206
SUBROUTINE STM 206
SUBROUTINE MIDSUM 207
SUBROUTINE MIDWIN 208
SUBROUTINE SUBSUM 209
SUBROUTINE SUBWIN 210
SUBROUTINE TROPIC 211
SUBROUTINE US 62 212
6S User Guide Version 2, July 1997
58
Miscellaneous
SUBROUTINE EQUIVWL 215
SUBROUTINE PRINT_ERROR 216
SUBROUTINE SPECINTERP 217
SUBROUTINE SPLIE2,SPLIN2,SPLINE,SPLINT 218
6S User Guide Version 2, July 1997
59
DESCRIPTION OF THE SUBROUTINES USED
TO COMPUTE THE GEOMETRICAL CONDITIONS
6S User Guide Version 2, July 1997
60
6S User Guide Version 2, July 1997
61
SUBROUTINE POSGE
Function: Same as POSMTO but for GOES East satellite. We use exactly the same scheme,
only we add the longitude of the subsatellite point, namely 75W, at the retrieval longitude. Let us
also recall that the dimension of the frame is 17331 12997 and the altitude of the satellite is 35729
km.
6S User Guide Version 2, July 1997
62
SUBROUTINE POSGW
Function: Same as POSMTO but for GOES West satellite. We use exactly the same scheme
but we add the longitude of the subsatellite point, namely 135W, at the retrieval longitude. Let us
recall that the dimension of the frame is 17331 12997 and the altitude of the satellite is 35769 km.
6S User Guide Version 2, July 1997
63
SUBROUTINE POSLAN
Function: To compute the geometrical conditions for the LANDSAT satellite. As the
dimensions of the frame are 180 180 km, the maximum observation angle is 5.5, so we put
V
=
0. The incident conditions are taken from the latitude and the longitude of the centre of the scene.
Reference:
NASA, 1981, GSFC specification for the Thematic Mapper Subsystem and associated test
equipment. Revision C., GSFC 400-8-D.210C, NASA/GSFC, Greenbelt, Maryland, U.S.A..
6S User Guide Version 2, July 1997
64
SUBROUTINE POSMTO
Function: To compute the geometrical conditions from the knowledge of the line number and
the pixel in the line (in Meteosat Frame 2500*5000). Firstly, we compute the latitude and the
longitude of the pixel, so the solar position (with the time conditions) can be computed and
secondly, the observation angle.
y
NL
NC
M
W
N
S
P
5000
2500
x
S
x
y
Description: Let S be the satellite, M the subsatellite point, P the observed point, and the
orientation of the axes according to Fig. 1.
From N
c
and N
1
, we obtain the two angles X, Y with respect to x and y axes.
If we refer now to the plane containing the points S, P, M and O the center of the earth (cf.
Fig. 2), we put H the altitude of the satellite and R
E
the earth radius in Equatorial plane.
The observed point P corresponds to the point P' on the earth surface and we have to
determine its coordinates x, y, z with respect to the axis-system centered at O.
6S User Guide Version 2, July 1997
65
Figure 2

z
O
H
S
M
N
P
P'
R
E

To obtain z, we put,
z = ON = (R
E
+H) - SN,
SN is obtained from the solutions of triangles OP'S and SNP'.
By solving OP'S, we have
SP'
2
- 2SP' cosq OS = OP'
2
- OS
2
,
so
SP' cos( )(H R
E
) (R
E
H)
2
cos( )
2
1 R
E
2
By solving SNP' we have SN = SP' cos ,
then SN cos( )
2
(H R
E
) cos( ) (R
E
H)
2
cos( )
2
1 R
E
2
Therefore :
z R
E
+ H cos ( )
2
H+ R
E
R
E
R
E
+ H
R
E



_
,

2

R
E
+ H
R
E



_
,

2
1
cos()
2






1
]
1
1
1
1
To estimate cos, we use the deviations X and Y. A simple trigonometrical identity shows
that :
6S User Guide Version 2, July 1997
66
cos
2
=cos
2
X.cos
2
Y
where
cos
2
X
1
1 tan
2
X
and
cos
2
Y
1
1 (tanY(1 ))
2
with
R
E
R
P
R
E
=
1
297
R
E
and R
P
being respectively the equatorial and polar radius, slightly different because of the earth's
oblateness.
O
y
x
z
x1
P"
y

x
z
N
y
To obtain x and y, we consider the Figure 3.
x = -SN tanY
and
y = SN/cosX.tanY
So we have the three coordinates x, y, z of the point P' and infer latitude and longitude .
To compute the longitude, we use, = arctan (x/z) (+ for East, - for West).
6S User Guide Version 2, July 1997
67
To compute the latitude, we have to consider the geoide with semiminor axis R
p
and
semimajor axis R
E
(Figure 4) by solving triangle P'OP". As = atan( y/x
1
)where y is the ordinate of
the point located on the ellipsod, so we have to compute x
1
.
Figure 4
R
p

x
1 R
E
x
y
The ellipse equation is written:
x
1
2
R
E
2
+
y
2
R
P
2
=1
so
atan(
y
R
P
2
y
2
(R
P
/R
E
)
)
atan(tan (R
P
/R
E
))
with = asin(y/R
P
).
6S User Guide Version 2, July 1997
68
To obtain observations angles (azimuthal and zenithal), we use the following simple
geometrical considerations.
For the zenithal angle
v,
Figure 5
R

v
P'
S
H

v
= asin ( (1+
h
R
) sin)
where is so as cos
2
= cos
2
X cos
2
Y.
For the azimuthal angle
V
, we solve the spherical triangle P'P''M
Figure 6
6S User Guide Version 2, July 1997
69

v
y
x
z
S

where A =
V
- ,
with tan A = tan (1/sin ),
S

v
so
V
= arctan[ tan (1/sin)] + .
From the line and column numbers in a METEOSAT Frame, we can compute the latitude ()
and the longitude () of the point, and the viewing direction from the normal at the point (azimuthal

V
and zenithal
V
angles). Moreover, if we know the date and the hour of the acquisition, we can
obtain the solar conditions (
s,

s
) from the subroutine POSSOL.
Reference:
MORGAN, 1981, Introduction to the Meteosat System, ESOC, Darmstadt, Germany.
6S User Guide Version 2, July 1997
70
SUBROUTINE POSNOA
Function: To compute the geometrical conditions for the NOAA series satellites. Generally,
we know the pixel number on a line, the longitude and the time of the ascendant node at the
equator, and the time of the acquisition. We obtain latitude and longitude of the viewed point, the
viewing angles and with the knowledge of the date, the solar geometrical conditions.
Description: The altitude of NOAA satellite is about H = 860 km, the orbit inclination
98.96and the time of one revolution is about 101.98 mn (6119 sec.). The 1/2 angle is of maximum
55.385and you have 2048 pixels for each line.
Figure 1
Gre. Mer.
P
Eq.
S
N
Na
98.96
6S User Guide Version 2, July 1997
71
Let AN be the hour movement in rad/sec, H
N
the hour at the ascendant node,
N
its longitude
and N
c
the pixel number.
Figure 2
S
N
Na
N
R
Na
i
P
U
B
S
G.M.

v
Consider Fig. 2, S is the subsatellite point, N the ascendant node and P the observed point.
The scan angle gives an angle noted at the centre of the earth.
By solving the triangle PRN, we have the latitude
P
so as:
sin
P
= sin (i+B) sin (NP)
Now, in triangle PSN,
sin (NP) = sin / sin B
and
tan (B) = tan / sin U
So,
sin
P
= cos i sin +sin i cos sinU
By solving PRN, we obtain the longitude
P
with respect to
N
,
6S User Guide Version 2, July 1997
72
sin
P
= sin NPR sin NP
with
sinNPR
cos(i B)
cos
P
so, we write
sin
P
cos(i B)sinNP
cos
P
or
sin
P
cos(i B)sin
cos
P
sinB
sin
P
sini sin cos cosi sinU
cos
P
To completly determine the longitude, we use the other relation which gives the cosine
cos
P
cos cosU
cos
P
The absolute longitude (Greenwich, Meridian reference) is given by
P N
T H
Na
2
86400
where T is the time of the acquisition, the last term is for taking the rotation of the earth between T
and H
Na
into account. Let us recall that the movement angle U is calculated from U=AN.(T- H
Na
).
Consider again Fig. 2 to determine the azimuthal and zenithal observation angles.
is so as
55.385
N
c
1024
1024
in deg.
the zenithal viewing angle
V
is defined by
V
asin[(1
H
R
)sin ]
The observation azimuthal angle
V
is determinated by solving the triangle NSP,
6S User Guide Version 2, July 1997
73
sin
V
sin(
S P
)cos
s
sin
and
cos
V
sin
S
sin
P
cos
cos
P
sin
where
S
and
S
are the latitude and the longitude of the subsatellite point P.
Reference:
The characteristics of the orbit have been taken from, NOAA POLAR ORBITER DATA USERS
GUIDE, 1985, U.S. Department of Commerce, NOAA: National Environment Satellite,
National Climatic Data Center, Satellite Data Service Division, World Weather Building,
Room 100, Washington DC 20233, U.S.A..
6S User Guide Version 2, July 1997
74
SUBROUTINE POSSOL
Function: To compute the solar azimuthal and zenithal angles (in degrees) for a point over
the globe defined by its longitude and its latitude (in dec. degrees) for a day of the year (fixed by
number of the month and number of the day in the month) at any Greenwich Meridian Time (GMT
dec. hour).
Figure 1
S
N

t
S
P
6S User Guide Version 2, July 1997
75
Description: Let P be the point determined by the latitude and the declination of the sun
at this period of the year, the hour angle is noted t. So the incident angle
S
can be determined by
spherical trigonometry expression
cos
S
= cos (

2
- ) cos (

2
- ) + sin (

2
- ) sin (

2
- ) cos t
or
cos
v
= sin sin + cos cos cos t
The solar declination depends upon the day of the year. We used the decomposition in Fourier
series of the declination based on astronomical data with the expression:
=
1
-
2
cos(A)+
3
sin(A) -
4
cos(2A)+
5
sin(2A)-
6
cos(3A)+
7
sin(3A)
where A=
2J
365
and J is the julian day

1
=.006918,
2
=.399912,
3
=.070257,

4
=.006758,
5
=.000907,
6
=.002697,
7
=.001480
The hour angle is computed from the following considerations. From the GMT time, we
compute the mean solar time (or local time) for the longitude
MST = GMT +

15
(dec.hour) .
The length of the day changes within the year (differences between +30 s and - 20 s), so we
have to correct the local time to obtain the true solar time (TST).
TST = MST + ET
where the equation of time ET is given by:
ET=
(
1
+
2
cos(B)-
3
sin(B)-
4
cos(2B)-
5
sin(2B))12

(dec.hour)
with
B =
2J
365
,
1
=.000075,
2
=.001868,
3
=.032077,
4
=.014615,
5
=.040849
We obtain the hour angle t
6S User Guide Version 2, July 1997
76
t = 15

180
(TST-12) (radians)
and can compute
S
.
Figure 2
N
t
S P

s
/2 - /2 -

To determine the azimuthal angle
S
, we solve the spherical triangle NSP:
sin
sin(

2
-)
=
sint
sin
S

where is the solar azimuthal angle measured from the south through the west.
Or
sin=cos
sint
sin
S
.
To determine the sign of we use the cosine
cos=
cossin+coscost
sin
S

so is completly defined.
To define the solar azimuthal angle
S
with respect to North, we write,

S
= + asin .
References:
6S User Guide Version 2, July 1997
77
Ch. PERRIN DE BRICHAMBAUT, Rayonnement Solaire et Echanges Radiatifs Naturels,
Monographies de Mtorologie, Gauthier-Villars, Paris, France, 1963.
N. ROBINSON, Solar Radiation, Elsevier Publishing Company, New-York, N.Y., 10017, 1966.
Figure 3: Simulation of solar angles for the 1
rst
May, at different Latitudes, versus Universal Time.
0
20
40
60
80
100
120
140
160
180
0 4 8 12 16 20 24
Lat: 45 south
Long: 0
S
o
l
a
r

z
e
n
i
t
h

a
n
g
l
e
Universal time
0
40
80
120
160
200
240
280
320
360
0 4 8 12 16 20 24
Long: 0
S
o
l
a
r

a
z
i
m
u
t
h

a
n
g
l
e
Universal time
Lat: 45 south
0
20
40
60
80
100
120
140
160
180
0 4 8 12 16 20 24
Long: 0
S
o
l
a
r

z
e
n
i
t
h

a
n
g
l
e
Universal time
Lat: 10 south
0
40
80
120
160
200
240
280
320
360
0 4 8 12 16 20 24
Long: 0
S
o
l
a
r

a
z
i
m
u
t
h

a
n
g
l
e
Universal time
Lat: 10 south
0
20
40
60
80
100
120
140
160
180
0 4 8 12 16 20 24
Long: 0
S
o
l
a
r

z
e
n
i
t
h

a
n
g
l
e
Universal time
Lat: 45 north
0
40
80
120
160
200
240
280
320
360
0 4 8 12 16 20 24
Long: 0
S
o
l
a
r

a
z
i
m
u
t
h

a
n
g
l
e
Universal time
Lat: 45 north
6S User Guide Version 2, July 1997
78
SUBROUTINE POSSPO
Function: To compute the geometrical conditions for the SPOT satellite. As the dimensions
of the frame are 60 60 km with an observation angle of maximum 2.06, we have considered that:
the zenithal observation angle is nul, so the azimuthal angle is not defined,
the incident conditions are the same that those computed for the center of the frame.
Note: We have not considered the off nadir viewing.
Reference:
M. CHEVREL , M. COURTOIS, G. WEILL (1981). The SPOT Satellite Remote Sensing Mission,
Photogrammetric Engineering and Remote Sensing, 47, 1163-1171.
6S User Guide Version 2, July 1997
79
DESCRIPTION OF THE SUBROUTINES USED
TO COMPUTE THE ATMOSPHERIC CONDITIONS
6S User Guide Version 2, July 1997
80
6S User Guide Version 2, July 1997
81
SUBROUTINE ABSTRA
Function: To compute the gaseous transmittance between 0.25 and 4 m for downward,
upward and total paths. We consider the six gases (O
2
, CO
2
, H
2
O, O
3
, N
2
O and CH
4
) separately.
The total transmission is put equal to the simple product of each ones. The spectral resolution is
equal to 10 cm
-1
.
Description: We have used two random exponential band models (Goody for H
2
O and
Malkmus for O
2
, CO
2
, O
3
, N
2
O and CH
4
) to compute the gaseous transmissions. If we consider an
homogeneous path, the transmission function is written,
for H2O
t
G
exp
N
0
k m
1
k m
0
1
2
for the other gases
t
M
exp
2
0
N
0
1
k m
0
1
2
1
where m is the absorber amount, N
0
the total line number in the frequency interval , k the
average intensity and
0
the average Lorentz half width, obtained from intensity Sj and half width
j of the j
th
spectral line by
k
S
j
j 1
N
0
N
0
6S User Guide Version 2, July 1997
82
k

1
4
S
j
j 1
N
0

S
j

j ( )
1/2
j1
N
0

1
]
1
1
1
2
(04)
The spectral resolution of 10 cm
-1
is sufficient and contains enough spectral lines to use a
random band model transmission function.
From a general point of view, the width of a spectral line corresponds to the convolution
product of the two shapes, Lorentz and Doppler and is therefore called a Voigt line. For an
atmospheric gas (O
2
, CO
2
, H
2
O, O
3
, N
2
O and CH
4
) the altitude where the Lorentz width and
Doppler width are equivalent, is about 30 km. So, according to the vertical distribution, only O
3
requires a more complex treatment to take into account a Voigt profile. The O
3
visible transmission
is computed by an other method detailed in the next part and the absorption in the solar infrared (3.3
m) is very small (cf. Fig. I-2 of the chapter 1, 1). Therefore, we have used the same formalism for
all gases. The approximation contributes no consequential error.
Equations (1) and (2) are valid for a homogeneous path, where pressure and temperature are
assumed to be constant. To take into account the variations of temperature and pressure along the
atmospheric path, we use the Curtis-Godson approximation which associates an amount m
weighted by temperature (thereby related to the line intensity), and a amount m weighted by
pressure and temperature (thereby related to the intensity and half width line)
m (z,z') (T) du
z
z'

, (05)
m(z,z') (T) du
z
z'

, (06)
with
=p/p
0
(p
0
is the standard pressure at which the measurements of spectroscopic parameters have
been made)
du=g(dz/) (g is the gaseous density and the cosine of the viewing angle).
The functions (T) and (T) are given by
6S User Guide Version 2, July 1997
83
(T)
S
j
(T)
j1
N
0

S
j
(T
r
)
j 1
N
0

, (07)
(T)
S
j
(T)
j0
(T)
[ ]
1/2
j 1
N
0

S
j
(T
r
)
j0
(T
r
)
[ ]
1/2
j1
N
0

1
]
1
1
1
2
,
(08)
with T
r
the reference temperature and j
0
the half-width at temperature T
r
and pressure p
0
.
To simplify, we fit these functions with,
(T) exp a(T T
r
) + b(T T
r
)
2
[ ]
, (09a)
(T) exp a'(T T
r
) + b'(T T
r
)
2
[ ]
. (09b)
The spectroscopic data are taken from the AFGL atmospheric absorption line parameters
compilation (1991 edition). We have selected the following parameters :
the position (in cm
-1
),
the integrated line strength Sj (T
r
) at 296 K (in cm
-1
/(molecules-cm
2
)),
the half width j
0
at 296 K and 1013 mb (in cm
-1
),
the energy of the lower transition state.
The half width at any temperature and pressure is obtained by

j
(p,T)
j0
p
p
0



_
,

T
r
T


_
,
1/2
(10)
and the intensity at any temperature can be computed from the vibrational and rotational partition
and the energy of the lower transition state.
Subsequently, we have taken T
r
= 250 K and computed (T) and (T) for 3 temperatures
(200, 250 and 300 K) to determine the coefficients a, a', b and b'.
Now we have a series of eight coefficients by steps of 10 cm
-1
:
6S User Guide Version 2, July 1997
84

k

S
j
j 1
N
0
(T
r
)

(11)

S
j
(T
r
)
j0
(T)
[ ]
1/2
j1
N
0

1
]
1
2
S
j
(T
r
)
j 1
N
0

(12)
a, a'
b, b'

low
the lower frequency of the interval, and
sup
=
low
+ 10 cm
-1
These coefficients are read in the subroutines, WAVA1 to 6 for H
2
O, OZON1 for O
3
,
OXYG3 to 6 for O
2
and DICA1 to 3 for CO
2
.
The weighted absorber amounts m and m are computed according to Eq. (5) and (6) and
the transmission functions (which correspond to Eq. (1) and (2) for a homogeneous path) are written
t

G
exp
k m

1 +
k

0
m
2
m



_
,

1
2




1
]
1
1
, (13)
t

M
exp

0

m
2m
1 +
4 k

0
m
2
m



_
,

1
2
1




_
,





1
]
1
1
. (14)
Due to the deficiency of spectroscopy data, the visible ozone transmission function is written,
t
O
3
() exp(A
O
3
() u
O
3
) (15)
where u
O
3
is the absorber amount, A
O
3
the absorption coefficient given by Kneizys and al. (1980).
These coefficients are given in steps of 200 cm
-1
between 13000 and 24200 cm
-1
and by step
of 500 cm
-1
between 27500 - 50000 cm
-1.
To take into account the water vapor continuum, we use the same expression with the
coefficients A
H
2
O
c
are given in step of 5 cm
-1
between 2350 and 2420 cm
-1
.
6S User Guide Version 2, July 1997
85
A comparison between MODTRAN2 and our results (6S) is shown in the following figures
(1-3). The difference observed at roughly 3.1m is due to the fact that we have not taken into
account the N
2
O continuum. This spectral range is already contaminated by water vapor and is not
an atmospheric window. Therefore, the 3.1m region is not used in remote sensing and its emission
in 6S generally unimportant.
References:
A.R. CURTIS, The computation of radiative heating rates in the atmosphere, Proc. Roy. Soc.
London, A236, p. 156-159, 1956.
R.G. ELLINGSON, J.C. GILLE, An infrared radiative transfer model. part 1: Model description
and comparison of observations with calculations, J. Atmos. Sci. 35, p. 523-545, 1978.
R.M. GOODY, Atmospheric Radiation 1, Theoretical Basis, Oxford University Press, 436 pp, 1964.
F.X. KNEIZYS, E.P. SHETTLE, W.O. GALLERY, J.H. CHETWYND Jr.L.W. ABREU, J.E.A.
SELBY, R.W. FENN, R.A. Mc CLATCHEY, Atmospheric Transmittance/Radiance:
Computer Code LOWTRAN 5, AFGL-TR-80-0067, Air Force Geophysics Laboratory,
Bedford, Mass. 1980.
W. MALKMUS, Random Lorentz Band Model with Exponential- Tailed S
-1
Line-intensity
Distribution Function, J. Opt. Soc. Am. 57, 3, p. 323-329, 1967.
J.J. MORCRETTE, Sur la Paramtrisation du Rayonnement dans les Modles de la Circulation
Gnrale Atmosphrique, Thse d'Etat no 630, Universit de Lille.
D.C. ROBERTSON, L.S. BERNSTEIN, R. HAIMES, J. WUNDERLICH, L. VEGA, 5 cm
-1.
Band
Model Option to Lowtran 5, Applied Optics, 20, p. 3218-3226, 1981.
C.D. RODGERS, C.O. WALSHAW, The Computation of Infrared Cooling Rates in Planetary
Atmospheres, Quart. J. Roy. Meteor. Soc., 92, p. 67-92, 1966.
6S User Guide Version 2, July 1997
86
L.S. ROTHMAN, R.R. GAMACHE, A. BARBE, A. GOLDMAN, J.R. GILLIS, L.R. BROWN,
R.A. TOTH, J.M. FLAUD, C. CAMY-PEYRET, AFGL Atmospheric Absorption Line
Parameters Compilation: 1982 Edition, Applied. Optics, 22, p. 2247-2256, 1983.
0.0
0.2
0.4
0.6
0.8
1.0
0.2 0.4 0.6 0.8 1.0 1.2
6S
MODTRAN
T
g
Wavelength (m)
Figure 1: Gaseous transmission between 0.25 and 1.2m (mid. lat. summer atmosphere)
6S User Guide Version 2, July 1997
87
0.0
0.2
0.4
0.6
0.8
1.0
1.2 1.4 1.6 1.8 2.0 2.2 2.4
6S
MODTRAN
T
g
Wavelength (m)
Figure 2: Gaseous transmission between 1.20m and 2.40m (mid. lat. summer atmosphere).
0.0
0.2
0.4
0.6
0.8
1.0
2.4 2.6 2.8 3.0 3.2 3.4 3.6 3.8 4.0
6S
MODTRAN
T
g
Wavelength (m)
Figure 3: Gaseous transmission between 2.40m and 4.00m (mid. lat. summer atmosphere).
6S User Guide Version 2, July 1997
88
SUBROUTINE AEROSO
Function: To compute the optical scattering parameters (extinction and scattering
coefficients, single scattering albedo, phase function, assymetry factor) at the ten discrete
wavelengths for the selected model (or created model) from:
(1) the characteristics of the basic components of the International Radiation Commission.
(1983).
dust-like component (D.L., SUBROUTINE DUST)
oceanic component (O.C., SUBROUTINE OCEA)
water-soluble component (W.S., SUBROUTINE WATE)
soot component (S.O., SUBROUTINE SOOT)
(2) pre-computed caracteristics,
now available are the desertic aerosol model corresponding to background conditions, as
described in Shettle(1984), a stratospheric aerosol model as measured Mona Loa (Hawaii)
during El Chichon eruption and as described by King et al. (1984), and a biomass burning
aerosol model as deduced from measurements taken by sunphotometers in Amazonia.
(SUBROUTINES BDM, STM and BBM)
(3) computed using the MIE theory with inputs (size distribution, refractive indexes...) given
by the user (see SUBROUTINES MIE and EXSCPHASE).
These models don't correspond to a mixture of the four basic components.
Description: From the MIE theory (see SUBROUTINE MIE), we have computed the phase
function P(), the extinction and scattering coefficients, the assymetry factor g for the basic
components defined by their size distributions and their refractive index. The computations were
performed at 10 wavelengths and 83 phase angles (80 Gauss angles, 0, 90, 180)
Note: We compute the resultant phase function for the scattering angle by linear interpolation
in the table of 83 values.
6S User Guide Version 2, July 1997
89
From the four basic components, three tropospheric aerosols types models have been selected
by mixing with the following volume percentages. By mixing, we suppose an idea of "external
mixing" in the model construction, so the resultant values are obtained by a weighted average using
the volume percentages Cj given by:
D.L. W.S. O.C. S.O.
Continental 0.70 0.29 0.01
Maritime 0.05 0.95
Urban 0.17 0.61 0.22
For each component, we know the volume concentration Vj and the particle number
concentration Nj (particle/cm
3
):
D.L. W.S. O.C. S.O.
Vj
3
/cm
3
113.98352 113.98352 10
06
5.14441 59.777553 10
06
Nj part/cm
3
54.73400 1.86850 10
+06
276.0500010 1.805820 10
+06
where
V
j

4
3
r
3
dN
j
(r)
dr
0
+

dr
and Nj is computed so as to normalize the extinction coefficient at 550 nm .
If Cj is the aerosol fraction by volume of the component j, we have Cj =vj / v with vj = nj Vj
where nj is the number of particles in the mixing so
n n
j
v
C
j
V
j j

then we can obtain the percentage density of particles

n
j
n

C
j
V
j
C
j
V
j j

so for example nj/n for the 3 selected models:

6S User Guide Version 2, July 1997
90
D.L. W.S. O.C. S.O.
Continental 2.26490 10
06
0.938299 0.0616987
Maritime 0.999579 4.20823 10
04
Urban 1.65125 10
07
0.592507 0.407492
To obtain the extinction coefficient of the resultant model, we compute
K
ext
()
n
j
n
K
j
ext
()
j

and we normalize also this coefficient at 550 nm. So we have to compute the equivalent number N
of particles by :
N
1
n
j
n
K
j
ext
(550)
j

Since K
ext
j
(550) = 1/Nj, we obtain
1
N

n
j
n
j

1
N
j
The other optical parameters are computed by the same way:
scattering coefficient :
K
sca
() N
n
j
n
j
K
j
sca
()
assymetry factor :
g()
N
K
sca
()
n
j
n
j
g
j
() K
j
sca
()
phase function :
P

()
N
K
sca
()
n
j
n
j
P

j
()K
j
sca
()
the single scattering albedo is directly obtained by the ratio

0
()
K
sca
()
K
ext
()
6S User Guide Version 2, July 1997
91
Notes:
- The data for extinction or scattering coefficients are in km
-1
- The following figures give us an order of magnitude of these terms for the 3 selected aerosol
models plus the desert aerosol model.
Reference:
World Meteorological Organization (CAS)/Radiation Commission of IAMAP Meeting of experts
on aerosols and their climatic effects, WCP 55, Williamsburg, Virginia, U.S.A., 28-30 March
1983.
E. P. SHETTLE, Optical and radiative properties of a desert aerosol model. Symposium of
Radiation in the atmosphere, 1 Deepak publishing), pp. 74-77, 1984.
M. KING, HARSHVARDHAN, and ARKING, A, A model of the Radiative Properties of the El
Chichon Stratospheric Aerosol Layer, J. Appl. Meteor., 23, (7), pp. 1121-1137, 1984.
6S User Guide Version 2, July 1997
92
Figure 1: Spectral dependence of the extinction coefficient for various aerosol models.
0.01
0.1
1
0.4 0.6 0.8 1.0 2.0 3.0
Continental
Maritime
Urban
Desertic
Biomass
Stratospheric
E
x
t
i
n
c
t
i
o
n

C
o
e
f
f
i
c
i
e
n
t
Wavelength (mm)
Figure 2: Spectral dependence of the single scattering albedo for various aerosol models.
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
0.4 0.6 0.8 1.0 2.0 3.0
Continental
Maritime
Urban
Desertic
Biomass
Stratospheric
S
i
n
g
l
e

S
c
a
t
t
e
r
i
n
g

A
l
b
e
d
o
Wavelength (mm)
6S User Guide Version 2, July 1997
93
Figure 3: Spectral dependence of the assymetry parameter for various aerosol models.
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.4 0.6 0.8 1.0 2.0 3.0
Continental
Maritime
Urban
Desertic
Biomass
Stratospheric
A
s
s
y
m
e
t
r
y

P
a
r
a
m
e
t
e
r
Wavelength (m)
Figure 4: Phase function at 0.550m versus scattering angle for various aerosol models.
0.01
0.1
1
10
1
10
2
10
3
0.0 30.0 60.0 90.0 120.0 150.0 180.0
Continental
Maritime
Urban
Desertic
Biomass
Stratospheric
P
h
a
s
e

F
u
n
c
t
i
o
n
Scattering Angle ()
6S User Guide Version 2, July 1997
94
SUBROUTINE ATMREF
Function: To compute the atmospheric reflectance for the molecular and aerosol atmospheres
and the mixed atmosphere. In 6S instead of an approximation as in 5S, we use the scalar Successive
Order of Scattering method (subroutine OS.f). The polarization terms of aerosol or rayleigh phase
are not accounted for in the computation of the aerosol reflectance and the mixed Rayleigh-aerosol
reflectance. The polarization is addressed in computing the Rayleigh reflectance (Subroutine
CHAND.f) by semi-empirical fitting of the vectorized Successive Orders of Scattering method
(Deuz et al, 1989).
Description: Three reflectance terms have to be computed by ATMREF.f, the aerosol
reflectance (
A
), the rayleigh reflectance (
R
) and the reflectance of the mixed Rayleigh-aerosol
(
R+A
). In addition three different configurations of sensor position are possible, ground based
observation, satellite sensor or airborne sensor.
In the case of ground based observations, we consider that there is no contribution of the
atmosphere below the sensor and the three reflectances are simply set to zero.
For the case of satellite based observations, we can consider that all the molecules and aerosol
are below the sensor. In that case, we use the subroutine OS.f to compute
A
and
R+A
and the
subroutine CHAND.f to compute
R
. The subroutine OS.f is able to deal with a mixture of
molecules and aerosols or with aerosol only or molecules only, by computing the signal in a set of
layers for which the proportion of molecules and aerosol can be adjusted. The computation of the
proportion of aerosol and molecules in each layer is optimized by the subroutine DISCRE.f to
divide the entire atmosphere in equal optical depth layers, the proportion depends on the aerosol
profile which is assumed to be exponential with a scale heigh of 2km.
6S User Guide Version 2, July 1997
95
For the case of airborne observation, the three components are computed by the OS.f
subroutine. In OS.f, a special layer is set so that the top of the layer corresponds to the aircraft's
altitude. When aerosol optical depth below the plane is provided by the user as encouraged, the
scale height of aerosol is computed again to match the total aerosol optical depth, the aerosol optical
below the plane and the plane altitude. If in that case, the scale heigh is found to be greater than
7km a warning message is issued and computation are aborted.
References:
Radiation Commission of IAMAP, Standart Procedures to compute Atmospheric Radiative
Transfer in a scattering Atmosphere. Edited by J. LENOBLE, Available from Dr. S. Ruttenberg,
NCAR, Boulder Colorado 80307, U.S.A., 1977.
D. TANRE, M. HERMAN, P.Y. DESCHAMPS, A. DE LEFFE, Atmospheric Modeling for Space
Measurements of Ground Reflectances including bidirectional properties, Appl. Opt. 18, no 21,
p. 3587-3594, 1979.
J. L. DEUZ, M. HERMAN AND R. SANTER, Fourier series expansion of the transfer equation in
the atmosphere-ocean system. Journal of Quantitative Spectroscopy and Radiative Transfer, 41,
6, 483-494, 1989.
6S User Guide Version 2, July 1997
96
SUBROUTINE CHAND
Function: To compute the atmospheric reflectance for the molecular atmosphere in case of satellite
observation.
Description: In 6S, to save computer ressources but maintain a good accuracy, we used the same
approach to compute the molecular scattering reflectance as it is detailled in a recent paper
(Vermote and Tanr 1992). The molecular reflectance, as computed from 6S is plotted vs the
reflectance computed from the SOS method for =0.35 in Fig. 1. Four values of the solar zenith
angle (0, 53, 66 and 70), 17 values of the viewing zenith angle (from 0 to 60 with a step of
3.3) and 19 values of the difference of the azimuth angles (from 0 to 180 with a step of 10),
covering a large range of possible geometrical conditions, have been selected. Multiple points fall
on the 45-degree line; the right-hand scale, which gives absolute differences between the two
results, clearly shows that the accuracy of 0.001 is achieved for the full range of geometric
conditions.
References:
E. F. VERMOTE and D. TANRE , Analytical Expressions for Radiative Properties of Planar
Rayleigh Scattering Media Including Polarization Contribution. Journal Of Quantitative
Spectroscopy and Radiative Transfer, 47, 4, 305-314, 1992.
Figure 1: Accuracy of CHAND.f
0.10
0.20
0.30
0.40
0.50
0.60
0.000
0.001
0.002
0.003
0.004
0.005
0.10 0.20 0.30 0.40 0.50 0.60
Reflectance
Absolute Difference
A
P
P
R
O
X
I
M
A
T
E

R
E
F
L
E
C
T
A
N
C
E
ACTUAL REFLECTANCE
A
B
S
O
L
U
T
E

D
I
F
F
E
R
E
N
C
E
6S User Guide Version 2, July 1997
97
SUBROUTINE CSALBR
Function: To compute the spherical albedo of the molecular layer.
Description: We integrate the transmission function of the different incident directions to
calculate the spherical albedo, s, that is:
s 1 T() d
0
1

(01)
Using the expression of T() derived in SCATRA (Eq. 01.), it can be shown that s reduces to:
s
1
4 + 3
[3 4E
3
() + 6E
4
()] (02)
where E
3
() and E
4
() are exponential integrals for the argument . These functions are easily
computable from expressions given in the reference below.
Figure 1 shows that the differences between the exact results and Eq. (02) are approximately
0.003 for 0.35 which results in an error of 0.0003 for a surface albedo of 0.10. In the red part of
the solar spectrum for which the surface albedo may be larger, the error is still below 0.001.
References
M. ABRAMOWITZ AND I STEGUN, Handbook of Mathematical Functions (New-York: Dover
Publications,Inc), 1970.
0.00
0.05
0.10
0.15
0.20
0.25
0.000
0.002
0.004
0.006
0.008
0.010
0.05 0.10 0.15 0.20 0.25 0.30 0.35
exact results
analytical expression
Error
S
P
H
E
R
I
C
A
L

A
L
B
E
D
O
A
B
S
O
L
U
T
E

D
I
F
F
E
R
E
N
C
E

OPTICAL THICKNESS
Figure 1: Accuracy of Eq. 02.
6S User Guide Version 2, July 1997
98
SUBROUTINE DISCOM
Function : To compute the optical properties of the atmosphere at the 10 discrete
wavelengths.
Description :
The 10 wavelengths, 0.400, 0.488, 0.515, 0.550, 0.633, 0.694, 0.860, 1.536, 2.250, 3.750,
have been selected because they correspond to the atmospheric windows used in remote sensing.
The computed quantities are
molecular optical depth (subroutine ODRAYL)
aerosol optical depth(subroutine ODA550)
atmospheric reflectances (subroutine ATMREF)
scattering transmittances (subroutine SCATRA)
spherical albedos (subroutine SCATRA).
The computations have been made respectively for the 3 types of atmosphere:
molecular
aerosols only
complete atmosphere with the two components.
6S User Guide Version 2, July 1997
99
SUBROUTINE DISCRE
Function: Decompose the atmosphere in a finite number of layers. For each layer, DISCRE
provides the optical thickness, the proportion of molecules and aerosols assuming an exponential
distribution for each constituants. Figure 1 illustrate the way molecules and aerosols are mixed in a
realistic atmosphere. For molecules, the scale height is 8km. For aerosols it is assumed to be 2km
unless otherwise specified by the user (using aircraft measurements).
Figure 1: Molecules and aerosol mixing in atmosphere.
molecule - aerosol content
layer2
layer3

ground
surface
molecule
aerosol
=0

=1
(z)=(0)exp(-z/h)
6S User Guide Version 2, July 1997
100
SUBROUTINE ENVIRO
Function: To compute the environment functions F(r) which allows us to account for an
inhomogeneous ground.
Description: For an accurate evaluation of F(r), Monte Carlo computations are necessary to
take into account
the altitude dependence of the phase function
the dependence of the phase function upon the aerosols type
the scaling factors which are different for the aerosols and the molecules.
Simulations for some different vertical distributions and phase functions show that the
variability of the environment function F(r) can be rather tractable.
The molecular scattering which is a major factor for the enlarged contribution of the
background can be linearized and accounted for by:
F(r)
t
d
R
(
v
)F
R
(r) + t
d
P
(
v
) F
P
(r)
t
d
R
(
v
) + t
d
P
(
v
)
(01)
where t
d
R
(
v
) and t
d
P
(
v
) are the diffuse fractions in the transmission functions respectively
for Rayleigh and aerosols.
F
R
(r) and F
P
(r) correspond to the environment functions estimated for Rayleigh and aerosols
taken into account separately, these functions are slightly dependent upon the wavelength.
We have computed these 2 functions for a mean atmosphere at the satellite level (Mc
Clatchey et al, 1971) and we propose the following approximations:
F
P
(r) 1 0.448 E
0.27r
0.552 E
2.83r
(02)
F
R
(r) 1 0.930 E
0.08r
0.070 E
1.10r
(03)
6S User Guide Version 2, July 1997
101
where r is in km.
If the actual aerosol model (type and vertical distribution) does not differ much from the mean
model, these approximations are reasonnable and we account for major part of the environment
effect. Figure 1 shows the two functions F
R
(r) and F
P
r. We note that the horizontal scales of the
environment effect are typically 1km for aerosol scattering and 10km for molecular scattering.
For the case of an airborne observation, we computed the altitude dependence of the Rayleigh
and aerosol environment function. For several typical altitude we have computed F
R
(r,z) and F
P
(r,z)
by the Monte Carlo method and we have derived an approximate expression (Eq. (2) and (3)).
Figures 2 and 3 show, for the selected altitudes, the environment functions For a plane flying at an
arbitrary altitude, we perform a linear interpolation between the closest simulated altitudes in 6S to
get the environment function at the altitude of the plane.
Effect of the view zenith angle.
For 6S, we look at the dependence of these two environment functions as a function of the
view zenith angle. Fig. 4a and 4b show for several values of the view zenith angle the environment
function of Rayleigh and aerosol. As it can be observed on Fig. 4a-b, there is a dependence of the
function F(r) on the view direction for view zenith angle larger than 30. In order to account for
this effect, we chose to fit the environment function at the desired view angle solely as a function of
the environment function computed for a nadir view as it is suggested by Fig. 4a-b. The results
presented on Fig. 4a-b (symbols) show that a simple polynomial function of nadir view
environment function whose coefficients depend on the logarithm of the cosine of view angle is
adequate. For molecules, the F function is fitted by the simple expression :
F
R
(
v
) F
R
(
v
0). ln(cos(
v
). 1 F
R
(
v
0) ( ) +1
[ ]
(04)
for aerosol, a polynomial of a higher degree is needed, that is:
F
A
(
v
) F
A
(
v
0).
1+ a
0
ln(cos()) + b
0
ln(cos())
2
[ ]
+
F
A
(
v
0). a
1
ln(cos()) + b
1
ln(cos())
2
[ ] +
F
A
(
v
0)
2
. (a
1
a
0
)ln(cos()) + (b
1
b
0
) ln(cos())
2
[ ]





1
]
1
1
1
(05)
with a
0
=1.3347, b
0
=0.57757, a
1
=-1.479, b
1
=-1.5275
6S User Guide Version 2, July 1997
102
However, it has to be pointed out that if the approximations (04 and 05) enable to take into
account adjacency effect for an arbitrary view angle, they implied uniformity of the background as a
function of azimuth. As contributions of the adjacent pixels for a large view angle don't comply to
the symmetry in azimuth, the 6S results, in case of large view angles, have to be interpreted more
like a sensitivity test to the problem of adjacency effect rather than an actual way to perform
adjacency effect correction .
Reference:
D. TANRE, M. HERMAN and P.Y. DESCHAMPS, Influence of the background contribution upon
space measurements of ground reflectance, Appl. Opt., 20, p. 3676-3684, 1981.
0.8
Radius (km)
F (r)
P
F (r)
R
0 2 4 6 8 10
0.6
0.4
0.2
0.0
1.0
Figure 1: Environment function at satellite level for Rayleigh and Particules.
6S User Guide Version 2, July 1997
103
0
0.2
0.4
0.6
0.8
1
0 2 4 6 8 10
z=0. 5km
z=2km
z=5km
z=7km
E
n
v
i
r
o
n
m
e
n
t

F
u
n
c
t
i
o
n

F
R
(
r
)
Radius (km)
Figure 2: Variation of Rayleigh environment function wih sensor altitude.
0
0.2
0.4
0.6
0.8
1
0 2 4 6 8 10
z = 0. 5km
z=2km
z=5km
z=7km
E
n
v
i
r
o
n
m
e
n
t

F
u
n
c
t
i
o
n

F
p
(
r
)
Radius (km)
Figure 3: Same as figure 2 but for particles.
6S User Guide Version 2, July 1997
104
0. 0
0. 1
0. 2
0. 3
0. 4
0. 5
0. 6
0 2 4 6 8 10
v
=0
v
=30
v
=60
fit F
r
(r,
v
=30) (10a)
fit F
r
(r,
v
=60) (10a)
F
r
(
r
)
r [km]
Figure 4a: Environment function for a pure molecular atmosphere (lines) for different view zenith
angle (
v
) compared to approximation used in 6S (symbols)
as a function of the distance to the imaged pixel (r).
0. 0
0. 2
0. 4
0. 6
0. 8
1. 0
0 2 4 6 8 10
v
=0
v
=30
v
=60
fit F
a
(r,
v
=30) (10b)
fit F
a
(r,
v
=60) (10b)
F
a
(
r
)
r [km]
Figure 4b: same as Figure 4a but for aerosol.
6S User Guide Version 2, July 1997
105
SUBROUTINE GAUSS
Function: Compute for a given n, the gaussian quadrature (the n gaussian angles and the
their respective weights). The gaussian quadrature is used in numerical integration involving the
cosine of emergent or incident direction zenith angle.
6S User Guide Version 2, July 1997
106
SUBROUTINE INTERP
Function: To estimate the different atmospheric functions (
S
,
v
,
S
,
v
), T() and S at any
wavelength from the 10 discret computations (subroutine DISCOM).
Description: The different atmospheric functions (noted f) have been assumed linear as a
function of optical depth , so the interpolation scheme is written,
f() = A
-
The constants A and are interpolated between 0.4 and 3.7 m and extrapolated for the two
extreme intervals 0.25-0.4 and 3.7-4 m.
The spectral dependances for Rayleigh ( = 4) and aerosols ( = 1) are quite different and we
considered the two types of atmosphere separately .
6S User Guide Version 2, July 1997
107
SUBROUTINE ISO
Function: Compute the atmospheric transmission for either a satellite or aircraft observation
as well as the spherical albedo of the atmosphere.
Description: The subroutine performs the computation on the basis of the Sucessive Orders
of Scattering method (see subroutine OS). The transmission is obtained directly by initially setting
the bottom of the atmosphere to a isotropic source of radiation. The spherical albedo is computed by
numerical integration (gaussian quadrature) of the transmission function (see Eq. (01) of
CSALBR.f).
6S User Guide Version 2, July 1997
108
SUBROUTINE KERNEL
Function: Compute the values of Legendre polynomials used in the successive order of
scattering method.
6S User Guide Version 2, July 1997
109
SUBROUTINE MIE (and EXSCPHASE)
Function: To compute, using the scattering of electromagnetic waves by a homogeneous
isotropic sphere, the physical properties of particles whose sizes are comparable to or larger than the
wavelength, and to generate mixture of dry particles.
Description: The interaction of an electromagnetic wave with a absorbing sphere is
described and expressed by the Mie theory (Mie, 1908). This theory has been particularly discussed
by Van de Hulst (Van de Hulst, 1981) and also in part by many other authors (as example Aden,
1951; Deirmendjian et al., 1961; Wyatt, 1962; Kattawar and Plass, 1967; Dave, 1969; Hansen and
Travis, 1974; Liou, 1980). Here, we outline the basic equations of the Mie scattering behind the
computation procedures.
1. Mie Scattering
Let represents the wavelength, r the radius of the sphere, x the Mie's parameter ( x 2 r ), m
the complex index of refraction ( m n
r
in
i
), and the direction of scattered radiation measured
from the forward direction. From the Maxwell's equations, we can defined two complex functions
S
1
(x,m, ) and S
2
(x,m, ) related to the amplitude of the scattered radiation, respectively,
perpendicular and parallel to the plane of scattering
S
1
(x,m, )
(2n 1)
n(n 1)
n 1
a
n
(x,m)
n
(cos ) b
n
(x,m)
n
(cos )
and
S
2
(x,m, )
(2n 1)
n(n 1)
n 1
a
n
(x,m)
n
(cos ) b
n
(x,m)
n
(cos )
1.1 Computation of a
n
(x,m) and b
n
(x,m)
The complex functions a
n
(x,m) and b
n
(x,m) are given by
a
n
(x,m)
n
'
(mx)
n
(x) m
n
(mx)
n
'
(x)
n
'
(mx)
n
(x) m
n
(mx)
n
'
(x)
and
b
n
(x,m)
m
n
'
(mx)
n
(x)
n
(mx)
n
'
(x)
m
n
'
(mx)
n
(x)
n
(mx)
n
'
(x)
6S User Guide Version 2, July 1997
110
where the prime denotes derivative of the function with respect of the argument (x or mx), and
where
n
(z x or mx) and
n
(z x) are the Ricatti-Bessel functions defined by
n
(z)
1
2
z
1
2
J
n
1
2
(z) z j
n
(z)
n
(z)
1
2
z
1
2
N
n
1
2
(z) -z n
n
(z)
n
(z)
1
2
z
1
2
H
n
1
2
(2)
(z) z h
n
(2)
(z)
n
(z)+i
n
(z)
where J
n
1
2
, N
n
1
2
and H
n
1
2
(2)
are respectively the Bessel functions of first, second, and third kind,
and where j
n
, n
n
and h
n
(2)
are the corresponding spherical Bessel functions. N
n
1
2
is also called the
Neumann functions and H
n
1
2
(2)
the half integral order Hankel function of the second kind.
In order to make the computational work more convenient, it is useful to introduce the logarithmic
derivative of the Ricatti-Bessel functions (Infeld, 1947; Aden, 1951; Kattawar and Plass, 1967)
D
n
(z)
d
dz
ln
n
(z)
G
n
(z)
d
dz
ln
n
(z)
making use of these equations, a
n
(x,m) and b
n
(x,m) may be rewritten
a
n
(x,m)
n
(x)
n
(x)
D
n
(mx) mD
n
(x)
D
n
(mx) mG
n
(x)
and
b
n
(x,m)
n
(x)
n
(x)
mD
n
(mx) D
n
(x)
mD
n
(mx) G
n
(x)
Expressions of a
n
(x,m) and b
n
(x,m) are now reduced to a ratio of Ricatti-Bessel functions involving
real arguments and a ratio of "D
n
(mx or x) and G
n
(x) " functions which are easily computable. We
reported Figures 1 and 2, examples of a
n
(x,m) and b
n
(x,m) for m = 1.33 - i 0.001 and for x=10 and
x=50 (which means respectively r 0.8 m and r 4.0m at 0.50 m).
Also in order to save time, we use in 6S the criterion defined by Deirmendjian et al., 1961:
"the quantities a
n
and b
n
are terminated either when (a
n
a
n
*
+ b
n
b
n
*
)/ n < 10
-14
"
1.1.1 Computation of the Ricatti-Bessel function.
The ratio of Ricatti-Bessel functions can be reduced to a ratio of spherical Bessel functions with a
real argument x as follow
n
(x)
n
(x)
j
n
(x)
h
n
(2)
(x)
j
n
(x)
j
n
(x) in
n
(x)
The spherical Bessel functions j
n
(x) , n
n
(x) or h
n
(2)
(x) , have different behaviors following they are
below or above the transition line defined by x
2
n(n + 1). Below the transition line
6S User Guide Version 2, July 1997
111
(n(n + 1) < x
2
), they behave as oscillating functions of both order and argument, whereas the
behavior becomes monotonic above the transition line (n(n 1) x
2
).
It has been shown by many authors that n
n
(x) or h
n
(2)
(x) can be processed using an upward
recurrence (what ever values of n and x). Functions n
n
(x) are computed using
n
n 1
(x) =
2n+1
x
n
n
(x) + n
n 1
(x)
with n
0
(x)
cos(x)
x
n
1
(x)
cos(x)
x
2
sin(x)
x
Figures 03 show examples of the n
n
(x) function for x=10 and x=50.
For j
n
(x) , we use a similar recurrence
j
n+ 1
(x) =
2n+1
x
j
n
(x) + j
n 1
(x)
but, has it is explained in the paper of Corbat and Uretsky, 1959, the function j
n
(x) cannot be
computed by an upward recurrence "since upward recursion (except in the region of the x-n plane
where j
n
oscillate) would bring about a rapid loss of accuracy". Then, a downward recurrence is
called for, but we have to define the starting value of n, and for that purpose we use the work of
Corbato and Uretsky which is summurized hereafter. Let N be the starting order of the recursion
with N(N 1) x
2
, in their paper, they show "that rather than accurately evaluate j
N
(x) and
j
N 1
(x) to start the process, a very approximately starting the recursion at a higher order will give
a set of numbers which are accurately proportional to the j
n
over the desired range of n from 0 to
N". Let j
n
be one of these numbers.
They propose to define the higher order by
N
'

ln
N
ln2
A +
B u
'
(2-u
' 2
)
2(1-u
' 2
)



1
]
1
where A=0.10 and B=0.35,
N
2
30
(this value comes from the fact that generally computers can store floating
points numbers with a 30 binary digit mantissa),
and u
'
2x (2N
'
+ 1) with N' =N or N'= x
1
2
+
ln
N
ln2
Bx such that be
the lower, with however N
'
N.
To avoid computational difficulties above the transition line, Corbato and Uretsky worked with the
ratio r
n
j
n+1
j
n
using the recurrence relation
r
n-1

x
2n + 1 xr
n
6S User Guide Version 2, July 1997
112
with the starting condition r 0 . The recursion is continued downward until a ratio r
n
which
exceeds unity is reached. Then, they set j
n+ 1
r
n
and j
n
1, and continue downward using the
recurrence relation
j
n 1
(x) =
2n+1
x
j
n
(x) + j
n+ 1
(x)
The positive number j
n
is defined by j
n
(x) j
n
(x) with a constant of proportionality obtained
from the relation
=( j
0
(x)- x j
1
(x))cos(x) + x j
0
(x)sin(x)
Figures 03 also show examples of the j
n
(x) function for x=10 and x=50.
1.1.2 Computation of the D
n
(mx or x) and G
n
(x) function.
As Kattawar and Plass, 1967, have pointed out, the procedure of computing D
n
(z) by an upward
recurrence is unstable, then a downward process is needed, and D
n
(z) is defined using
D
n 1
(z) n z
1
D
n
(z) n z
Calculations have to be started at an order n
'
>> z with a starting value which is not really
important because the serie converges rapidly to the exact value (then D ' (z) 0 is a convenient
value). When n z , D
n
(z) becomes oscillatory, and then there is no problem for the calculation in
using the recurrence relation. For practical reasons, we selected in 6S
'
as defined for j
n
.
Kattawar and plass have also shown that G
n
(x) may always be calculated using an upward process
with a starting value G
0
(x) i
G
n
(x) n x
1
G
n1
(x) n x
Figure 04 reportes examples of the D
n
(x) function, Figures 05 the D
n
(mx) function, and Figures 06
theG
n
(x) function.
1.2 Computation of
n
(cos ) and
n
(cos )
Functions
n
and
n
depend of only the scattering angle . They are related to the associated
Legendre polynomials P
n
1
(cos )
n
(cos )
1
sin
P
n
1
(cos )
n
(cos )
d
d
P
n
1
(cos )
and are computed from upward recurrence relations defined as follow
n
n+1
(cos ) (2n+1) cos
n
(cos ) - (n+1)
n-1
(cos )
n+1
(cos ) (n+1) cos
n+1
(cos ) - (n+2)
n
(cos )
6S User Guide Version 2, July 1997
113
with the starting values
0
(cos ) 0
1
(cos ) 1
Examples of functions
n
and
n
are shown Figures 07 for n=1 to 6 and for 0<<90.
2. Computation of the physical properties of a particle (see for example Liou, 1980).
2.1 Extinction
The extinction cross section
e
, which denotes the amount of energy removed (scattered and
absorbed) from the original beam by the particle, is obtained considering a point in the forward
direction ( =0) in the "far field". If we consider an isotropic homogeneous sphere, the extinction
cross section is given by
e
( ,r,m)
4
2 ( )
2

e
S(x,m, 0) [ ]
with
S(x,m, 0) S
1
(x,m, 0) S
2
(x,m, 0)
1
2
(2n 1)[a
n
(x,m) b
n
(x,m)
n 1
]
Thus the extinction efficiency Q
e
is defined by
Q
e
( ,r,m)
e
( ,r,m)
r
2
2
x
2
(2n 1)
e
[a
n
(x,m) b
n
(x,m)
n 1
]
2.2 Scattering
The scattering cross section
s
is derived by a similar way, but considering a scattered light in an
arbitrary direction, by
s
( ,r,m)
x r
2
[S
1
(x,m, )S
1
*
(x,m, ) S
2
(x,m, )S
2
*
(x,m, )]sin d
0
Owing of the functions
n
and
n
, we have to integrate products of the associated Legendre
polynomials. Using the orthogonal and recurrence properties of these polynomials, the scattering
cross section can be written
s
( ,r,m)
2
x r
2
(2n 1)[a
n
(x,m)a
n
*
(x,m) b
n
(x,m)b
n
*
(x,m)]
n 1
where the asterisk denotes the complex conjugate value, and the scattering efficiency Q
s
can be
evaluated by the relation
Q
s
( ,r,m)
s
( ,r,m)
r
2
2
x
2
(2n 1)[a
n
(x,m)a
n
*
(x,m) b
n
(x,m)b
n
*
(x,m)]
n 1
2.3 Absorption
The absorption cross section
a
and the absorption efficiency Q
a
can be deduced from
6S User Guide Version 2, July 1997
114
a
( ,r,m)=
e
( ,r,m)
s
( ,r,m) Q
a
( ,r,m) Q
e
( , r, m)-Q
s
( ,r,m)
2.4 Phase function
On the basis of the Stokes parameters, the intensity I of the electromagnetic waves at each point and
in any given direction can be related to the incident intensity I
0
by
I =M
11
I
0
with
M
11
( ,r,m, )=
1
2x
2
[S
1
(x,m, )S
1
*
(x,m, )+S
2
(x,m, )S
2
*
(x,m, )]
The angular distribution of the scattered energy for a single sphere (also called Phase function)
P
11
( , r,m, ) can be defined by
M
11
( , r,m, ) =
s
( ,r,m)
4 r
2
P
11
( , r,m, )
then
P
11
( ,r,m, )=
2
(x r)
2
s
( ,r,m)
[S
1
(x,m, )S
1
*
(x,m, )+S
2
(x,m, )S
2
*
(x,m, )]
It can be check that
P
11
( ,r,m, )
0

0
2

sin d d 4
3. Physical properties of a sample of identical particles
We now consider a sample of identical particles whose size is described by the size distribution
n(r) (in cm
-3
m
-1
) such that
n(r) dr
0

=
dN(r)
dr
dr
0

=1
where d N(r) represents the number of particle per unit volume having a radius between r and
r+dr.
In 6S, we selected several possibilities to represent the size distribution, thus the user will be
allowed to choice between 4 options:
1- a Junge power-law function . Junge, 1952, showed that the size distribution of aerosols whose
radii are larger than 0.1m may be described by
dN(r)
dlogr
ln(10) c r
0

1
r


_
,
1
or
dN(r)
dr
c r
0

1
r


_
,

with varying between 3 and 5, c the number density of particles with radius r
0
and r
0
an
arbitrary radius.
Figure 08-a shows an example of Junge Power-Law function which is the "Model C"
defined by Deirmendjian, 1969, for c.r
0

=1 and =4
6S User Guide Version 2, July 1997
115
2- a Modified Gamma distribution function. Used by Deirmendjian, 1964, to compute scattering
properties of water clouds and haze and to fit aerosol measurements. Also employed by Mie
in the Mie and diffraction calculations.
dN(r)
dr
A r r
0
( ) exp -b r r
0
( )
( )
with r
0
=1m
An example of Modified Gamma distribution function is given Figure 08-b (Volcanic Ash
defined in WCP 112, A=5461.33, =1.0, =0.5, b=16).
3- a Log-Normal distribution function . Based on the Junge power-law function, Davies, 1974,
introduced this function to take into account large particles.
dN(r)
dlogr

N
2 log
exp -
1
2
log r -log r
M
log



_
,
2



_
,

where r
M
is the mean radius of the particle, and the standard deviation of r.
We reported Figure 08-c examples of Log-Normal distribution functions which are the 3
three components of the "Continental Model" defined in WCP 112 (see AEROSO to find r
M
and ).
4- sun photometer measurements . You enter directly d V(r)/ dlogr r
4
dN(r) / dr .
The Figure 09 shows the same function than Figures 08 but for dV(r)/dlog(r).
Under the assumption of "independent scattering" which means that particles are sufficiently far
from each other compared to the incident wavelength to consider just one scattering, it is possible to
add scattered intensities independently of the phase of the wave. Then we can defined the radiative
characteristics upon the particle size distribution by
The extinction (e), scattering (s) and absorption (a) coefficient
k
e,s,a
( ,m) Q
e,s,a
( ,r,m) r
2
rmin
r
max
dN(r)
dr
dr
The normalized phase function
P( ,m, )
1
k
s
( ,m)
M
11
( ,r,m, ) 4
rmin
r
max

r
2
dN(r)
dr
dr
We now introduce the single scattering albedo
0
which represents the percentage of
energy removed from the incident beam which will reappear as a single scattered
radiation.
0
( , m)
k
s
( , m)
k
e
( , m)
6S User Guide Version 2, July 1997
116
Computationally, k
e,s,a
( ,m) and P( ,m, )are integrated step by step following:
k
e,s,a
( ,m) Q
e,s,a
( ,r,m) r
2
r
min
r
max

dN(r)
dr
r
and
P( ,m, )
1
k
s
( ,m)
M
11
( ,r,m, ) 4
r
min
r
max

r
2
dN(r)
dr
r ,
where r is defined by
log
r + r
r


_
,
0. 03
The value 0.03 has been selected in order to preserve a good accuracy with a reasonable
computational time. For example D'Almeida used a very small step width, 0.011, for the
computations given in his book (D'Almeida et al., 1991). The logarithmic expression of r comes
from the fact that size distributions can be frequently described by a logarithmic shape (Junge,
1952; Davies, 1974).
Finally, and, in order to save computational time, we defined a criterion on the summation such that
the computations are not performed either when
n
i
n
r
2
dN(r)
dr
r


_
,
1
< 10
8
where n
i
n is the percentage density of particles (cf. subroutine AEROSO for some examples). The
latter criterion has been tested between 0.4 and 4.0 m.
4. Physical properties of a mixture of aerosol type
We now consider a mixture of particles originating from different sources (4 max.). The mixing is
treated in the same way that the one used to generated the data base in the AEROSO subroutine.
Let us recall that the mixture of individual components (or type) of an aerosol is characterized by
the percentage density of particles n
i
n , and if we assume that the particles are spherical, each type
i is described by its size distribution (then by its microphysical identity: r
M i
and
i
see Table 1 for
some examples (Shettle and Fenn, 1976; World Climate Programme, 1986), and by its complex
refractive index m
i
(see Table 2, from (Shettle and Fenn, 1976; Shettle and Fenn, 1979; World
Climate Programme, 1986; D'Almeida et al., 1991). For the size distribution, the Log-Normal
distribution is well adapted to emphasize the individual components of a mixture (Davies, 1974,
D'Almeida et al., 1991).
6S User Guide Version 2, July 1997
117
5. Examples and comparisons
The comparison of the computed normalized at 550 nm K
e
, K
s
and
0
values that obtained by 6S
with those given by World Climate Programme, 1986 are reported for a Continental and an Urban
dry aerosol model respectively Table 3 and Table 4. Also reported in theses tables, are the
asymmetry parameters g. In 6S, this parameter is already computed in an another subroutine, but we
can compute it here using
g
cos P( ,m, )
1
+1

dcos
P( ,m, )
1
+1

dcos
The comparison of the phase function of a Continental model (WMO/WCP-112) computed by the
MIE subroutine with those by a precise code (AEROSO subroutine) is reported Figure 10 for
several wavelengths. Also, we show, Figure 11, the phase function computed using the volumic
distribution dV/dlogr provided by a CIMEL sunphotometer during the SCAR-A field experiment
(Sulfate Clouds Aerosol and Reflectances - America) that took place in July 1993 in the Eastern
USA.
6S User Guide Version 2, July 1997
118
The parameters are:
For IAER=8 Multimodal Log Normal (up to 4 modes)
r
min
,r
max
,icp
then for k=1 to icp, enter:
, r
M
, C
ij
r
n
(
j
), j=1,10
r
i
(
j
), j=1,10
where r
min
and r
max
are the radii min and max of the aerosol
icp the number of mode (component)
and r
M
are parameters of the Log-Normale size distributions
Cij is the percentage density of particles (see SUBROUTINE AEROSO)
r
n
and r
i
are the real and imaginary index of refraction of each component
with r=r
n
-i r
i
. You have to enter these parameters for the 10 wavelengths
used to compute the atmospheric signalwhich are:
0.400, 0.488, 0.515, 0.550, 0.633, 0.694, 0.860, 1.536, 2.250, 3.750
For IAER=9 Modified gamma distribution
r
min
,r
max
, b,
r
n
(
j
), j=1,10
r
i
(
j
), j=1,10
where , b, and are the parameters of the Modified Gamma size distribution
For IAER=10 Junge Power-Law distribution
r
min
,r
max

r
n
(
j
), j=1,10
r
i
(
j
), j=1,10
where is the parameter of the Junge Power-Law size distribution
For IAER=11 Sun Photometer distribution (50 values max)
irsunph
for k=1 to irsunph enter: r and dV/dlogr
r
n
(
j
), j=1,10
r
i
(
j
), j=1,10
where irsunph is the number of value and dV/dlogr is usually profided by sunphotometers.
6S User Guide Version 2, July 1997
119
Table 1: Microphysical characteristics of the aerosol type (dry particles)
used for the comparisons shown Tables 3 and 4 (from WMO-WCP112).
Dust-Like Water Soluble Oceanic Soot
r
Mi
(m) 0.500 0.0050 0.30 0.0118

i 2.990 2.990 2.51 2.00

Table 2: Complexe refractive indexes of the aerosol types (dry particles)
used for the comparisons shown Tables 3 and 4 (from WMO-WCP112).
Dust-Like Water Soluble Oceanic Soot
(m) n
r
n
i
n
r
n
i
n
r
n
i
n
r
n
i
0.400 1.530 8.00E-3 1.530 5.00E-3 1.385 9.90E-9 1.750 0.460
0.488 1.530 8.00E-3 1.530 5.00E-3 1.382 6.41E-9 1.750 0.450
0.515 1.530 8.00E-3 1.530 5.00E-3 1.381 3.70E-9 1.750 0.450
0.550 1.530 8.00E-3 1.530 6.00E-3 1.381 4.26E-9 1.750 0.440
0.633 1.530 8.00E-3 1.530 6.00E-3 1.377 1.62E-8 1.750 0.430
0.694 1.530 8.00E-3 1.530 7.00E-3 1.376 5.04E-8 1.750 0.430
0.860 1.520 8.00E-3 1.520 1.20E-2 1.372 1.09E-6 1.750 0.430
1.536 1.400 8.00E-3 1.510 2.30E-2 1.359 2.43E-4 1.770 0.460
2.250 1.220 9.00E-3 1.420 1.00E-2 1.334 8.50E-4 1.810 0.500
3.750 1.270 1.10E-2 1.452 4.00E-3 1.398 2.90E-3 1.900 0.570
6S User Guide Version 2, July 1997
120
Table 3: Comparison between 6S and WMO-WCP112
for a Continental model (normalized value-dry particles)
K
ext
K
sca

0 g
6S WMO 6S WMO 6S WMO 6S WMO
0.400 1.40 1.40 1.27 1.27 0.902 0.901 0.643 0.646
0.488 1.14 1.14 1.03 1.03 0.900 0.898 0.637 0.640
0.515 1.08 1.08 0.967 0.967 0.899 0.897 0.635 0.638
0.550 1.00 1.00 0.893 0.891 0.893 0.891 0.634 0.637
0.633 0.849 0.849 0.755 0.754 0.890 0.888 0.629 0.633
0.694 0.760 0.760 0.671 0.669 0.881 0.879 0.628 0.631
0.860 0.577 0.577 0.487 0.486 0.844 0.841 0.629 0.633
1.536 0.282 0.283 0.212 0.212 0.753 0.750 0.641 0.645
2.250 0.150 0.151 0.115 0.115 0.765 0.761 0.738 0.741
3.750 0.101 0.103 0.0796 0.0805 0.790 0.785 0.777 0.779
Table 4: Comparison between 6S and WMO-WCP112
for an urban model (normalized value-dry particles)
K
ext
K
sca

0 g
6S WMO 6S WMO 6S WMO 6S WMO
0.400 1.48 1.48 0.980 0.976 0.664 0.660 0.600 0.600
0.488 1.16 1.17 0.766 0.762 0.658 0.654 0.594 0.593
0.515 1.09 1.09 0.715 0.711 0.655 0.651 0.592 0.592
0.550 1.00 1.00 0.651 0.647 0.651 0.647 0.591 0.591
0.633 0.828 0.829 0.535 0.532 0.646 0.641 0.587 0.587
0.694 0.733 0.733 0.466 0.462 0.635 0.631 0.585 0.585
0.860 0.542 0.542 0.322 0.319 0.593 0.588 0.584 0.583
1.536 0.242 0.243 0.111 0.111 0.460 0.455 0.564 0.565
2.250 0.123 0.124 0.0428 0.0426 0.347 0.342 0.583 0.585
3.750 0.0647 0.0659 0.0177 0.0181 0.274 0.274 0.579 0.587
6S User Guide Version 2, July 1997
121
Figures 01: Examples of a a
n
(m,x) function.
0. 0
0. 20
0. 40
0. 60
0. 80
1. 0
1. 2
0 10 20 30 40 50 60 70
Re[a
n
(m,x=10)]
Re[a
n
(m,x=50)]
R
e
[
a
n
(
m
,
x
)
]
Order n
m = 1.33 - i 0.001
-0. 60
-0. 40
-0. 20
0. 0
0. 20
0. 40
0. 60
0. 80
0 10 20 30 40 50 60 70
Im[a
n
(m,x=10)]
Im[a
n
(m,x=50)]
I
m
[
a
n
(
m
,
x
)
]
Order n
m = 1.33 - i 0.001
Figures 02: Examples of a b
n
(m,x) function.
0. 0
0. 20
0. 40
0. 60
0. 80
1. 0
1. 2
0 10 20 30 40 50 60 70
Re[b
n
(m,x=10)]
Re[b
n
(m,x=50)]
R
e
[
b
n
(
m
,
x
)
]
Order n
m = 1.33 - i 0.001
-0. 60
-0. 40
-0. 20
0. 0
0. 20
0. 40
0. 60
0. 80
0 10 20 30 40 50 60 70
Im[b
n
(m,x=10)]
Im[b
n
(m,x=50)]
I
m
[
b
n
(
m
,
x
)
]
Order n
m = 1.33 - i 0.001
Figures 03: Examples of spherical Bessel functions j
n
(x) and n
n
(x)
-0. 15
-0. 10
-0. 05
0. 00
0. 05
0. 10
0. 15
0 10 20 30 40 50 60 70
j
n
(x=10)
j
n
(x=50)
j
n
(
x
)
Order n
-0. 20
-0. 15
-0. 10
-0. 05
0. 00
0. 05
0. 10
0. 15
0. 20
0 10 20 30 40 50 60 70
n
n
(x=10)
n
n
(x=50)
n
n
(
x
)
Order n
6S User Guide Version 2, July 1997
122
Figure 04: Examples of a D
n
(x) function.
-20.00
-15.00
-10.00
-5. 00
0. 00
5. 00
10. 00
15. 00
20. 00
0 20 40 60 80 100
Re [D
n
(x=10)]
Re [D
n
(x=50)]
R
e

[
D
n
(
x
)
]
Order n
Figures 05: Examples of a D
n
(mx) function.
-10.00
-5. 00
0. 00
5. 00
10. 00
0 20 40 60 80 100
Re [D
n
(m,x=10)]
Re [D
n
(m,x=50)]
R
e

[
D
n
(
m
x
)
]
Order n
m=1.35 - i 0.001
10
- 3
10
- 2
10
- 1
10
0
10
1
10
2
0 20 40 60 80 100
Im [D
n
(m,x=10)]
Im [D
n
(m,x=50)]
I
m

[
D
n
(
m
x
)
]
Order n
m=1.35 - i 0.001
Figures 06: Examples of a G
n
(x) function.
-1. 40
-1. 20
-1. 00
-0. 80
-0. 60
-0. 40
-0. 20
0. 00
0 20 40 60 80 100
Re[ G
n
(x=10)]
Re[ G
n
(x=50)]
R
e
[
G
n
(
x
)
]
Order n
-1. 20
-1. 00
-0. 80
-0. 60
-0. 40
-0. 20
0. 00
0 20 40 60 80 100
I m[ G
n
(x=10)]
I m[ G
n
(x=50)]
I
m
[
G
n
(
x
)
]
Order n
6S User Guide Version 2, July 1997
123
Figures 07: Examples of functions
n
() and
n
() for n=1 to 6.
- 5
0
5
10
15
20
25
0 30 60 90

n
Scattering Angle
1
2
3
4
5
6
-20
-10
0
10
20
0 30 60 90

n
Scattering Angle
1
2
3
4
5
6
Figure 08-a: Junge Power-law function: Model C (Deirmendjian, 1954)
10
-10
10
-7
10
-4
10
-1
10
2
10
5
10
-4
10
-3
10
-2
10
-1
10
0
10
1
10
2
d
N
(
r
)
/
d
r
r (m)
Model C (Deirmendjian 1969)
Figure 08-b: Modified Gamma distribution function: Volcanic Ash (WCP 112)
10
-10
10
-7
10
-4
10
-1
10
2
10
5
10
-4
10
-3
10
-2
10
-1
10
0
10
1
10
2
d
N
(
r
)
/
d
r
r (m)
Volcanic Ash
(WMO/WCP 112)
6S User Guide Version 2, July 1997
124
Figure 08-c: Log-Normal distribution function: Continental Model (WCP 112)
10
-10
10
-7
10
-4
10
-1
10
2
10
5
10
-4
10
-3
10
-2
10
-1
10
0
10
1
10
2
d
N
(
r
)
/
d
l
o
g
r
r (m)
Dust Like
Soot
Water Soluble
Continental Model
(WMO/WCP 112)
Figure 09: Same as Figures 8 but represented for dV/dlogr
10
- 6
10
- 5
10
- 4
10
- 3
10
- 2
10
- 1
10
0
10
1
10
- 3
10
- 2
10
- 1
10
0
10
1
10
2
Model C (Deirmendjian 1954)
Volcanic Ash (WMO/WCP 112)
Continental Model (WMO/WCP 112)
d
V

/

d
l
o
g
r
r (m)
6S User Guide Version 2, July 1997
125
Figure 10: Phase function (dry particle) as computed by the MIE subroutine and
by the one generated by AEROSO subroutine (exact case).
0.1
1
10
100
0 30 60 90 120 150 180
Continental Model
From AEROSO (Exact Case)
P
h
a
s
e

f
u
n
c
t
i
o
n
Scattering angle
3.750m
0.860m
0.440m
From MIE
Figure 11: Phase function as computed by MIE subroutine using the dV/dlogr provided
by a sunphotometer CIMEL during the SCAR-A experiment (Hog Island, July 11, 1993).
0.1
1
10
100
0 30 60 90 120 150 180
441 nm
873 nm
P
h
a
s
e

f
u
n
c
t
i
o
n
Scattering Angle
6S User Guide Version 2, July 1997
126
REFERENCES
A.L. ADEN, Electromagnetic Scattering from Spheres with Sizes Comparable to the Wavelength. J.
Appl. Phys., 22, 5, 601-605, 1951
F.J. CORBATO and J.L. URETSKY, Generation of Spherial Bessel Functions in Digital
Computers. J. Assoc. Computing Machinery, 6, 366-375, 1959
G.A. D'ALMEIDA, P. KOEPKE, and E.P. SHETTLE, Atmospheric Aerosols Global Climatology
and Radiative Characteristics, A. Deepak Publishing, Hampton, 1991.
J.V. DAVE, Scattering of Visible Light by Large Water Spheres. Appl. Opt., 8, 1, 155-164, 1969
C.N. DAVIES, Size distribution of atmospheric particles. J. Aerosol Sci., 5, 293-300, 1974
D. DEIRMENDJIAN, Scattering and Polarization properties of water clouds and hazes in the
visible and infrared. Appl. Opt., 3, 187-196, 1964
D. DEIRMENDJIAN, R. CLASEN, and W. VIEZEE, Mie Scattering with Complex Index of
Refraction. J. Opt. Soc. Am., 51, 6, 620-633, 1961
D. DEIRMENDJIAN, Electromagnetic scattering on spherical polydispersion. Elsevier Ed. New-
York, 1969
J.E. HANSEN and L. TRAVIS, Light scattering in planetary atmospheres. Space Sci. Rev., 16, 527-
610, 1974
L. INFELD, The Influence of the Width of the Gap upon the Theory of Antennas. Quart. Appl.
Math., 5, 2, 113-132, 1947
C.E. JUNGE, Gesetzmigkeiten in der Droenverteilung atmosphrischer Aerosole ber dem
Kontinent. Ber d. Deirsch Wetterdienst U.S.-Zone, 35, 261-277, 1952
G.W. KATTAWAR and G.N. PLASS, Electromagnetic Scattering from Absorbing Spheres. Appl.
Opt., 6, 8, 1377-1382, 1967
K.N. LIOU, An Introduction to Atmospheric Radiation, Academic Press Inc., San Diego, 1980.
G. MIE, Beigrade zur Optik trber Medien, speziell kolloidaler Metallsungen. Ann. Physik., 25,
377-445, 1908
E.P. SHETTLE and R.W. FENN, Models of atmospheric aerosols and their optical properties, in:
Optical Properties in the Atmosphere, AGARD-Cp-183, NTIS, ADA 028615, 1976.
E.P. SHETTLE and R.W. FENN, Models for the aerosol of the lower atmosphere and the effect of
humidity variations on their optical properties, AFGL-TR-79-0214, Environmental Research
Paper No 675, NTIS, ADA 085951, 1979.
H.C. VAN DE HULST, Light Scattering by Small Particles, Dover Publications, New York, Dover,
1981.
6S User Guide Version 2, July 1997
127
WORLD CLIMATE PROGRAMME, WCP-55, Report of the expert meeting on aerosols and their
climatic effects (Eds. A. Deepak and H.E. Gerber), World Meteorological Organization,
Geneva, 1983.
WORLD CLIMATE PROGRAMME, WCP-112, A preliminary cloudless standard atmosphere for
radiation computation, World Meteorological Organization, WMO/TD-No 24, Geneva, 1986.
P.J. WYATT, Scattering of Electromagnetic Plane Waves from Inhomogeneous Spherically
Symmetric Objects. Phys. Rev., 127, 5, 1837-1843, 1962
6S User Guide Version 2, July 1997
128
SUBROUTINE ODA550
Function: To compute the extinction cross section and the aerosol optical depth at = 550
nm from the vertical distributions of the particle density (in particules/cm
3
).
Description: We have considered the 2 profiles, suggested by Mc Clatchey et al (1971),
corresponding to a visibility of 23 (clear) and 5 km (hazy) at ground level. The total numbers of
aerosols for the clear atmosphere have been adjusted so that the total extinction coefficient at =
550 nm becomes identical to the values used by Elterman (1964).
This total extinction coefficient K (in km
-1
) is obtained from
K
550
(z)
550
10
3
N(z)
where s is the extinction cross section in mm2 and N(z) the particules density (in part/cm3) (the
factor 10
-3
is to obtain an extinction coefficient in km
-1
). was computed with the same aerosol
model as the one defined by Mc Clatchey, index of refraction equal to 1.50 and size distribution
similar to Deirmendjian's model "C" (1969) (cut off has been extended from 5 to 10 m). The
computed value of
550
is 0.056032.
The optical thickness is defined by

550
K
550
(z) dz
0
+

We obtain the optical thicknesses at 550 nm, 0.235 and 0.780 respectively for the two
standard visibilities 23 and 5km. For another visibility, we compute a new profile particle density
from those defined for 23 and 5km. The calculations were made using the following interpolations,
N(z)
a(z)
VIS
+ b(z)
6S User Guide Version 2, July 1997
129
We obtain for example :
= 0.152 for V = 50km,
= 0.520 for V = 8km.
Reference :
R.A. Mc CLATCHEY, R.W. FENN, J.E.A. SELBY, F.E. VOLZ, J.S. GARING. Optical properties
of the Atmosphere (revised), AFCRL 71-0279, Env. Research Paper 354, Bedford,
Massachusetts, U.S.A., 1971.
D. DEIRMENDJIAN, Electromagnetic Scattering on Spherical Polydispersions , 1969.
L. ELTERMAN , Rayleigh and Extinction Coefficient to 50km for the region 0.27m to 0.55m.
Applied Optics, 10, 1139-1145, 1964.
6S User Guide Version 2, July 1997
130
SUBROUTINE ODRAYL
Function: To compute the molecular optical depth as a function of wavelength for any
atmosphere defined by the pressure and temperature profiles.
Description: The optical depth is written

(z) dz
0
+

where

(z) is the molecular extinction coefficient at altitude z and for wavelength . It can be
obtained from

(z)

N
r
(z)10
5
,
with N
r
(z) is the molecules number/cm
3
at altitude z, and

the extinction (or scattering)

cross section in cm
2
.
These two quantities are defined by

8
3
(n
s
2
1)
2
3
4
N
s
2
6 + 3
6 7
and
N
r
(z) N
s
P(z)
1013.25
273.15
T(z)
where P(z) and T(z) are respectively the pressure and the temperature at the altitude z. Recall that n
S
is the air refractive index, N
S
the molecular density at z=0 in STP conditions, and the molecular
depolarization factor.
We have taken:
* for refractive index
(n
s
1).10
8
8342.13 +
2406030
130
2
+
15997
38.9
2
6S User Guide Version 2, July 1997
131
where is the frequency in cm
-1
* N
S
= 2.54143.10
19
* and the depolarization factor = 0.0279 following Young's (1980).
This depolarization factor is also used to compute the Rayleigh phase function (see routine
CHAND.f) according to:
P()
3
4
1
1 + 2
(1+ cos
2
) +
3
1 + 2
where is the scattering angle, and = /(2-).
References :
B. EDLEN, The refractive Index of Air. Meteorologia, 2, 71-80, 1966
L. ETERMAN, Rayleigh and Extinction Coefficients to 50 km for the Region 0.27 m to 0.55 m
Appl. Opt., 10, p. 1139-1145, 1964
D.V. HOYT, A redetermination of the Rayleigh Optical Depth and its Application to selected Solar
Radiation Problems, J. Appl. Meteor, 16, p. 432-436, 1977.
A. T. YOUNG, Revised Depolarization Corrections for Atmospheric Extinction. Applied Optics,
19, 3427-3428, 1980
6S User Guide Version 2, July 1997
132
SUBROUTINE OS
Function: Compute the atmospheric intrinsic reflectance for the case of either satellite or
aircraft observation. Also compute the downward radiation field needed by the integral formula of
and ' (see Chapter I, 2.5.1, Eqs. 25 and 26) used in the computation in case of a non
lambertian target.
Description: The general purpose of the successive order of scattering is to solve
numericaly the equation of radiative transfer for upward (Eq. 01.) and downward radiation (Eq. 02.)
at any optical thickness . If
1
is the total optical thickness and the cosine of the view angle, then
we can write:
I(;, ) I(
1
;, )
(
1
)/
e + J(' ; , )
(' )/
e
d'

(1 0) (01)
I(; , ) I(0; , )
/
e + J( ' ;, )
(' )/
e
d'

(1 0) (02)
where the source function, J(;,) accounts for the interaction of the present radiation field with
the particles of the layer located at , so that:
J( ;, )

0
4
I(;' , ' ) P(, ;', ' ) d' d'
1
1

0
2

0
4
F
0
P(, ;
0
,
0
)
/
0 e (03)
The second term of equation (03) represents the sun source F
0
transversing the path along
(
0
,
0
) directly to the level and then being scattered in direction (,) (primary scattering).
To solve this differential equation, one has to fix boundary conditions which are:
I(0;,) =0 (04)
I(
1;
,) = 0 (05)
6S User Guide Version 2, July 1997
133
These express the fact that there is no diffuse downward nor upward radiation at the top and the
bottom of the finite atmosphere.
The convention is to describe the atmosphere with the top at =0 and the bottom at =
1
. The
upward radiation correspond to + and the downward to with (1>0) as depicted by figure 1.
Figure 1: Schematic view of the radiative transfer problem for a plane parralel atmosphere.
Top
Bottom
= 0
= 0
=
1
I(0;,)
I(
1
;,)
I(;,)
I(
1
;-,)
I(;-,)
The successive orders of scattering consist of solving numerically the equation of transfer by
iteration. First, the equation is solved for each layer considering only the primary scattering
radiation (one interaction between the source (sun) and the atmosphere), giving for Eqs. (01) and
(02):
I
(1)
(; , )

0
4
F
0
P(, ;
0
,
0
)
/
0 e (1 0) (06)
I
(1)
(; , )

0
4
F
0
P(, ;
0
,
0
)
/
0 e (1 0) (07)
6S User Guide Version 2, July 1997
134
Then for higher order of scattering we write:
I
(n)
(
j
; , )
1

J
(n)
(
j
;, )
(
j
)/
[ ] e
i j
p

(08)
I
(n)
(
j
; , )
1

J
(n)
(
j
;, )
(
j
)/
[ ] e
i 1
j

(09)
where p represents the number of layers used for the decomposition of the atmosphere,
j
the optical
thickness at level j and the increment in optical thickness between two successive layer. J
(n)
is
computed from I
(n-1)
by:
J
(n)
(;, )

0
4
I
(n)
(;', ' )P( , ;' , ' )d' d'
1
1

0
2

(10)
In the code a numerical integration of Eq. 10 is performed using the decomposition in Fourier series
(for ), the legendre polynomial and Gaussian quadrature (for ).
6S User Guide Version 2, July 1997
135
SUBROUTINE SCATRA
Function: To compute the scattering transmission functions for the three atmospheric models,
rayleigh, aerosol and a mixture of both on the two paths (downward and upward). We also compute
the spherical albedo.
Description: As in ATMREF.f, we have to compute the transmission function and albedo in
three different cases and three sensor configurations. Again, the accuracy of the computation of 6S
was a concern, so the approximation adopted in 5S has been replaced by using the sucessive order
of scattering method (ISO.f) for the aerosol case and the mixed case, or when the sensor was inside
the atmosphere on board an aircraft. For the Rayleigh atmosphere we used an accurate analytical
formula which has sufficient accuracy and enables us to save computer time. The formula is
explicity coded into SCATRA.f for the transmission and call CSALBR.f for the albedo. For ground
measurements, the upward transmission is set to 1.0 and the spherical albedo to 0.0, because we
neglect the atmosphere between the sensor and the target.
We only give here the formula of the Rayleigh transmission which is based on the two stream
method adapted to the case of a single scattering albedo equal to 1.0 (Rayleigh case). The total
transmission on path of length , T() can be approximated by:
T()
[(2/ 3) +] +[(2/ 3) ]e

(4 / 3) +
R
(01)
where tR is the Rayleigh optical thickness.
Figure 1 compares the accuracy of Eq. 01 to the "exact" computation (Successive Orders of
Scattering).
6S User Guide Version 2, July 1997
136
References
J.H. JOSEPH, W.J. WISCOMBE, J.A. WEINMAN, The Delta-Eddington Approximation for
Radiative Flux Transfer,.J. Atmos. Sci. 33, p.2242-2459, 1976.
W.E. MEADOR, W.R. WEAVER, Two-Stream Approximations to Radiative Transfer in Planetary
Atmospheres: a Unified Description of Existing Methods and a New Improvement, J. Atmos.
Sci. 37, p. 630-643, 1980.
R.H.WELCH, W.G. ZDUNKOWSKI, Back Scattering Approximations and their Influence on
Eddington-Type Solar Flux Calculation, Beitr. Phys. Atmosph., 55, no 1, p. 28-42, 1982.
W.G. ZDUNKOWSKI, R.M. WELCH, G. KORB, An Investigation of the Structure of Typical
Two-Stream Methods for the Calculation of Solar Fluxes and Heating Rates in Clouds, Beitr.
Phys. Atmosph.53, no 2, p. 147-166, 1980.
Figure 1: Accuracy of Eq. 01
0.00
0.20
0.40
0.60
0.80
1.00
0.000
0.002
0.004
0.006
0.008
0.010
0 10 20 30 40 50 60 70
exact results
analytical expression
absolute difference
Zenith observation angle ()
T
r
a
n
s
m
i
s
s
i
o
n

f
u
n
c
t
i
o
n

T
(
)
A
b
s
o
l
u
t
e

d
i
f
f
e
r
e
n
c
e
6S User Guide Version 2, July 1997
137
SUBROUTINE TRUNCA
Function: decompose the aerosol phase function in series of Legendre polynomial used in
OS.f and ISO.f and compute truncation coefficient f to modify aerosol optical thickness and single
scattering albedo
0
according to:
' (1
0
f)

0
'

0
(1 f)
(1
0
f)
References
J. LENOBLE, Radiative Transfer in Scattering and Absorbing Atmospheres: Standard computional
procedures, pp 83-84, A. Deepak Publishing, 1985