130
50
H
TT
TV-
TF
Hi
P?
ip
t+tt
i
:r.i:
TTÎT
TTi-t:
TTÎT
TTiT
ii
TT
TV*
A):
;i
-t.-fTT
•ijH
T-fTT
lii'.Î.T:
-;-T-TT
itïîTT
i
-*- 4 -
•t-i*
l- 4 --
.....
-;-v 4 r.
•h 4 +*
.....T
•î*î 4 t*
H
it
44 *
■TV
Ü
-T-V-f-T-
■TrU
T-V-t-T-
TV-T*
TT?-
ilii
.i-i-.t.i.:
TT?-Î-
.TTt-T-'
•TT-"-;-
i
•U;
TV
1
'.t.V
T-H
T-j
T
T
tiff
-:-TTV-
HH
Üv
!jf
'* 7 *T 4 t
Ü
i
I’i*
g
T
é
t.T
T-
VT
Hi
::E
ttH’
iti
Üi
i?!?TT
i
tt
H
T
il
:
■T-v-tt.;
gg
H;
•TV;j/î\
TT?-?-
iii
ttTT
't
iL
T
TV-t-T;'
Ìli:
T-V-t-T*
'Ml
Ili
T-V-t-T;
ifii
±
0 20 40 60 80 100
Alb*do(X)
(a) Albedo of 20% and visibility of 20km
(b) Albedo of 70% and visibility of 30km
Figure 2. Intersection of contours under ideal conditions
0 20 40 60 80 100
Alb«io<X)
iv
.t.T.+T- ‘
ü
Tü
Üi
lîH
-.T T vii'
T
-f.T-4-
-j-fÎT-
T
HT?-?-"
.TV.t.ï.
TTl'.r:
'V
-4-4--
: ■ 4- 4* -
.;.*-4-4--
-r-t-rr-
TT
TT
TT?-?"*
Ì-TT
•T-T-
-T-TT
'.T.v.TT
’-•-•TT
'.t.t.TT
-|--'-TT
ir'v'fT
T-TT
ill
•-T.-fTT
1
i:
.t.VTT
■-Ì-TV-
ijH
TV-t.T
-T-V-|-->
T-V-tT
TT?--**
iiÜT-T
TT**-*-"
.iiTT
TT?---*
Ì-T.‘
TV-t-7-'
■'T'-Ìt
-tirr
r.r.tTi
-4-Tt-T-
vr.iVT
TT
ii:
- 4-4f
VM
t:t.Tiv
.....<-4--
TÉ-
Hi
s
■t-4-i-*-
Hii
TTt't'j
i
-H-
Titi
H-h
ìli'
1 1
n-.i-.t.-f;
Sii!
TV.Î.T
TT?-?-"
ü
T
: t-T'
T
T-r-T-T-
tt.TT
•j.iTT
•T.ÌTT
±
ü
T}.-TV-
-.t.ÌTT
Tt-TT
-.Î.ÎT+-
TiTT
Tt-VT
lï
Trj
TT?-?-*
ür
TV-t-îv
Pi
irti
Ih
Pi
•T-V-t-T;
T-V-t-T.
T-V-t-lT
T-V-t-t.
1
0 20 40 80 80 100
Mbado(X)
(b) Albedo of 20%, urban aerosol
model with visibility 10km
(a) Nadir reflectance =20%,
along-track reflectance = 22.5%
Figure 3 . Intersection of contours resulting in erroneous parameter retrieval
A
Figure 4
3.2. Sensitivity to
The atmospheric c
he boundary layer;
properties of the a
single scattering al
aerosol compositio
meteorological co
boundary layer cor
layer aerosols from
Four standard atm
urban, desert and
generated. To expl
atmosphere radiam
and visibility cons
ground visibility,
radiance surfaces f
visibility of 32.96k
simulated data, a question not addressed therefore is sensitivity to inexact forward modelling of the atmosphere.
Errors due to the simplified two-stream multiple scattering approximation used by LOWTRAN-7 may be
particularly significant. However a more exact multiple scattering approximation is recommended for
operational use. Sources of error resulting from non-tropospheric atmospheric variation, such as sub-pixel cloud
are not explored here.
3.1. Sensitivity to non-Lambertian reflectance
The atmospheric correction method makes the assumption that the surface reflectance is Lambertian in all
channels. Sensitivity of retrieval of albedo and visibility to non-Lambertian reflectance was investigated using
simulated top-op-atmosphere radiances. Figure 3a shows the superimposition of contours formed on the nadir
and along-track radiance surfaces for channel 1 by a surface which has reflectance of 20 % at nadir and a
reflectance of 22.5% at 50° for a visibility of 10km. It can be seen that the nadir contour passes through the
point (20,10) while the along-track contour passes through the point (22.5,10). The intersection of these two
contours is far from the correct values, approximately at the point (11.2,3.1). The limitation of the method is
underlined, as the result shows that a Lambertian surface with reflectance 11.3 viewed at nadir and 50° through
an atmosphere of visibility 3.1km has the same appearance as the non-Lambertian surface viewed through a
10km atmosphere. Figures 4a and 4b show the parameters retrieved from top-of-atmosphere radiances from a
surface with nadir reflectance 20% and along-track reflectance varying from 17.5% to 22.5% for an atmosphere
of 10km visibility. Further results on non-Lambertian surfaces are given in section 3.4..
The results of parai
in table 1 , and for;
be reasonably robi
aerosol model. An
atmosphere radian«
albedo. The retriev
this paper, the var
aerosol types throu
33. Sensitivity to i
For the sensitivity
channels will be fr
although the indiv
stipulated at 20:1 ;
signal noise. Howe
atmospheric correc
range equivalent u
show R.M.S. error
channel dynamic n
