X 
y 
0 
f•tan( 0 ) 
(5) 
Here is introduced an example of' the results. The geometric 
correction of thermal infrared Image of VIMR (band 3) was done with 
18 GCPs and resampled into the pixel size of 20” in latitude and 30 
In longitude. Photo. 2 shows the surface temperature distribution 
of the east part of Japan. The image is represented by color level 
slicing. 
4. Geometric Correction of SEASAT SAR Imagery with DTM 
A radar imagery over mountainous areas has geometric distor 
tions owing to topographic effects since it is acquired at the big 
off-nadir angle and processed on the assumption that the ground 
surface is plane. Especially the distortion on the range direction 
is very big and cannot be ignored. The study on correction of these 
distortions of SEASAT SAR using elevation data has been carried out 
It is assumed that the earth is globe and the swath area of 
radar is plane as shown in Fig. 3. Topographic effects are examined 
on the reference plane B-C which is defined by the intersection B 
between the earth surface and the beginning point of swath and by 
the intersection C between the earth surface and the main axis of 
fan beam. The reference plane of uncorrected imagery is supposed to 
be the plane B-E which is normal to the vertical line from sensor 
and contains point B. The relation between corrected image coordi 
nate (u,v) and uncorrected one (u’,v') is expressed as below, where 
u and u’ is on the range direction and v and v' is on the azimuth 
direction. Let the origin of coordinate to be ( 0 ) in (u,v) and 
( 0' ) in (u',v’), respectively. 
r 2 = t*+ s 2 = ( T - z ) 2 + u 2 
= ( H + m ) 2 + u ' ‘ 
v = v ' 
(6) 
where 
r : slant range 
T : distance from sensor to plane B-C 
H : altitude of sensor 
z : ground height above plane B-C 
s : position on plane B-C corresponding to u’ on plane B-E 
m : difference between distance from sensor to plane B-E 
and altitude of sensor ( H ) 
The ground height (z) is prepared on a geographic coordinate 
system(x.y). The transformation functions between (u,v) and ( x,y ) 
is determined as polynomials by GCPs as below. 
u = f ( x, y ) (7) 
v = g ( x, y ) 
The several polynomials of different; order have been examined with 
35 GCPs. The results indicated that the range direction (f) could 
be sufficiently expressed by a binomial and the azimuth direction 
(g) could be by a monomial with accuracy of less than one pixel. 
Photo. 3 shows one of the results, which covers Pasadena area 
803