International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part Bl. Istanbul 2004 
SWMT € ch.24 250m * ch.20 
a & ch.25 
* ch.21 
¥ € ch.26 : Sh 
e. ch.27 u ch.22 
-025 à ch.30 05 | ch.23 
ch. 34 à ch.28 
x x = t 
® + A ch.35 S = À ch.29 
> Ach36. 5 ° 
a 05 = 0 à 
e £c 
9 3 2 
© A «© 
-0.75 E -05 
9 
zi zi 
zi -075 -05 -025 0 7-05 -025 0 0.25 0.5 
cross track cross track 
Figure 6 Geometric Accuracy after the Correction of 
Sat/GLI Alignment (mean error, unit: pixel) 
2.4 Correction of Difference of Mirror A/B 
Even though the corrections mentioned above are done, the 
errors for mirror A and B are different from each other. The 
parameter “A/B” is only related with the difference of scan 
mirror A and B. So this parameter should be corrected on next 
step. The difference of mirror A/B side is expressed by the 
angle (81. 02) around the scanning axis. The angle is defined by 
relative values to side A. In other words, the effect of those 
parameters appears only on the image scanned by the mirror B. 
Such a difference has the effect of distortion on the image as 
shown Figure 7. 
Similar to the equation (5), the observation equations are 
written as follows. The correction values as shown in Table 6 
were acquired by the orientation using the GCP data set (Q3). 
  
  
  
  
  
  
  
  
  
  
D OF. OF. 
EF. s m € À — e À = 0 
00, 00, (6) 
à GF. or, 
FC py —————Ó n 
: : 00, co, 
Mirror A Mirror B B-A 
ch23 | -026 .058 ‚017 2702 .044 -.760 
ch 28 | -116 .091 2.072 2.672 .044 = 763 
ch. 29:1 - 135 .065 -.084 - 613 .052 -.678 
  
  
  
  
  
  
  
  
Table 6 Difference of Errors for Mirror A/B 
(partial, used GCP: Q3, unit: pixel) 
  
  
  
  
  
Figure 7 Effect of Mirror A/B Difference on Image 
     
   
    
    
   
   
   
    
  
   
  
   
   
    
    
  
  
  
   
    
  
    
   
   
    
  
  
   
   
  
    
   
  
  
  
    
  
GCP ài | 5 
Q3 AR 
  
  
  
  
  
Table 7 Corrected Mirror A/B Difference (arcsec.) 
2.5 Geometric Accuracy after All Corrections 
Through up to section 2.4, the parameters of Sat./GLI, detector 
address, the difference of Mirror A & B concerning the view 
vector were corrected. The geometric accuracy after all 
corrections was evaluated using GCP data sets of Q4 and K3. 
The results of some channels are shown in Table 8. In the table, 
the accuracy was evaluated by the R.M.S.E., mean errors and 
the difference between the mean errors from mirror A/B. As an 
example, the error distribution of ch.23 is shown in Figure 8. 
The Table 8 and Figure 8 indicate that the error differences of 
Mirror A/B were decreased specially in pixel direction. The 
geometric accuracy (16) by system correction will be expected 
about 1 pixel for 250m channel and less than one pixel for 1km 
channel image. 
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
ch FAB CP AE V. V 
A T3550 | 0993 : 0739 = 1187 
ch.23 T T 
B 38.1 0951 10792 1.222 
250 page [oda 225 0.3811. 0.754 ! 1.159 
m er B 226 | 0.931 0.783 1217 
aeu 186 | 0836 | 0.706 . 1.095 
en. B isl | 0868 0.789 | 1.173 
3o | ^ 556 | 0.592 | 0.587 0,333 
en. Bp | 502 | n5i1 0478 0.700 
Ke] co | À soi 1 03558 043%... 0.73% 
eit B | 551 | 0515 0516 0.720 
ion ^ 377 | 0510 0543 | 0.745 
en. B 331 0.500 | 0.600 | 0.780 
(a) R.M.S.E 
ch A/B Vi. Vp B-A 
A 0400 0.029 
ch 23 B -0.040 -0.033 .060 i -.062 
250 A | -0.066 0.003 
= ch.28 B -0.107 -0.020 .041 - 023 
A | 0.036- 0.101 
ch.29 B -0.083 -0.094 -.047 .007 
A | 05924 0.015 
c9 don ds e salu. 05 
lkm A | 0045 0.016 ^ 
ato 0.009 DOSE TOM 
A 0.027 0.019 
eh | y onto |-041 1000 
  
  
  
  
  
  
  
  
  
(b) Mean Errors and Difference of Mirror A/B 
Table 8 Geometric Accuracy after All Corrections 
(used GCP set: Q4, K3, unit: pixel)