Since we have got the coordinates of the character points and 
their corresponding ground coordinates, so we can evaluate the 
geolocation accuracy using the following formula: 
AX 
AX 
n 
= the coordinates difference in X direction 
= the coordinates difference in Y direction 
= the number of the check points 
RMSEx = 
RMSEy = 
IA Y* 
RMSE = yj RMSEx 2 + RMSEy 2 
Where 
RMSE = Root Mean Square Error 
RMSEx = Root Mean Square Error in X direction 
RMSEy = Root Mean Square Error in Y direction 
2.3 Result 
The ASTER Level-1A data of research area was observed at 
31 st August, 2004, and the Level-3 A data was generated at 21 st 
December, 2005 by Earth Remote Sensing Data Analysis 
Center (ERSDAC). They orthorecitified the same ASTER 
Level-1A image again at 18 th October, 2006. These 2 ASTER 
Level-3A images were named old version and new version to 
distinguish them, and the geolocation accuracy was evaluated 
respectively. 
Figure 2-3 the geolocation accuracy of ASTER Level 3-A image 
We can see the geolocation accuracy form figure 2-3, the arrow 
showed the displacement of character point corresponding to 
the ground check point, the start of the arrow is the coordinates 
of ground check point, and the end is the coordinates of image 
character point. The value of the geolocation error was 
extended 1000 times to show clearly. The black arrow showed 
the old version’s geolocation accuracy and the red arrow 
showed the new version’s. We can see that the new version can 
locate more precise than the old version. 
The Mean, Standard Deviation(Stdev.) and Root Mean Square 
Error(RMSE) of the character points are listed in the below 
table: 
Old Version 
New Version 
AX(m) 
AY(m) 
AX(m) 
AY(m) 
Mean 
-5.52 
13.65 
-0.70 
3.56