The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B7. Beijing 2008 
65 
captured on undulating earth surfaces., These terrain 
illumination effects have serious consequences for the 
application of quantitative methods, including classification of 
land cover types and linear spectral analysis. So the main 
objective of this paper is to quantitatively evaluate the accuracy 
uncertainty of LSMM affected by the terrain undulation with 
the hope that this study could offer new theory basis for LSMM 
applications in mountainous area and furthermore improve the 
unmixing accuracy of LSMM. To this end, via taking unmixing 
vegetation abundance as an example, we choose the ASTER 
data equipped with relatively broad spectral range, fine spatial 
resolution, a large number of bands, and concentrated our study 
on part of suburb in Fuzhou, China. The following section is 
described to introduce the test site and the ASTER data 
followed by ASTER data processing. Then the C terrain 
undulation correction was used and subsequently LSMM was 
applied to both original ASTER data and C terrain undulation 
corrected ASTER image covering test site. Finally a regression 
analysis between vegetation abundances and NDVI, MVI 
extracted from originally or terrain corrected ASTER data 
using systematically sampling was implemented and the 
correlation coefficient as a indicator index was selected to 
quantitate the terrain undulate impact.