|
Mixture spectra are common in remote sensing due to the limitations of spatial resolution and the heterogeneity of land
surface. During the past 30 years, a lot of subpixel model have developed to investigate the information within mixture
pixels. Linear spectral mixture model (LSMM) is a simper and more general subpixel model. LSMM also known as
spectral mixture analysis is a widely used procedure to determine the proportion of endmembers (constituent materials)
within a pixel based on the endmembers' spectral characteristics. The unmixing accuracy of LSMM is restricted by
variety of factors, but now the research about LSMM is mostly focused on appraisement of nonlinear effect relating to
itself and techniques used to select endmembers, unfortunately, the environment conditions of study area which could
sway the unmixing-accuracy, such as atmospheric scatting and terrain undulation, are not studied. This paper probes
emphatically into the accuracy uncertainty of LSMM resulting from the terrain undulation. ASTER dataset was chosen
and the C terrain correction algorithm was applied to it. Based on this, fractional abundances for different cover types
were extracted from both pre- and post-C terrain illumination corrected ASTER using LSMM. Simultaneously, the
regression analyses and the IKONOS image were introduced to assess the unmixing accuracy. Results showed that
terrain undulation could dramatically constrain the application of LSMM in mountain area. Specifically, for vegetation
abundances, a improved unmixing accuracy of 17.6% (regression against to NDVI) and 18.6% (regression against to MVI) for R2 was achieved respectively by removing terrain undulation. Anyway, this study indicated in a quantitative
way that effective removal or minimization of terrain illumination effects was essential for applying LSMM. This paper
could also provide a new instance for LSMM applications in mountainous areas. In addition, the methods employed in
this study could be effectively used to evaluate different algorithms of terrain undulation correction for further study.
|
