133
reads

Hsien-Wei Chen

2011-06-10
13:50:00 - 14:50:00

307 , Mathematics Research Center Building (ori. New Math. Bldg.)

Topography effects, which relate to the sun direction, elevations, and atmospheric conditions, usually extremely limit the applications of remote sensing images. Therefore, it is important to find a correction method for eliminating topography effects. However, the reflectance of ground surface, which contains only the ground information, is the only parameter that does not conduct terrain information and atmospheric conditions. Therefore, a perfect reflectance estimation result will automatically lead to a perfect topography correction. This study extends the radiometric control area base path radiance estimation algorithm into a radiometric control area base surface reflectance estimation algorithm. In this approach, topography effects are characterized by the ground slope and the shape factor. For the estimation of ground slope, a straightforward but effective calculation is applied. However, the estimation of shape factor, which is the essential part of this study, is much more complicated. First, this study will show that the downwelled radiance is not homogenous at all. Therefore, the solid angle calculation for estimating shape factor is not ppropriated. Second, this study successfully uses covariance matrix decomposition techniques in modeling the approximated linear relationship between shape factor and elevation differences. Therefore, the connection between physical machination and statistical model can be constructed. Third, this study defines the term shaded ground sample, and proves that the shaded ground samples are essentially carrying the information of shape factor. Therefore, the inference for shape factor can be made from shaded ground samples. Forth, although the estimation of shape factor is base on statistical assumption, the extremely large sample makes the estimator with little uncertainty, i.e., this study is essentially using statistical concepts to establish a government equation for shape factor without any uncertainties. Therefore, the reflectance estimation also has little uncertainty. The final result shows no topography effects in the reflectance image. Also the comparison between the scaled digital numbers and the reflectance shows the correction effects of this method. In addition, the minimum requirement for the method is digital image and DTM data has a same spatial resolution.