Portland State University. Department of Geography
Jiunn-Der (Geoffrey) Duh
Date of Award
Master of Science (M.S.) in Geography
Remote-sensing images, Image processing, Topographical surveying
Satellite imagery is commonly used to study land-use, land-cover change in mountainous areas. Classification of land-cover types is particularly difficult in this type of terrain because topography affects the spectral response of surface features. Several techniques have been developed to help compensate for this topographic effect and to increase classification accuracy.
The purpose of this study is to compare topographic correction methods to determine which produces the most accurate results. Method 1 does not use any topographic correction, serving as a control. Method 2 is based on the Lambertian model, and uses topographical data as variables in an equation. Method 3 adds an aspect band and a vegetation index band to the image prior to classification. Both of these latter methods produced significantly more accurate results than the control (method 1) in the classification of agricultural lands, but did not improve accuracy in forestland classification.
Rosen, David Michael, "Methods for Correcting Topographically Induced Radiometric Distortion on Landsat Thematic Mapper Images for Land Cover Classification" (2005). Geography Masters Research Papers. 12.