Performance Assessment of Different Topographic Correction Techniques over Subpixel Classification

Abstract

Over mountainous region, the topographic effects in the form of shadow due to rugged terrain are very common which imposes the negative impact on the applicability of satellite dataset. Since the past decade, different topographic correction methods have been developed and implemented over hard classifiers to compensate for these topographic effects. But the impact of different topographic corrections on subpixel classification is still needed to be addressed over undulating terrain like the Himalayas. The subpixel classification methods deliver the information at subpixel level from low and coarse resolution satellite imagery which can be utilized in mapping or monitoring of land use and cover changes. Previous literature has shown that the performance of subpixel classification methods is highly influenced by topographic effects. In order to select an appropriate method of topographic correction for both subpixel classification and Himalaya region, different topographic correction methods have been implemented on AWiFS satellite dataset and performance of each topographically corrected image is analyzed over linear spectral mixing (LSM) as subpixel classification method. The experimental outcomes have been computed using visual analysis, graphical analysis, statistical analysis and accuracy assessment procedures. The study provides the necessary information regarding accurate analysis of change detection, glacier mass-balances, and monitoring.