Topographic Effects On The Determination Of Hyperspectral Vegetation Indices: A Case Study In Southeastern Brazil

We investigated the sensitivity of 16 hyperspectral vegetation indices (VIs) to topographic effects in southeastern Brazil. Using Hyperion data acquired over ombrophilous forest and the topographic C-correction, we calculated the relative changes and statistical differences (Wilcoxon test) between corrected and non-corrected data, and the magnitude of the effect size of topographic correction. The analysis considered 12 terrain illumination classes obtained from cross-comparison between aspect and slope images. The results showed that the most anisotropic indices were the Anthocyanin Reflectance Index (ARI), Enhanced Vegetation Index (EVI) and Photochemical Reflectance Index (PRI). During the comparison between C-corrected and non-corrected data, these VIs presented the largest relative changes, the lowest p-values and the largest metric of effect size. The sensitivity was stronger with increasing slope over south- and west-facing shaded surfaces. The least sensitive VIs were the Visible Index Green (VIG), Visible Atmospherically Resistant Index (VARI) and Structure Insensitive Pigment Index (SIPI).