Vegetation indices for Landsat images

Satellite-based remote sensing of vegetation densities is a growing technology with great future potential owing to the proliferation of increasingly sophisticated space-based sensors and the computational tools required to analyze the wealth of data. We explore algorithms that create vegetation mappings from the satellite image data. Specifically, we study and compare the Normalized Difference Vegetation Index (NDVI), and its derivatives, the Atmospherically Resistant Vegetation Index (ARVI) and the Soil Adjusted Vegetation Index (SAVI). We use Landsat TM images taken over the southern tip of Texas. For the set of images used in the study the NDVI worked well in identifying vegetation densities. The ARVI did not work for this locale and set of TM images. The SAVI worked slightly better than the NDVI in that the mappings that it produced were more distinct (i.e., better contrast). However, SAVI suffered from slightly more noise than did NDVI.