Comparison of Techniques for Forest Change Mapping Using Landsat Data in Karnataka, India

Abstract

The potential for forest change monitoring in the state of Karnataka, India using Landsat imagery was evaluated. Imagery from 1986 and 2003 was analyzed using two change detection techniques: (1) image differencing of the Normalized Difference Vegetation Index (NDVI), the second principal component (PC2), and the Kauth‐Thomas greenness index (KT‐G), and (2) post‐classification comparison (PCC). As field validation data did not exist for 1986, extensive visual assessment was conducted to locate and identify errors of commission and omission in the change maps. The image difference vegetation maps did not display obvious errors of omission, but the NDVI difference performed better than KT‐G and PC2 differences in terms of errors of commission. It was therefore classified into a deforestation/reforestation map and evaluated against the PCC forest change map. PCC was able to more accurately detect changes over the 17‐year period. Analysis of the literature and the forest change maps showed that deforestation was primarily a result of submergence by reservoirs created in hydroelectric developments, whereas reforestation was mainly due to significant increases in forest plantations, as a result of various social forestry projects.