Remote sensing technology provides the opportunity to greatly enhance socio-economic and environmental research, however it remains vastly under-used by researchers in these fields due to obstacles in expertise and experience with the technology. Here a case study is used to elucidate the integration of this technology into such applications. In order to evaluate the environmental impact of the Green Revolution in a semi-arid region of Western India, vegetation change data was derived from satellite imagery during the years 1972-1980. Because methodologies to determine vegetation cover from remotely sensed data range in complexity, sophistication, and accuracy, two commonly used methods were compared in order to determine the level of sophistication necessary to derive reliable and useful vegetation data for social and environmental applications. The Normalized Difference Vegetation Index (NDVI) was compared with Spectral Mixture Analysis (SMA). Both methods were applied to a total of six images, two scenes for three separate dates. The results show that NDVI and SMA both yield similar results for the majority of vegetation types in the scene, however, some discrepancies exist in areas of high vegetation abundance, especially complex forest structures. The vegetation change results highly correlated between the two methods, although forest change analysis using NDVI yielded extremely different results than a methodology using multiple SMA endmembers. The data generated from this analysis was used to determine the major land cover trends and environmental changes that took place over the study region during this time period. Dams were responsible for the majority of large-scale vegetation loss in the region. The major increases in vegetation cover occurred in large-scale agricultural plots, mostly likely due to increased irrigation.

Simultaneously vegetation cover increases in villages were disproportionately low, indicating sharp inequalities in the benefits of irrigation projects. Additionally desertification effects were analyzed in small areas using SMA multiple endmember datasets.

View the complete thesis at:
http://envstudies.brown.edu/thesis/2000/undergrad/lfirestone/title.htm