Attempts to study shifting cultivation landscapes are fundamentally impeded by the difficulty in mapping and distinguishing shifting cultivation, settled farms and forests. There are foundational challenges in defining shifting cultivation and its constituent land-covers and land-uses, conceptualizing a suitable mapping framework, and identifying consequent methodological specifications. Our objective is to present a rigorous methodological framework and mapping protocol, couple it with extensive fieldwork and use them to undertake a two-season Landsat image analysis to map the forest-agriculture frontier of West Garo Hills district, Meghalaya, in Northeast India. We achieve an overall accuracy of ~80% and find that shifting cultivation is the most extensive land-use, followed by tree plantations and old-growth forest confined to only a few locations. We have also found that commercial plantation extent is positively correlated with shortened fallow periods and high land-use intensities. Our findings are in sharp contrast to various official reports and studies, including from the Forest Survey of India, the Wastelands Atlas of India and state government statistics that show the landscape as primarily forested with only small fractions under shifting cultivation, a consequence of the lack of clear definitions and poor understanding of what constitutes shifting cultivation and forest. Our results call for an attentive revision of India’s official land-use mapping protocols, and have wider significance for remote sensing-based mapping in other shifting cultivation landscapes.
Modeling Agricultural Change through Logistic Regression and Cellular Automata: A Case Study on Shifting Cultivation
