Satellite Image Analytics, Land Change and Food Security

Changing patterns and reduction in agricultural land are among the fundamental problems that impacts food security in developing regions like India. Rapid economic growth coupled with increasing populations and changes in climatic patterns are among the main factors impacting availability of agricultural land. On a macroscopic scale, due to the lack of good quality data, governments do not have a complete and clear cut picture of changes in land usage patterns. In this paper, we present the design of a satellite image analytics engine that we use to perform a detailed analysis of changes in agricultural land patterns over a 13-year time period (2000-2012) in West Bengal, India, traditionally considered one of the most fertile areas in the world. Our satellite analytics engine can perform a fine-grained analysis of macro-granular satellite images (elevation of 11 km) and classify small portions of land in each image into different categories: agricultural, developed, forest and water bodies. Our analytics engine can analyze temporal changes in land patterns and compute the percentage of change in land under each category. Based on detailed food production data gathered in collaboration with the bureau of statistics of West Bengal, we analyze the correlations between changes in agricultural land patterns and corresponding changes in food production (normalized by change in yield patterns). Our tool can be used at varying levels of spatial granularities ranging from macroscopic analysis at a state level to fine-grained analysis at sub-district levels. This analytics tool is targeted for government and non-governmental policy makers to analyze land pattern changes and correlate them with food security metrics.