Predicting the spatiotemporal changes of an agriculturally vulnerable region of Bangladesh

Abstract

Agricultural land, the primary factor of food production, is essential for ensuring food security. Land constraints have led policymakers to promote agricultural intensification to achieve higher production, which is no longer sustainable. In Bangladesh, the consistent decline of agricultural land at a regional scale is a rising concern for food security. This study intends to assess the spatiotemporal changes in agricultural lands concerning food security, including temporary cropland, permanent cropland, and fallow land. LANDSAT satellite imagery for 1995, 2010, and 2022 were categorized using a hybrid image classification method. However, the study limits to produce higher accuracy as compromised due to the spatial resolution of LANDSAT imagery. MLP-CA Markov chain model was used to predict the agricultural land for 2041 by employing driver variables. The study finds around 15% loss in agricultural land from 1995–2022 with significant losses (12%) between 2010–2022. The built-up area is doubled after each of the time periods. Temporary crop-producing lands are declining quickly and converted rapidly (around 30%) to built-up areas between 2010–2022. Notably, agricultural land near riverine zones rapidly converts to built-up areas, hinting at potential environmental consequences. The model predicts around 10% loss in agricultural land with a likely conversion around cities and riverine areas, driven by infrastructure development. Contradictory sectoral policies have driven such conversion without effective land use policy. Hence, the study implies formulating a physical plan and urbanization policy for growth control and management, as well as land zoning and master plan for protecting valuable agricultural land.