Land Use and Land Cover Dynamics in Dinajpur District: Spatio-Temporal Trends and Future Scenario Using Google Earth Engine and CA–Markov Modeling

Abstract

Investigation on the detection and prediction of changes in land use and land cover (LULC) is essential for reinforcing sustainable management and planning. As a result, the purpose of this research was to discover LULC changes in the Dinajpur area of Bangladesh between 2000 and 2020, as well as anticipate future changes for 2040. Landsat-5, 7, 8, and 9 pictures, as well as vegetation indices and topographic parameters, were used to classify LULCs in 2000, 2005, 2010, and 2020. Classification was performed using the random forest (RF) machine learning method, which is embedded into the cloudbased platform Google Earth Engine (GEE). The classification accuracy evaluation found excellent agreement between the classified maps and the validation dataset, with kappa coefficients of 0.904, 0.919, 0.924, and 0.901 for the LULC maps of 2000, 2005, 2010, and 2020, respectively. The Spatio-Temporal Trends study shows that from 2000 to 2020, built-up land grew by 82.8%, barren land grew by 41.1%, and vegetation and agriculture shrank by 74.1% and 15.1%, respectively. The LULC state in 2040 was simulated using Cellular Automata-Markov. CA-Markov forecasts indicate that agricultural land would decline by 18%, while built-up and barren areas will rise by 27% and 22.6%, respectively. Furthermore, declining forest cover raised concerns about long-term ecological balance. The recent developments in Dinajpur highlight the need for a unified land strategy that balances urban growth and environmental protection.