Multi-Temporal LULC Change and Future Urban Expansion in Faridpur Municipality, Bangladesh: A CA-ANN Based Simulation

The study examines land use and land cover (LULC) conversions and forecasts future urban growth in Faridpur Municipality, Bangladesh. Despite substantial urbanization in recent years, smaller cities such as Faridpur have been largely absent from LULC studies. Using Landsat data from 2004, 2014, and 2024, the research employed Maximum Likelihood Supervised Classification (MLSC) to identify land cover changes. It used an Artificial Neural Network (ANN) with Cellular Automata (CA) to anticipate LULC trends for 2034, 2044, and 2054. Findings indicate that urbanized areas grew from 2.18% in 2004 to 7.31% in 2024, initially replacing vegetation and cropland, and are projected to expand further to 15.04% by 2054. An R² of 0.9164 represents a strong statistical fit of the plotted built-up-area trend; however, for greenspace, farmland, and water, such expansion into the projected future is not promising. Geospatial analysis suggests the most developed areas are in the northeast and southwest regions, which offer greater connectivity and are the two growth magnets. Validation of the model consistently showed an accuracy of 92.56%, Kappa of 0.878, Kappa (histogram) of 0.9829, and Kappa (location) of 0.8940, which indicates that the model reproduces the observed spatial pattern reasonably well. Thus, the findings suggest that urban planners need to pay attention to trends in sustainable expansion and in the creation of agricultural space. By conveying practical understandings into growth patterns, it supports policymakers, local authorities, and urban planners in monitoring and predicting future urbanization in Faridpur Municipality.