Quantifying land use and land cover (LULC) change is necessary for tracking the impacts of urbanisation on the natural environment and for planning sustainable land use. This study analyses LULC transformations in Nashik city, Maharashtra, India, using multi-temporal Landsat satellite data acquired from 1996 to 2025. The Random Forest (RF) algorithm was applied for image classification, yielding overall accuracy values ranges from 88–91% and Kappa coefficients in the range 0.84–0.88. Change analysis shows that the built-up area expanded from 42.52 km² (15.97%) in 1996 to 116.63 km² (43.80%) in 2025, while vegetation cover contracted from 105.14 km² (39.50%) to 50.40 km² (18.93%). An Artificial Neural Network–Cellular Automata (ANN–CA) model was subsequently used to simulate LULC for 2030, with OpenStreetMap (OSM)-derived spatial drivers — proximity to roads, railways, water bodies and existing built-up land — incorporated alongside terrain variables. The 2030 simulation indicates a further increase in built-up area to 126.49 km² (47.57%) and a continued decline in vegetation to 43.84 km² (16.49%). Region-wise density analysis was also performed for all administrative zones of Nashik, revealing a centre-to-periphery gradient in which central regions carry high built-up density and outer regions retain comparatively more vegetation cover. The results demonstrate that combining infrastructure-based spatial drivers, long-term RF classification and region-level density mapping within a single workflow provides practical information for urban planning and land management.