Work

A deep learning-geospatial analytics fusion framework for predicting land use land cover dynamics in eastern region of India

Year: 2026

Type: article

Abstract: Globally, rapid urbanisation and unregulated urban expansion have driven widespread Land Use and Land Cover (LULC) changes, resulting in deforestation, loss of vegetation, and depletion of water bodies. In India, cities and peri-urban areas are experiencing similar transformations, making it essential to monitor and manage urban growth effectively. This study aims to analyse LULC changes in Berhampore, West Bengal, from 2001 to 2023, identify key drivers of urban expansion, and predict future land use for 2033 using remote sensing, GIS-based spatial analysis, and Deep learning techniques (ANN and CA models). The results show a sharp increase in built-up areas from 20.39 sq. km (2001) to 43.85 sq. km (2023), leading to significant reductions in vegetation (-52.13 sq. km) and water bodies (-1.74 sq. km). Forecast for 2033 shows continued urban development where agricultural territories represent sixty-two percent of the land while forests face a fifty percent reduction thus boosting risks of biodiversity depletion along with enhanced flood potential. This research proves that land sustainability policies and green infrastructure planning combined with strategic urban strategies must urgently develop to balance development with environmental endurance. This research shows that ANN and CA operate as predictive methods to enhance LULC accuracy by helping administrators execute measurement policies. This study incorporates deep learning architecture parts that integrate CNNs and LSTMs to establish advanced LULC modelling frameworks which excel at spatial-temporal pattern detection. Future research on urban growth prediction will benefit from the proposed structures built to improve predictive models by integrating hierarchical features based on long-term dependency according to academic publications.

Source: Results in Engineering

DOI: 10.1016/j.rineng.2026.109201