Spatio-Temporal Urban Land Use Change in Mumbai, India: Analysis and Prediction of 2030 Using Satellite Data and a Cellular Automata-Markov Chain Model

Abstract

Mumbai, India’s economic hub, continues to grow due to its strategic location, robust infrastructure, and thriving economy. This rapid development, driven by nationwide migration, is transforming resource rich land into urban spaces. Understanding past land transformations and predicting future urban growth is crucial for sustainable planning. This study analyses Land Use and Land Cover (LULC) changes in Mumbai using multi-temporal Landsat data (1990, 2000, 2010, and 2020) and forecasts changes for 2030 and 2040. Supervised classification techniques Support Vector Machine (SVM), Random Forest (RF), and Maximum Likelihood Classifier (MLC) were applied. SVM outperformed by achieving 95.63% overall accuracy and a Kappa coefficient of 0.92, and thus used for future predictions. A Python-based Cellular Automata–Markov Chain (CA-MC) model was developed using multiple inputs to simulate future land transformations. The model was validated by predicting 2020 LULC and comparing it to actual classified data, achieving 95.24% accuracy. Results show a steady rise in urban land from 2000 to 2020, accompanied by notable declines in water bodies, water vegetation, dense vegetation, and barren land. Urban areas currently cover 45.83% of the 654.20 km² study area, projected to increase to 47.25% by 2030 and 55.84% by 2040. Between 2000 and 2020, water bodies declined by 16.80%, dense vegetation by 11.26%, and barren land by 37.29%. Urban expansion originated in South Mumbai and continues northward, converting natural landscapes into built-up areas. These insights support data-driven urban and environmental planning.