Modeling and Predicting Land Use/Land Cover Change Using Deep Learning in southern Malawi

Abstract Monitoring land use and land cover (LULC) change is crucial for ecological sustainability and urban planning. This study employs a Multi-Layer Perceptron (MLP) neural network and Random Forest classifier to model LULC dynamics in Malawi’s Michiru Mountain Forest Reserve from 2004 to 2024. During this period, vegetation cover declined from 85.77 km² to 31.55 km² (63% loss), while bare land increased by 123% and built-up areas nearly doubled from 4.58 km² to 8.77 km². Spatial analysis shows 70% of urban expansion occurred within 500 meters of roads, emphasizing the role of infrastructure. The optimized MLP model achieved 99.87% accuracy, a 0.9974 skill measure, and a Kappa coefficient of 0.85, with minimal overfitting (training RMS: 0.0368, testing RMS: 0.0396). Forecasts for 2034 project built-up areas to reach 23.20 km² and further vegetation decline, highlighting ecosystem degradation and the need for integrated governance.