Predicting Urban Growth of Kathmandu Valley Using Artificial Intelligence

The rapid global expansion of urban areas in the twenty-first century presents a significant challenge, evident in the Kathmandu Valley in Nepal, where urbanization reshapes the Himalayan foothills. Over the past four decades, the Kathmandu Valley has witnessed substantial changes in land use and land cover (LULC). Employing pixel-based unsupervised classification with Landsat and Sentinel imagery from 1992, 2003, 2013, and 2021, the research analyzed LULC alterations and expansion of the built-up area. The most significant growth occurred between 2003 and 2013, restructuring the cityscape concentrically. Proximity of the central geographical features and rural–urban migration steered this expansion. The study uses QGIS with the MOLUSCE plugin and MLP-ANN model to forecast urban growth, achieving 93.82% accuracy in urban growth prediction. The research optimized spatial patterns for future urban growth under three scenarios: baseline, environment-friendly, and resource-efficient sustainable development. Anticipated spatial patterns provide insights into urban growth trajectories by 2041, which is crucial for informed decision-making on land availability. The study highlights the utility of LULC and CA-ANN models in identifying future trends and aiding governments, planners, and stakeholders in estimating the potential repercussions of policy options.