Optimizing Residential Construction Site Selection in Mountainous Regions Using Geospatial Data and eXplainable AI.

The rapid urbanization of Abha and its surrounding cities in Saudi Arabia's mountainous regions poses challenges for sustainable and secure development. This study aimed to identify suitable sites for eco-friendly and safe building complexes amidst complex geophysical, geoecological, and socio-economic factors, integrating natural hazards assessment and risk management. Employing the Fuzzy Analytic Hierarchy Process (Fuzzy-AHP), the study constructed a suitability model incorporating sixteen parameters. Additionally, a Deep Neural Network (DNN) based on eXplainable Artificial Intelligence (XAI) conducted sensitivity analyses to assess the parameters' influence on optimal location decision making. The results reveal slope as the most crucial parameter (22.90%), followed by altitude and land use/land cover (13.24%), emphasizing topography and environmental considerations. Drainage density (11.36%) and rainfall patterns (9.15%) are also significant for flood defense and water management. Only 12.21% of the study area is deemed "highly suitable", with "no-build zones" designated for safety and environmental protection. DNN-based XAI demonstrates the positive impact of variables like the NDVI and municipal solid waste generation on site selection, informing waste management and ecological preservation strategies. This integrated methodology provides actionable insights for sustainable and safe residential development in Abha, aiding informed decision making and balancing urban expansion with environmental conservation and hazard risk reduction. [ABSTRACT FROM AUTHOR]