A decision-making framework for landfill site selection in Saudi Arabia using explainable artificial intelligence and multi-criteria analysis

Abstract

The rapid urbanisation in Abha, Saudi Arabia, especially in the mountainous landscape, makes it necessary to identify the optimal locations for environmentally friendly building complexes. This study introduces a transparent decision making framework for landfill site selection that combines multi-criteria decision making, fuzzy set theory, GIS and eXplainable Artificial Intelligence (XAI). We focused on the complex interplay of geophysical, geoecological and socio-economic parameters and used fuzzy analytical hierarchy process (AHP) for parameter weighting to address the challenges of site selection in mountainous urban centre in developing country. An index map, called the Landfill Site Potential Index (LSPI), was generated, integrating all key parameters to indicate suitable zones for landfill sites. The LSPI was classified into different suitability zones, ranging from very high to very low suitability, using Self-Organising Maps (SOM) in combination with k-means clustering. The use of XAI models, in particular an optimised bagging ensemble model, provided crucial insights into the factors influencing site suitability. The LSPI with a mean value of 0.681 was categorised into five zones, ranging from very low (7.25%) to very high suitability (20.60%). Within the most suitable zone, ten sub-zones were defined for the prioritised development of landfill sites. The model showed high accuracy, with the SHAP and LIME analyses providing a deeper understanding of the global and local determinants of landfill site suitability. This study not only provides a comprehensive and transparent decision-making framework for waste management in developing countries, but also improves the understanding of the critical factors in landfill site selection.