Explainable artificial intelligence in geospatial disaster modeling: A systematic review of methods, applications, and challenges

Recent advances in Artificial Intelligence (AI) have significantly improved geospatial disaster modeling applications, particularly in flood, landslide, wildfire, and earthquake susceptibility assessments. Machine learning (ML) and deep learning (DL) models integrated with geographic information systems (GIS) and remote sensing technologies provide high predictive performance; however, many of these models operate as “black-box” systems with limited transparency and interpretability. This limitation has increased the importance of Explainable Artificial Intelligence (XAI) approaches in disaster risk analysis and spatial decision-support systems. This study presents a PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses)-based systematic review of 25 studies published between 2021 and 15 June 2026 investigating XAI applications in geospatial disaster modeling. Using a PRISMA-based methodology, studies indexed in Web of Science and Scopus databases were evaluated to identify major research trends, dominant AI models, commonly used explainability techniques, and emerging research directions. The review indicates that SHapley Additive exPlanations (SHAP), Local Interpretable Model-agnostic Explanations (LIME), feature importance analysis, and attention-based methods are the most widely used XAI approaches in disaster susceptibility mapping. The findings show that landslide (40%) and flood (32%) studies dominate the reviewed literature, whereas wildfire (12%) and earthquake (8%) applications remain comparatively underrepresented; the remaining studies (8%) address hydro-morphological susceptibility and flood-related urban exposure. Despite the growing adoption of XAI methods, challenges related to spatial autocorrelation, uncertainty, transferability, computational complexity, and model reliability continue to limit the development of trustworthy geospatial AI systems. Overall, this review highlights the increasing role of explainable and human-centered AI frameworks in disaster management and emphasizes the future potential of GeoAI, digital twins, physics-informed AI, and trustworthy AI approaches for transparent and reliable spatial decision-support systems.