Explainable Artificial Intelligence (XAI) Model for Earthquake Spatial Probability Assessment in Arabian Peninsula

Among all the natural hazards, earthquake prediction is an arduous task. Although many studies have been published on hazard assessment (EHA), very few use of artificial intelligence (AI) in spatial probability (SPA). There a great deal complexity observed SPA modeling process due to involvement seismological geophysical factors. Recent shown that insertion certain integrated factors such as ground shaking, seismic gap, and tectonic contacts AI model improves accuracy extent. Because black-box nature models, this paper explores explainable (XAI) SPA. This study aims develop hybrid Inception v3-ensemble extreme gradient boosting (XGBoost) shapely additive explanations (SHAP). The would efficiently interpret recognize factors’ behavior their weighted contribution. work explains specific responsible for importance inventory data were collected from US Geological Survey (USGS) past 22 years ranging magnitudes 5 Mw above. Landsat-8 satellite imagery digital elevation (DEM) also incorporated analysis. Results revealed SHAP outputs align with v3-XGBoost (87.9% accuracy) explanations, thus indicating necessity add new gaps contacts, where absence these makes performs poorly. According interpretations, peak accelerations (PGA), magnitude variation, epicenter density are most critical recent Turkey earthquakes (Mw 7.8, 7.5, 6.7) active east Anatolian fault validate obtained AI-based results. conclusions drawn algorithm depicted relevant, irrelevant, futuristic modeling.