From Hazard Maps to Decision Intelligence: A Critical Review of GIS, Remote Sensing and GeoAI for Disaster Risk Management and Engineering Resilience

Disaster risk is inherently spatial, yet geographic information systems (GIS) are still frequently used to end analysis at a classified hazard map rather than to support an engineering or operational decision. This critical review traces how GIS, integrated with remote sensing (RS), multi-criteria decision analysis (MCDA), Web-GIS and geospatial artificial intelligence (GeoAI), has evolved across the disaster management cycle, and evaluates whether the resulting products are validated, uncertainty-aware and usable in civil-engineering practice. Twenty foundational studies (1992–2017) covering general emergency management, floods, earthquakes, landslides, cyclones and coastal hazards, drought and forest fire are appraised against a structured set of dimensions and are updated with recent systematic reviews of Web-GIS, AI/ML-enabled GIS, drone-based hazard sensing and AI-based landslide analysis. Four analytical families (knowledge-driven, data-driven, simulation-based and integrated Web-GIS/GeoAI) are compared. Knowledge-driven methods such as AHP and weighted overlay remain valuable where event inventories are scarce but are sensitive to expert-assigned weights; statistical and machine-learning models can improve prediction but are vulnerable to spatial leakage, non-representative inventories and limited interpretability; operational Web-GIS and GeoAI systems are advancing quickly but are rarely validated against independent events or evaluated by their effect on decisions. The most persistent gap is the translation of hazard classes into engineering quantities such as flood depth and velocity, design ground motion, slope safety factors, damage states and lifeline reliability. The review proposes (i) an eight-stage data-to-decision framework linking GIS, RS, GeoAI and engineering analysis, (ii) a minimum reporting and validation checklist for GIS-based hazard studies, and (iii) a prioritised research agenda covering multi-hazard modelling, near-real-time Web-GIS, uncertainty-aware GeoAI, digital twins, BIM–GIS integration and decision-centric validation.