COSI-SAFE: A GIS-Based Multi-Criteria Framework for Evaluating Urban Open Space Suitability for Post-Earthquake Emergency Sheltering

Emergency response preparedness is a critical component of comprehensive earthquake risk management. The selection of suitable sites for post-disaster emergency sheltering during the response phase is central to effective contingency planning in densely populated urban areas. However, conventional planning often relies on ad-hoc methods that fail to systematically balance competing suitability factors. We propose COSI-SAFE, a Comprehensive Open Space Suitability Index (COSI)-based Suitability Assessment for Emergency-sheltering (SAFE) framework that evaluates the suitability of open spaces for temporary shelter establishment in the immediate aftermath of an earthquake in a densely populated urban area (Mirpur area in Dhaka City, Bangladesh), utilizing a GIS-based multi-criteria decision-making (MCDM) approach. The evaluation framework defines suitability based on three key dimensions — quality, capacity, and accessibility. Quality indicators were derived from field surveys and secondary data sources. Capacity and accessibility assessments were conducted using spatial analysis tools, including network and proximity analysis in the ArcGIS platform. The Analytic Hierarchy Process (AHP) was applied to combine the three criteria and derive the COSI for the identified open spaces. The results indicate that most open spaces are either moderately or poorly suited for use as emergency shelters, with only a small fraction deemed highly favorable in a post-earthquake context. Specifically, 1.2% of the open spaces were found to be completely unsuitable, 23.5% had low suitability, 56.8% were moderately suitable, and only 18.5% were fairly suitable for temporary shelter establishment. When weighted by total open-space area rather than site counts, the distribution was: unsuitable 0.8%, low 18.0%, moderate 59.5%, and fair 21.7%, indicating that a limited number of large open spaces contribute disproportionately to the total available shelter area. The findings highlight the inadequacy of existing open spaces to accommodate potential sheltering demand, emphasizing the need for proactive planning in the studied region. Assuming a deterministic Mw 7.5 event and displacement rates of 80% for partially damaged and 100% for extensively damaged buildings, we estimate a shelter-seeking population of 576,678, of whom 242,381 (42.0%) could be accommodated within a 700 m walking radius at 1.7 m²/person. The developed methodology offers a replicable planning tool for policymakers, emergency managers, and responders to strengthen earthquake response preparedness and inform post-disaster response strategies in high-density, earthquake-prone urban areas.