Spatial Modeling of Urban Fire Risk in Nagpur City Using Kernel Density Estimation (KDE) and GIS

Abstract

Urban fires are one of the most common and devastating hazards in Indian cities because of high-density development with inadequate infrastructure. This study applies Kernel Density Estimation (KDE) integrated with GIS for identifying and analyzing the spatial fire risk patterns across Nagpur City in Maharashtra over three years, including incident data on fire from 2020 to 2022. A total of 487 geocoded fire events have been processed in order to create a continuous surface density that represents the intensity of fire incidents. The KDE model was run within ArcGIS 10.8 using a 500 m bandwidth for the city, classifying it into four classes of risk: Very High, High, Moderate, and Low. Strong clustering was observed around the Gandhibagh-Satranjipura-Lakadganj zones, which contributed almost 46% to the incidents for only 15% of the urban area. Further validation with the KDE surface through the correlation and ROC analysis resulted in r = 0.81 and AUC = 0.86, meaning that this surface is highly accurate. Spatial overlay with fire station coverage shows severe service gaps in many high-risk wards and peripheral residential sectors having relatively low exposure. These spatial insights show that urban morphology and land-use composition are a strong driving force for fire risk in Nagpur. This study underlines the utility of KDE as a low-cost, reproducible tool for mapping urban fire risk and also supports policy recommendations on optimization of fire stations, safety zoning, and evidence-based resilience planning. The results add to the emerging area of spatial disaster analytics and provide a feasible paradigm for municipal-level fire risk mitigation in rapidly urbanizing Indian cities.