A spatiotemporal segmentation-based framework for mining the spatial and temporal laws associated to urban fire hazards using data obtained through urban grid management is proposed to address the urgent need for preventive measures in urban fire management and prevention. The framework processes spatial and temporal distribution patterns successively. Firstly, spatial distribution patterns of urban fire hazards under different spatial grid resolutions are explored to obtain collections of samples with similar spatially distributed patterns. Subsequently, coexistence patterns and lagging effects of the above-mentioned set of hazards are analyzed to delve into their possible spatio-temporal distribution patterns. Empirical research is conducted with nine typical fire hazard types in Guangming district, Shenzhen, as a case study. The research shows that there are four distinct groups of similarity combinations regarding the spatiotemporal distribution of fire hazards across various spatiotemporal scales. Further spatiotemporal similarity mining indicates the coexistence relationship among hazards, with some hazards exhibiting lagging impact relationships. It can be concluded that the proposed framework is feasible for analyzing and mining spatiotemporal distribution similarity patterns of urban fire hazards. The resulting analytical conclusions of urban fire hazards coexistence, spatiotemporal similarity, and lag effects may provide substantial support for urban fire prevention tasks as well as for other similar studies.

urban fire hazards, spatiotemporal pattern analysis, hazard coexistence and lagging, urban grid management, hazard prevention