Abstract: The fire risk assessment at the city level mainly includes fire hazard, fire harmfulness, and fire emergency response capability. This article selects fire hazard assessment for targeted research. In the era of big data, causal relationships could be replaced with correlation. Thus, multi-source data is used for weighing and scoring the evaluation indicators in order to obtain a spatiotemporal distribution of fire danger in the urban area of Fuzhou. First, address analysis of the fire alarm records in the past few years is performed using the AMap API tool, and nearly 10 000 fire alarm locations are spatially assigned mapping the spatial distribution of historical fires in Fuzhou. Furthermore, multi-source and heterogeneous data, such as the structure of urban land use, property of land use development, and thermal map of population distribution, is further integrated to explore its correlation with the spatial distribution of historical fires. Finally, taking Fuzhou as an example, the results of dynamic fire risk assessment are formed, and the fire risk assessment method with sufficient mathematical support is developed, which provides strong support for urban fire control planning.

Key words: big data, multi-source spatiotemporal data, multiple linear regression, fire hazard assessment, Fuzhou