In recent years, in order to deal with the frequent occurrence of disasters, master and predict the law of disaster occurrence, explore the nature, disaster knowledge map construction and analysis and application technology has become an important research hotspot and direction. However, existing knowledge graph structures for natural disasters often focus only on temporal or spatial properties, lacking the integration of spatiotemporal properties. Therefore, this paper proposes a technique for constructing a typical disaster knowledge graph integrated with its spatial and temporal characteristics. First of all, through the analysis of typical disaster events, established the “event-space-time-causal” triplet model, the disaster three factors relationship, formed a comprehensive, systematic disaster knowledge framework, and then use deep learning technology of semantic analysis of large amounts of text data and relationship extraction, to automatically build a disaster knowledge map. In the test verification, this paper takes a natural disaster as an example to construct a typical disaster knowledge map integrating spatial and temporal characteristics, and conducts empirical analysis. The results show that the knowledge graph construction method can better express the spatial and temporal characteristics and influencing factors of some natural disasters, and provide more comprehensive and accurate disaster prevention and control and response strategies for decision-making agencies. In conclusion, the typical disaster knowledge map construction technology integrating spatial and temporal characteristics is a novel and effective disaster response method with great application potential.