Objectives On December 18, 2023, an Ms 6.2 earthquake occured in Jishishan County, Gansu Province, China. The purpose of this study was to construct a spatiotemporal evolution model of population casualties in Jishishan by integrating multi-source data, and to estimate the casualties in Jishishan area in the complex environment of Jishishan by combining the spatiotemporal distribution characteristics of Jishi-shan.

Methods This study developed a spatiotemporal model for predicting population casualties in earthquake disasters using multi-source data. By integrating the coseismic deformation field obtained from Sentinel-1 satellite's ascending and descending radar images with time series records from various near-field stations (N0028, LXJS, Yinchuan, Dawu, and Guyuan) as well as accurate positioning results of aftershocks, we were able to analyze the spatiotemporal distribution characteristics of the Jishishan area and predict population casualties resulting from the earthquake.

Results The results show that the earthquake has the characteristics of thrusting. The deformation of the ascending and descending coseismic deformation fields along the line of sight is mainly uplift, which is 7.51 cm and 7.91 cm respectively. The precise location of aftershocks shows that the earthquake may have occurred on an NW or SE striking fault. At the same time, after analyzing the precise location of the aftershock and the time series signal of the near-field station, the acceleration of the seismic station (N0028) indicates that the earthquake has caused sustained damage. The time series of global navigation satellite system station (LXJS) indicates that the permanent displacement of 1.986 cm and 1.377 cm in the north and east directions is produced by the earthquake. The Yinchuan, Dawu and Guyuan stations have found that the fault activity in the Jishishan region caused signi-ficant coseismic gravity variation. To determine the casualties of the earthquake disaster and explore the characteristics of the “coseismic severe disaster”. The spatiotemporal evolution model of earthquake casualty is constructed by multi-source data, and the parameters of the Jishishan earthquake case are used as input factors of the model. The predicted result is 149 people died in the main earthquake, and the accuracy is 98.45%. Through the aftershock time series prediction, 3 people died in the aftershock, 152 people died in total, which is 1 person more than the actual 151 people, which shows that the model can predict the population casualties very well.

Conclusions It can be found that the study on the spatiotemporal evolution of the population casualties in the Jishishan earthquake disaster area under the complex mountain environment can not only provide scientific guidance and support for post-disaster rescue and reconstruction, but also provide valuable reference information for the prevention, monitoring, and mitigation of similar disasters in the future.