<p>This study proposes a geological hazards susceptibility evaluation method based on the spatial–temporal evolution of evaluation factors. The variation trend slope (Slope) is pre-processing a model to obtain the spatial–temporal evolution of evaluation factors. Then substitute the spatial–temporal evolution data of evaluation factors into ensemble models of information content method and logistic regression (ICM-LR), and information content method and artificial neural network (CIM-ANN), predicting the susceptibility of geological hazards, providing a reference for future prediction of geological hazards. In this study, the annual precipitation and normalized difference vegetation index (NDVI) of Jiuzhai gully from 2016 to 2021 are used as basic data to build Slope model. Train and validate using historical geological hazard data after the Mw 7.0 earthquake in 2017. Evaluate the model through curve area under curve (AUC) and compare it with previous research results. The study results indicate that the geological hazards susceptibility of ICM-LR and ICM-ANN obtained using the Slope model has high applicability. It can effectively improve the accuracy of short-term geological hazard prediction and provide a reference value for post-earthquake geological hazard prediction and post-earthquake reconstruction.</p>