Geological hazards caused by strong earthquakes have caused continuous social and economic losses and destruction of the ecological environment in the hazard area, and are mostly manifested in the areas with frequent occurrence of geological hazards or the clustering of geological hazards. Considering the long-term nature of earthquakes and geological disasters in this region, this paper takes ten earthquake-stricken areas in Wenchuan earthquake zone as examples to collect shallow landslide data in 2010, combined with the spatial location of landslides and other factors. Kernel density estimation (KDE) method is used to analyze the spatial characteristics of shallow landslide. Taking the space of shallow landslide as the characteristic variable and fully considering the regulating factors of earthquake-induced landslide: terrain complexity, distance to river, distance to fault, distance to road, lithology, normalized vegetation difference index (NDVI) and ground peak acceleration (PGA) as independent variables, based on KDE and polynomial logistic regression (MLR), A quantitative model of shallow landslide in the earthquake area is constructed. The results show that: (1) PGA has the greatest impact on landslide in the study area. (2) Compared with the two-category logistic regression (two-category LR) model, the susceptibility map of landslide prediction results based on the KDE-MLR landslide susceptibility prediction model is more consistent with the actual situation. (3) The prediction accuracy of the model validation set is 70.7%, indicating that the landslide susceptibility prediction model based on KDE-MLR can effectively highlight the spatial characteristics of shallow landslides in 10 extreme disaster areas. The research results can provide decision-making basis for shallow landslide warning and post-disaster reconstruction in earthquake-stricken areas.

In the present study, a spatial quantitative model of landslide hazards based on a deep belief network (DBN) is constructed. Firstly, environmental similarity-based sampling (ESBS) was used to determine the negative sampling area. Secondly, multiple data sets are constructed. Each data set contains seven landslide-conditioning factors; 70% of the data are used for training; and 30% are used for validation. The performance evaluation index of the spatial quantitative model of landslide hazards was established; that is, the AUC mean (AUCmean) was used to measure the stability of the model, and the AUC standard deviation (AUCSD) was used to measure the uncertainty of the model. Finally, the accuracy of the prediction results of the DBN model is analyzed. The results show that the area with negative sample reliability greater than 0.51 is the best negative sample sampling area, and the stability of the DBN model is maintained at a relatively good level in both the training step (AUCmean = 0.9597) and the validation step (AUCmean = 0.8897). The standard deviation of AUC is close to 0 (AUCSD = 0.0081 in the training step and AUCSD = 0.0085 in the validation step), indicating that the selected negative samples have a weak impact on the performance of the model. The susceptibility areas of very high obtained by the DBN model (landslide points in the susceptibility areas of very high accounted for 55.03%) are realistic. Therefore, the DBN model constructed in the present study is effective and can be used in the field of landslide hazard spatial prediction.