Landslide susceptibility assessment in Wenchuan County after the 5.12 magnitude earthquake

Wenchuan was designated as one of the earthquake-stricken areas after the 2008 5.12-magnitude Wenchuan earthquake. During the decade following the earthquake, many post-earthquake landslides have occurred, seriously threatening the property of the local population’s lives and property. At present, there are few long-term spatiotemporal changes studies on landslides generated after the earthquake. The main aim of the present study is to use Landsat multi-period remote sensing image data, combined with post-earthquake landslide interpretation point data, to extract landslide point information from 2011, 2014, and 2018 to analyze the spatial evolution of landslides and define reliable susceptibility models for landslides in 2018 using an assessed certainty factor (CF), an integrated CF with random forest (RF), and an integrated CF with logistic regression (LR). The area under cure (AUC) was used to test the accuracy of the three models and obtain the most reflective susceptibility map. The results show that the RF-CF model produced the optimal result in terms of its AUC value (0.84). Over the past decade, the high-density area has become more concentrated than that of other grades and its coverage small, and its proportion has gradually decreased; the proportion of landslides in low-density areas has been increasing, and its distribution has become more concentrated. These research results fill the gap in the study of temporal and spatial changes in landslides after an earthquake and provide a scientific basis for disaster management and prediction in landslide-prone areas such as Wenchuan.