Based on the improved SCGM(1,1)c and WIV rainfall landslide susceptible area prediction model
| Autor: | Qian Zhang, Shujie Cao, Yanliang Du, MingYuan Du, Yixuan Zhao, Yaoqi Nie |
|---|---|
| Jazyk: | angličtina |
| Rok vydání: | 2024 |
| Předmět: | |
| Zdroj: | Geomatics, Natural Hazards & Risk, Vol 15, Iss 1 (2024) |
| Druh dokumentu: | article |
| ISSN: | 19475705 1947-5713 1947-5705 |
| DOI: | 10.1080/19475705.2024.2399667 |
| Popis: | Rainfall is an important factor inducing landslide deformation. Rainfall induced landslide warning is of great significance for regional landslide prevention and control. Based on ArcGIS software, the rainfall data of Guodashan area in southwestern China from 1991 to 2021 were analyzed and processed. On the basis of the single factor system cloud grey model (SCGM (1,1)c), an improved SCGM (1,1)c model is proposed based on Markov prediction theory and CS algorithm optimization to predict rainfall. According to the prediction results, a total of 10 evaluation factors are considered, including altitude, slope, aspect, profile, engineering rock group, geological age, river, road, fault, etc. A GIS based weighted information model was established to divide the study area into five categories: extremely low, low, medium, high, and high, and to evaluate the susceptibility of landslides. The results showed that the residual corrected MaxAPE and MAEP values of the improved SCGM (1,1)c rainfall prediction model were 8.8 and 2.57%, respectively. After susceptibility zoning, historical landslide frequencies indicate that 92.4% of landslide points are located above the high susceptibility zone, and the AUC value of the model obtained from the ROC curve is 0.8654, indicating that the weighted information model has good accuracy in predicting susceptibility zoning. This coupled model can accurately predict the spatiotemporal distribution of regional rainfall and landslide susceptible areas, and predict the distribution of landslide deformation and geological disaster susceptible areas that are close to the actual landslide distribution areas, providing theoretical support for regional landslide warning and prediction. |
| Databáze: | Directory of Open Access Journals |
| Externí odkaz: |
|