| As one of the common geological disasters,landslides are closely related to people’s lives,and landslide disasters are often sudden.According to statistics,landslide disasters accounted for more than 50% of the geological disasters that occurred in China every year from 2005 to 2020,and the highest proportion reached 86%,posing a certain threat to people’s normal production and life.Jingyuan County is located in the central and eastern part of Gansu Province,and is divided into north and south parts by Pingchuan,and there are differences in geological structure,geological environment conditions and human engineering activities in the north and south,and there are few studies on the susceptibility assessment of landslide disasters,and the data reference on the distribution law of landslides is also limited.In order to explore the development law of landslides in Jingyuan County,this paper takes the northern part of Jingyuan County as the research area,combines the DEM,Landsat8,landslide data and engineering geology data in the northern part of Jingyuan County,selects a 25 m grid as the evaluation unit,completes the landslide susceptibility assessment in the study area based on the GIS platform,and obtains the landslide susceptibility zoning map,which provides basic data for the geological disaster risk assessment in the research area and provides a reference for the development of disaster prevention and mitigation work in the area.The relevant results achieved are as follows:(1)The relationship between the disaster-causing factors and landslide disasters was analyzed,and the evaluation index system of the district was constructed,and ten low-degree related disaster factors were screened out,including the landform type,distance from ravine,distance from highway,lithology,SPI,elevation,distance from river,land type,NDVI and slope,which not only improved the accuracy of the susceptibility assessment results,but also provided a reference for the selection of influencing factors in the future similar susceptibility evaluation process.(2)The best computing cell for evaluation was obtained,and the evaluation results obtained by removing the leading edge of the landslide with the surface element were the most accurate.If the point unit is used as the computing cell,the point in the slope can best represent the developmental characteristics of the landslide,and the resulting zoning map of susceptibility results is more reasonable.(3)For logistic regression and perceptron models,the accuracy of the susceptibility assessment results in the study area can be further improved by coupling the model with the bivariate statistical model,and when the support vector machine is used for evaluation in this research area,the single support vector machine model obtains the best effect,and model coupling is not suitable.(4)The best results of evaluation results are selected in various models,and the results are fused by principal component analysis,which will not change the accuracy of the results obtained by each model,and can obtain the susceptibility zoning map in line with the actual situation,and can also make up for the shortcomings of bivariate statistical models and machine learning models,that is,improve the problem of poor recognition ability of bivariate statistical models and support vector machine models in flat land,and also optimize the polarization of logistic regression and perceptron model evaluation results.(5)According to the final susceptibility zoning map,the characteristics of landslide susceptibility zoning distribution in the study area were obtained.The high susceptibility area was mainly distributed in the area from Shimen Township to Yongxin Township in the west of the study area,and also distributed in the south of Dongsheng Town to Jing’an Township,and the lithologic types in the area were mainly structural erosion of medium and low mountains and erosion and erosion of loess hills,and the engineering rock group type was mostly loess single-layer soil,and the slope of the mountain was between 25-63°.The medium and low susceptibility areas are mainly distributed outside the high susceptibility areas;Non susceptibility areas are mainly distributed in the flat east of Beitan Town,and the northern part from Dongsheng Town to Jing’an Township.In addition,the distribution characteristics of the high susceptibility area were similar to the distribution characteristics of landslide disasters in the study area,mainly along both sides of ravines in the study area. |
