Study On Landslide Risk Assessment In Mountainous Villages And Towns

China is one of the most seriously affected countries by landslides,especially in mountainous villages and towns.Due to the complex topographical,geological conditions and the increasing engineering activities brought by urbanization,landslides occur frequently.The focus of landslide prevention research should be on exploring a theoretical way to study of regional landslide risk assessment from the case of single landslide prevention,so as to play a more effective and positive role in avoiding,controlling and reducing the macro landslide loss in mountainous villages and towns.Landslide risk assessment is the most economical and effective tool for landslide prevention advocated internationally,which includes landslide susceptibility assessment,vulnerability assessment and value analysis.The problems existing in landslide risk assessment include the difficulty of determining the parameters of unit division,the subjectivity of index system construction and the low spatial adaptability of assessment model.Considering the lack of a desirable system to satisfy the needs of practical application,the mentioned problems are are analyzed depending on the Research and Demonstration of Integrated Disaster Prevention and Mitigation Information System in Villages and Towns supported by National Science and Technology Ministry.The main works and conclusions are as follows:(1)A method for determining the size of grid unit is established based on the image quality of index layers and the classification results.Through analyzing the impacts of flow thresholds of DEM and inversed DEM on slope division,the division method based on orthogonal method is built.Those methods solve the difficulty of determining the parameters of unit division.(2)In consideration of the subjectivities in of index system construction,the loop-rejection clustering algorithm is established to classify the locational indexes under grid condition,and the screening method of index based on information entropy and correlation analysis is built.Those methods could not only eliminate sample data those are less correlated with landslides but also guarantee the independence among assessment indexes and the correlation between each index and landslides.(3)The region-based linear/logical regression model considering the spatial location is obtained by analyzing the impacts of indexes on landslide susceptibility at different locations.This model avoids the problem that a single regression equation cannot make a reasonable assessment while the types and influence patterns of each index are different and improves the spatial adaptability.(4)A landslide susceptibility assessment system for mountainous villages and towns is developed,based on which landslide susceptibility assessments under different units(grid and slope)models(regionalized linear/logical regression model and support vector machine model)are compared.The result shows that the way of data extraction is the main controlling factor that affects the susceptibility assessment at different conditions.(5)A landslide risk assessment considering the resistance to landslides and loss rate of disaster-bearing body is proposed,and the slope division is introduced to the value analysis.On this basis,the landslide risk assessment is completed for Songbai Town.The result shows that landslide risk in Songbai Town is mainly concentrated around road and construction land,and human engineering activity is the dominating factor affecting the distribution of landslide risk.