Regionalization Analysis Research Of Landslide Susceptibility In The Three Gorges Area,China

Landslides are one of the natural disasters that cause huge losses and human casualties in the world today.Landslides is the second geological disaster only to earthquakes and poses a great threat to the safety of people's lives and property and to economic construction.In view of the severe situation of landslide disasters and the huge loss they bring,the effective investigation and monitoring of landslide distribution and activity conditions,landslide prediction and assessment,and the provision of scientific basis for the prevention and treatment of landslide hazards are urgent needs of the affected areas and landslide scientists.In this paper,the key issue in the study of geological hazards is landslide susceptibility evaluation.Landslide susceptibility is also called landslide sensitivity and it belongs to the category of landslide spatial scope prediction.Taking the Zigui-Badong section of the Three Gorges Reservoir area as the research area,a study was conducted to try to integrate GIS,RS,and data mining methods into landslide susceptibility assessment to resolve landslide susceptibility.In the evaluation,problems such as low efficiency,poor accuracy,and time-consuming and labor-intensive problems were met,and landslide geological disasters were informatized and scientific.This paper combines GIS with traditional logistic regression models,information theory,K-means clustering analysis,and other methods to perform landslide-susceptibility maps on the Zigui-Badong section of the Three Gorges reservoir area based on previous landslide hazards.To achieve the goal of reducing disaster losses,protecting the lives and property of local residents and economic construction.The main research content includes the following aspects:(1)Landslide causative factor extraction and reclassification study.The extraction of causative factor is a crucial part of landslide susceptibility analysis.Landslide causative factor can be divided into geomorphology(elevation,slope,aspect,slope form,etc.),geology(lithology,rock bed structure,etc.),hydrology(catchment slope,catchment area,distance to river,etc.),and surface coverage(land use,normalized vegetation index,normalized water index,etc.),rainfall,and human engineering activities.These factors are the key to landslide susceptibility studies and the source of data for the entire study.The impact of each causative factor on the occurrence of landslide events varies depending on the geographical environment in which the study area is located.After the extraction of landslide causative factors is completed,the extracted factors need to be reclassified.In previous studies,the number of class was determined artificially for each factor,and the subjectivity was relatively large.In this paper,the author used the information theory which basis of Shannon's entropy to divides each hazard factor into corresponding classes so that each causative factor can maximize the amount of information it can provide in landslide susceptibility prediction.(2)Research on the regionalization of the study area.In recent years,the regionalization of the study area has become a hot spot for landslide susceptibility research.The regionalization of the study area is to use some technical means to divide the study area into several sub-areas,so as to achieve landslide susceptibility analysis in each sub-area to achieve higher prediction accuracy.The key to the regionalization of the study area is the regionalization method it adopts and the landslide susceptibility analysis after the regionalization is based on the hazard factor or model basis.In this study,the author first used the deterministic coefficient method to calculate the deterministic coefficient value(CF)for each category of each hazard factor,and then based on the landslide property of each computation unit,it was determined whether the computation unit used the landslide susceptibility prediction.The hazard factors,so as to obtain the combination of hazard factors for each calculation unit(a two-dimensional vector consisting of "0" and "1").Then,the K-means clustering method is used to cluster the two-dimensional vectors of all the computational units in the study area,so that all the computational units are grouped into different categories,so that the computational units within each category have similar factor properties,and the difference in the factor attribute of the calculation unit in the category is maximized.Experiments show that regionalization of the study area can significantly improve the accuracy of landslide susceptibility prediction.(3)Impact of landslide susceptibility prediction on the development of local towns.Because the study area is a typical mountainous landform,landslide disasters are the main threat to the construction of local towns.In this paper,the author takes the landslide susceptibility prediction result(K=4)as one of the factors of urban development,and uses the Analytic Hierarchy Process(AHP)to obtain the stability map of local urbanization development,then providing decision support for local township constructors.