| Wenchuan earthquake has induced tens of thousands of geological disasters such as landslides and collapses,which are seriously destructive and pose a serious threat to the safety of human life and property.Under the external conditions such as heavy rainfall,the slope in the disaster area after the earthquake is likely to happen again,and even turn into the source of debris flow,which further promotes the destruction of geological disasters.In this paper,Yingxiu Town and Yinxing Township in Wenchuan strong earthquake area are selected as the study areas.Based on the landslide interpretation of remote sensing images of four periods after the earthquake in the study area,the spatio-temporal data set of landslide disaster is established,the frequency distribution characteristics of landslide area in strong earthquake area are explored,and the spatial frequency of landslide area is predicted and evaluated,aiming at providing certain theoretical basis and reference value for landslide disaster prediction and disaster prevention and mitigation in strong earthquake area.Considering the geological structure,topography,slope lithology and hydrological conditions of earthquake landslide in the study area,10 influencing factors of earthquake landslide are preliminarily selected,and the spatial distribution law of landslide in these factors is analyzed.Factors multicollinearity analysis and information gain rate method are used to screen out the main controlling factors of landslide distribution.A variety of statistical distribution functions are compared and analyzed,and the distribution function with high fitting degree to the landslide area frequency curve in the study area is selected to establish the landslide scale frequency distribution model.The characteristics of landslide area frequency distribution in the study area are analyzed,and the frequency variation trend of landslide disasters of any scale at different attribute levels of each landslide main control factor is explored.Finally,combined with landslide scale frequency distribution model and spatial interpolation method,the spatial distribution map of landslide area frequency in the study area is established,and the spatial distribution of landslide area frequency is predicted and evaluated,and the following conclusions are drawn:(1)Based on the remote sensing images of the study area in 2010,2011,2015 and2017,the landslide is interpreted,and the geological environment influencing factors of landslide are extracted from topographic and geological maps,and the landslide disaster spatio-temporal database is established.According to statistics,the area and quantity of landslides are decreasing year by year on the time scale.(2)After the earthquake,the spatial distribution of landslides on four remote sensing images is statistically analyzed,and it is found that the distribution characteristics of landslides in different years are similar in elevation,slope,aspect and topographic relief.Landslides are mainly concentrated in the elevation range of 1458-2491m;The slope range of 30-60 is beneficial to the occurrence of landslides in the study area;The greater the topographic relief,the more favorable it is for the occurrence of landslides.There is a certain negative correlation between ground roughness and landslide.There are many landslides in the south and southeast directions.Landslides on convex slopes are more distributed than those on concave slopes,and landslides on plane slopes are rarely distributed.The lithology of the study area is mainly granite and diorite,and the landslide mainly develops in granite area,followed by diorite area.The closer the distance from faults,water systems and roads is,the more landslides are distributed.On the contrary,the farther away from faults,water systems and roads,the less landslides occur.(3)Information gain rate method can evaluate the influence degree of different geological environment factors on landslide occurrence,so as to screen out the factors that have less influence on landslide.The results of factor screening show that seven factors,such as elevation,slope,aspect,topographic relief,ground roughness,lithology and distance from road,have great influence on landslide occurrence.Therefore,these seven main control factors will be selected to analyze the scale and frequency distribution characteristics of landslide.(4)Six distribution curves,lognormal,three-parameter lognormal,gamma,threeparameter inverse gamma,Weber and three-parameter Weber,are used to fit and compare the landslide area frequency distribution in the study area.Finally,the threeparameter lognormal distribution with high fitting degree is selected to establish the landslide scale frequency distribution model,thus quantifying the frequency distribution characteristics of earthquake landslides in the study area.The exact solutions of three parameters of the four-stage landslide area frequency distribution model are obtained by using the maximum likelihood estimation method.Generally speaking,with the increase of landslide area,the frequency of landslide occurrence has a sharp decreasing trend,that is,the study area is dominated by small and mediumsized landslides,and large-scale landslides are less distributed.(5)The distribution characteristics of landslide area frequency in seven main control factors,such as elevation,topographic relief,slope,surface roughness,slope direction,lithology and distance from road were explored.In high-altitude areas,the frequency of large-scale landslides is low;With the increase of topographic relief,the more large-scale landslides are distributed;The steeper the slope,the more conducive to the occurrence of large-scale landslides;Ground roughness plays a certain role in controlling large-scale landslide changes.Large-scale landslides in diorite area are more distributed than those in granite area.With the increase of the distance between landslide and road,the frequency of landslide area decreases gradually.(6)Based on the principle of landslide similarity and similarity by inverse distance weighted interpolation method and the principle of landslide local weighting by Kriging interpolation method,the spatial interpolation analysis and comparison of landslide area in the fourth stage of the study area are carried out,and the spatial distribution map of landslide area is obtained.Combined with the spatial distribution of landslide area and landslide scale frequency model,the spatial distribution map of landslide area frequency is established,and its zoning analysis is carried out.In order to verify the results of landslide area frequency zoning in the study area,70% of landslide data were randomly selected as the test set for spatial interpolation and the remaining 30%landslide data for verification.The results of landslide area frequency distribution zoning were verified by information entropy theory.The results showed that the greater the landslide area frequency,the greater the entropy value of verification set,which was more beneficial to the occurrence of landslide,indicating that the results of landslide area cumulative percentage zoning were ideal. |
PDF Full Text Request