Research On Landslide Fractal Features And Hazard Assessment In Mid-Yunnan Water Diversion Reservoir Area, Jinsha River

China is one of the countries rich in hydropower resources in the world.According to the general survey of national hydropower resources, China’sexploitable hydropower and annual energy output both rank first in the world.However, the distribution of hydropower resources is extremely uneven, and most ofthe water resources are mainly concentrated in the southwestern mountainous areas,where terrain vertical drop is very high, rainfall concentrates and there develops largeamounts of rivers. Hence, this area is abundant in hydropower resources. SouthwestChina is located between the first and the second ladder, southeastern margin of theQinghai-Tibet Plateau. Because of the high mountains and steep slopes, complicatedgeological condition, frequent seismic activity and fragile ecological environment,various geological hazards occurs frequently. Mountain landslide disasters especiallydevelop,which not only inflicts enormous life and property losses, but also restrictsthe local economic development. Therefore, it is highly significant to study hazardassessment of regional landslides.The present paper is supported by2010Non-Profit Scientific Special ResearchFunds of Ministry of Water Resources (Grant no.201001008), which is named as“Research on the hazard effect and countermeasures of large reservoir bank landslidesin the southwest of China: a study on the characteristics survey and geneticmechanism classification of landslides. The study area is in the mid-Yunnan waterdiversion reservoir of Jinsha River. Based on the3S (RS, GIS, GPS) technology andcombined with field investigation and remote sensing interpretation, a detailed surveyon geological hazards of the landslides is carried out and the landslides’ location,quantity, distribution and scale in the research area are totally ascertained. Thecharacteristics of landslide spatial distribution in the study area is summarized basedon the above research work. And then, the genetic mechanism of the landslides isanalyzed in detail. Moreover, the fractal feature of the landslides is studied and thehazard assessment of the landslides is mainly made.Based on the fractal geometry theory, spatial analysis function and mapping function of ARCGIS, the fractal characteristics of geometric configuration and spatialdistribution of landslides are fully studied in great details. Meanwhile, the basicdiagrams of elevation, slope and aspect are extracted from the digital elevation modelof study area. Then the variable dimension fractal method is adopted to evaluate thequantitative relationship between the inducing factors such as elevation, slope, aspectand the distribution area of landslips, which further gives support to the fractalgeometry basis for establishing the evaluation system of landslide hazard assessment.The fractal dimension of landslide geometric shape is attempted as one of therisk index factors of landslides. Those selected twelve evaluation factors are extractedto reduce dimension and eliminate the multicollinearity and redundancy between theoriginal data based on the kernel principal component analysis. The intelligencemodel of landslide hazard assessment is established on the basis ofmulti-classification support vector machine theory. Furthermore, particle swarmoptimization is used for optimizing the kernel function parameter δ and penaltycoefficient C to obtain a better effect evaluation. The results of recognition are wellcoincident with the field survey.