Analyzing Landslide Susceptibility In The Upper Mingjiang Basin

With the development of society, the improvement of economic level and the increasing of population, the scope of human activities and the scale of engineering activities are increasing, the landslide hazard get worse year by year. The landslide has the characteristics of regional, multiple and serious in our country, which seriously threaten the safety of human life and restrict the development of social economy. So, it is significant to study the related research of landslide hazard.The research area of this paper is the upper MinJiang basin, which is located in the northwest of Fujian Province. Lots of mountains and rivers in the area, and the ecological environment of the area is fragile, so landslide hazard is very serious. On the basis of summarizing the experience about landslide research of numerous experts and scholars, this paper explored the distribution law and the formation mechanism of the landslide in the study area, selected the evaluation factors of landslide susceptibility, and studied on regional landslide susceptibility assessment based on different evaluation models. The research result is reliable, which has certain practical significance. The research in this paper mainly includes the following aspects:?1?This paper summarized the landslie characteristics, temporal and spatial distribution law of landslide hazard in the upper MinJiang basin, based on the research results of other experts and field survey data, analysised the formation reason of the landslide, selected the landslide evaluation factors according to the above. Finally the 9 factors, such as elevation, slope, relief, soil type, leaf area index, rainfall, river, road and landuse types, were selected to explore the relationship between each factor and the landslide distribution, then evaluated the regional landslide susceptibility.?2?Leaf area index ?LAI? not only reflect the vegetation cover, but also closely relate to soil moisture, and can better reflect the relationship between vegetation and landslides, so this paper used LAI factor to evaluate the regional landslide susceptibility.The content research included the field measurement of the leaf area index, the determination of the remote sensing inversion model, and the extraction of the area LAI. The empirical model of LAI is LAI= 8.6×NDVI^3.232, the determining coefficient of which is 0.685.?3?According to the selected evaluation factors, information value model and Logistic regression were used to establish the mathematical model for the evaluation of the landslide susceptibility in the upper MinJiang basin. In addition, an unsupervised classification method based on the combination of ISODATA clustering and maximum likelihood method was introduced into the area of regional landslide susceptibility, the applicability of this method in practical application was researched.?4?Analysised of three results of landslide susceptibility from three models, explored the characteristics of each model, and combined the precision accuracy of ROC curve. The results had showed that the Logistic regression model had the best prediction results, and the unsupervised classification method based on the combination of ISODATA clustering and maximum likelihood method got the most poor result, but each accuracy were similar, precision accuracy were all at a higher level, so each models were suit for the landslide susceptibility in the upper MinJiang basin. The unsupervised classification method based on the combination of ISODATA clustering and maximum likelihood method is simple and easy to operate, less process directly the evaluation factor data were quantified and normalized can be carried out landslide susceptibility division, and the result of this method has a high degree of agreement, and landslide susceptibility can be simple division in lack of landslide distribution data, so it has provided a new method for the simple accurate and fast classification of landslide susceptibility.