Study On Logistic Regression Model Applied To Regional Slope Stability Evaluation Based On GIS

I. Purpose, Significance& TheoryWith the characteristics of wide distribution, frequent happening and quick causing calamity, landslide is one of the main disasters in the world and endangers the country's economical development. It has been more than 20 years since the study of regional landslide disaster evaluation and mapping in the international world and it has made remarkable progress during the recent years. The landslide disaster must abide by a series of basic theories and procedures. Varnes has put forward three ultimate theories on partition landslide disaster, which suggested by the UN in 1984:(1) Past and present conditions are the key factors to predict the future;(2) The main reason that cause the landslide can be found;(3) Landslide risk and disaster probability can be evaluated.The methods of landslide disaster spatial evaluation can be concluded as statistics, sensitivity mapping, information and all kinds of experience, etc.Although landslide forecast theories have different evaluation methods, they are based on mechanism analyses, favorableness, and their similarities are to be taken fully consideration of overlapping for various geological environments to evaluate the possibility of causing landslide in certain geological environment and to circle the relatively unsafe "dangerous sector" in the hazardous area center. Different ways of overlapping are adapted in evaluation process, and the prediction result is expressed by section graphs.The characteristics of landslide prediction study are intersected and penetrated by various kinds of theories so that different kinds of modern theories such as information theory, systematic theory, fuzzy mathematics, expert system, neural network, as well as computer technology are all applying to landslide spatial prediction.The above-mentioned methods have the unique advantages and characteristics, which play quite important roles in qualitative and quantitative slope stability evaluations, but there are still some weak points. Generally speaking, it can be summarized as the following respects: First, the model's fitness ability cannot be assessed; Second, the regression coefficient can not be tested; Third, landslide sensitivity degree can not be offered. And these points are exactly the most important ones in the engineering practice. The model can't carry on the whole assessment, which...