Abstract:

Based on previous research experiences and a large number of field trips, a number of important impact factors of debris flow are selected to forecast debris flow risk range more objectively and more accurately. Hundreds of debris flows in Wudongde region classified by their accumulation areas are evaluated with multiple regression analysis and the prediction models about the morphological parameters of different types of debris flows are obtained. A new method which can search a prediction mathematical model for regional debris flow risk range by changing the steps is presented. Every possible mathematical model is searched through the computer and optimal prediction models are found by comparing average error. This method and the traditional forecasting methods are both applied as examples in Wudongde. The results show that the error calculated by the stepwise regression analysis method of literature\[3\] is much bigger than the error calculated by the suggested method. The former is from 10.5% to 29.6%, and the later is between 6.7% and 9.2%.

Key words: debris flow, risk range, multiple regressions, search the optimal solution, Wudongde region