Study On The Prediction Model Of Landslide Induced By Rainfall

Based on a large number of landslides and rainfall data in the Chongqing region, this paper conducted a in-depth research on the following issues:statistical regularities of rainfall-induced landslides in Chongqing region, statistical threshold of rainfall landslide of Chongqing, the landslide probability prediction model based on statistical analysis, etc, besides, on this basis, the author selected a landslide prototype, developed the rainfall program and carried out rainfall-induced landslide model test, then getting the main achievements as follows:(1) This paper analyzes the rainfall characteristics and the basic characteristics of rainfall-induced landslide of the Chongqing region, then getting the landslide quantity and annual rainfall trend of Zhong County.(2) Through analysis of Rainfall-induced landslide data, in this paper, the author achieved the rainfall intensity-lasted expression of Chongqing, cumulative rainfall three days before landslide occurred and the day rainfall threshold expression, cumulative rainfall three days before landslide occurred and the day maximum hourly rainfall threshold expression.(3)The author analyzed the rainfall landslide data by SPSS software, got the logistic regression model of rainfall-induced landslide in Chongqing region, this model have passed the software fitting goodness test and been tested by use of historical rainfall landslide data instances of Chongqing, the correct rate is90%. This paper also discussed the relationship between maximum hourly rainfall induced by landslides and the day rainfall without antecedent rainfall on the basis of the logistic regression model.(4) The author analyzed the correlation and partial correlation coefficients between rainfall factors on the occurrence of landslides and landslides quantity through SPSS software, finally selected4d cumulative rainfall for landslide probability analysis. When the early rainfall attenuation coefficient K is0.6, the correlation coefficient of4d effective cumulative rainfall to landslides occurrence and landslides quantity is the highest. On this basis, the author respectively established the4d effective cumulative rainfall-landslide probability model, the day maximum hourly rainfall-landslide probability model,4d effective cumulative rainfall-landslide occurrence probability correction model. By the inspection of historical rainfall landslide instance, the correction model prediction accuracy rate still has to be improved. (5) The author carried out the landslide model test under four rainfall conditions, four groups of test experienced different degrees of collapse, slippery stream, the failure of overall sliding. Under different rainfall pattern, critical rainfall is about20mm when slope began to destory, the middle peak starts to break down earlier than incremental rainfall.(6)When the rainfall parameters applied into the model test, the calculation results of the two sets of predictive models are more consistent with the model test. The P in correction prediction model is more consistent than the effective cumulative rainfall-landslide occurrence probability P (R’) in the pre-trial stage on test phenomenon. Cumulative rainfall (low intensity, long duration) can more easily induced landslides overall damage; But the short-time heavy rainfall is more likely to induce the partial collapse and stream slippery destruction of landslides. Certain cumulative rainfall coupled with high-intensity rainfall can easily induce landslide damage.