| Zhen’an County is located in the southwest of Shangluo City,Shaanxi Province.The geo-logical structure is complex,the mountains are forested,the ravines are vertical and horizontal,the terrain is complex and changeable,the deep faults are developed,and the geological envi-ronment is poor.At the same time,the rainfall in this area is large and concentrated in the rainy season,which is the main factor causing geological disasters.This paper studies the sensitivity of geological disasters based on machine learning and the continuous probability threshold model of rainfall-induced geological disasters based on rainfall intensity-rainfall duration,and analyzes and studies the early warning model and program of rainfall-induced geological dis-asters in Zhen’an County.It provides theoretical support for early warning and prevention of geological disasters.The specific research contents and results are as follows:（1）A geohazard sensitivity evaluation index system including four criteria layers with a total of 14 index factors was constructed for Zhen’an County.Machine learning methods such as random forest and logistic regression were used to evaluate the sensitivity of geological haz-ards in the study area,and an integrated model was established by fusing the random forest model and logistic regression model.In order to optimize the machine learning model a Bayes-ian optimization algorithm based on TPE was used for hyperparameter optimization,and after comparison and analysis it was found that the integrated model had better comprehensive eval-uation indexes and higher AUC values,which were more in line with the actual situation.（2）A continuous probability threshold model of rainfall-induced geological hazards in Zhen’an County was established.The fractal power exponential relationship between landslide frequency and cumulative rainfall was used to model the pre-effective rainfall in Zhen’an County,and the attenuation coefficient of the pre-effective rainfall was obtained.Then,a thresh-old model about the average effective rainfall intensity-duration（I-D）was established,and the critical threshold curve was determined by quantile regression to obtain the threshold models with different critical probabilities.And the logistic regression model was applied to obtain the equation 1/P=1+e（2.52214+0.02422*D-0.05005*I*D）of the continuous probability threshold model for rainfall-induced geohazards,whose goodness of fit is 0.945,which is a good fit.Finally,the threshold model was adjusted and optimized by considering the effect of rainfall on the day.（3）The geological disaster weather warning model of Zhen’an County is constructed,and the geological disaster weather warning procedure of Zhen’an County is proposed.The weather warning model of Zhen’an County is constructed by using the probability quantification model of geological disaster-causing factors,and the weight coefficients are determined and optimized by the entropy weighting method to obtain the warning model,whose expression is W=0.389H+0.611Y,and four warning levels are divided according to the different warning val-ues W.Based on the meteorological warning model,a warning procedure is developed for rain-fall-induced geological disasters in Zhen’an County,and the meteorological warning for geo-logical disasters in Zhen’an County is realized. |
PDF Full Text Request