Research On The Town-scale Geological Hazard Susceptibility Assessment

The southwest of Zhejiang Province is dominated by mountains and hills,showing complex geological environmental conditions.In recent years,under the influence of extreme weather,the geological hazards in southwest Zhejiang are frequent.To protect the safety of people and property and support the geological hazard prevention and control work of the local government,it is urgent to carry out a high-precision geological hazard risk investigation and evaluation for the township(town,street)at a slope scale.However,with the reduction of the study area and the relatively fewer historical hazard samples,it is challenging to evaluate the regional susceptibility of geological hazards with high precision under small sample data.Taking Chongtou Town,Yunhe County,Lishui City,Zhejiang Province as the research area,this paper analyzed the mechanism of geological hazard in this area utilizing field investigation and remote sensing interpretation and obtained 9 evaluation factors of geological hazard susceptibility.Combining the respective advantages of semi-quantitative and quantitative single evaluation methods,two kinds of geological hazard susceptibility evaluation models were established,which were coupled with grey relational degree(GRD)and analytic hierarchy process(AHP),and frequency ratio(FR)and multiple regression analysis methods.The reliability verification methods of evaluation models were established from three aspects:index distribution,prediction accuracy,and zoning rationality.It solves the problem of low accuracy of model prediction for the limited sample data of historical hazards.On this basis,a total of 8coupled evaluation models in the above two systems were used to evaluate the geological hazard susceptibility of Chongtou Town,and the reliability of the evaluation results was analyzed.Finally,the optimal evaluation model was recommended,and the high-precision geological hazard susceptibility zoning map was obtained.The main conclusions are as follows:(1)The investigation and analysis of hazards in the study area show that:the main geological hazards in this area are rainfall-induced landslide,falling-type topple,and slop-type debris flow;from the perspective of spatial distribution,geological hazards are mainly distributed in hilly gullies and low hills where human engineering activities are frequent;from the time distribution,it was found that the occurrence of geological hazards in this area is closely related to the influence of typhoons and rainstorms.(2)The hazard mechanism in the study area was revealed.According to the common characteristics of geological hazards,lithological and geological parameters,slope type,overburden thickness,distance from the fault,slope gradient,altitude difference,slope aspect,slope shape,and the height of the cut slope are the best combination of evaluation factors.The results of frequency ratio analysis show that:the geological hazards are most likely to occur in rock fracture structures with a continuous weak structural plane,loose structure,down-slope drift in geotechnical composite slope,and in the states that the overburden thickness is 6～12 m,the distance from the fault<100 m,the slope gradient is 25～35°,the altitude difference is 100～300 m,the slope aspect is east,the slope shape is convex and the height of cut slope is 12～25 m.(3)Using the advantage of GRD analysis in processing small sample data,the GRD between 9 evaluation factors and historical hazards was analyzed,and three discriminant matrices of AHP were constructed based on the GRD values,the exponential scale,and the ranking of GRD values respectively.Three coupled models of GRD and AHP(G-A1,G-A2,and G-A3)and a single GRD model(G)were established.Finally,reliability analysis was carried out on the evaluation results from distribution characteristics,ROC curve,susceptibility intensity index,and others.It was found that:1)In terms of susceptibility index distribution,coupled model G-A1 and single model G are reasonable;2)In terms of prediction accuracy,the AUC value of the ROC curve of coupled model G-A1 is the best;3)In terms of the susceptibility zoning rationality,coupled model G-A1 has the smallest proportion of high-susceptibility area,the largest high-susceptibility intensity index.(4)By frequency ratio(FR)analysis,the normalized FR values of 9 evaluation factors in different states were obtained,and it was utilized in multiple regression analysis with the potential of processing small sample data.Five coupled evaluation models(FR-LR,FR-PLSR,FR-RR,FR-SR,FR-DA)were established based on logistic regression(LR),partial least squares regression(PLSR),Ridge regression(RR),stepwise regression(SR)and discriminant analysis(DA),respectively.The reliability of the evaluation results was analyzed from the susceptibility index distribution,ROC curve,and susceptibility intensity.It was found that:1)In terms of the distribution characteristics of susceptibility index,FR-PLSR,FR-RR,and FR-DA were close to the standard normal distribution and showed better performance;2)In terms of prediction accuracy,the AUC indexes of the five models were relatively large,ranging from 0.86to 0.87;3)In terms of the susceptibility zoning rationality,FR-PLSR and FR-RR showed better performance,with relatively small proportion of high-susceptibility area and relatively large high-susceptibility intensity.(5)Pearson correlation analysis was carried out on the two coupled systems and their evaluation results.The results showed that the recommended coupled models G-A1,FR-PLSR,and FR-RR are highly correlated(Pearson correlation degree above0.8)in the ability to judge historical hazards and high-susceptibility areas.And they are superior to other similar coupled models.Finally,the G-A1 model with better engineering applicability was used to evaluate the geological hazard susceptibility of the study area,and a high-precision geological hazard susceptibility zoning map was obtained.The areas of high,medium,low,and extra low slop units are 5.01 km~2,87.78km~2,97.33 km~2,and 38.32 km~2,respectively.They account for 2.19%,38.43%,42.61%,and 16.77%,respectively.Our research has an important reference for geological hazard susceptibility assessment in southwest Zhejiang Province and provides an effective coupled evaluation model for high-precision geological hazard assessment and prevention and control.