Study On The Risk Assessment Of Extreme Precipitation-Induced Geohazard Chain In Shuicheng County,Guizhou Province Based On Scenario Simulation

Geohazards are widely distributed in China,often bringing economic losses and casualties,which seriously affect the sustainable development of society and economy.Shuicheng County,Guizhou Province,is one of the counties where geohazards are frequent and serious,and extreme precipitation is their main trigger.Once the extreme precipitation-induced geohazards chain occurs,it will pose a serious threat to the safety of people’s lives and property in mountainous areas of Shuicheng County.Carrying out risk assessment of extreme precipitationinduced geohazard chains in Shuicheng County can provide great support and help to the relevant local management departments to prepare for the prevention and resistance to disasters in advance,and to make rescue and relief decisions during disasters.In this study,the risk assessment of extreme precipitation-induced geohazard chain in Shuicheng County,Guizhou Province based on scenario simulation was carried out according to the regional disaster system theory,natural disaster risk formation theory and disaster chain formation mechanism.This paper analyzed the correlation of extreme precipitation-induced geohazard chains with their influencing factors and the formation mechanism of extreme precipitation-induced geohazard chains.The Bayesian optimized multivariate machine learning models and their integrated models were used to construct a susceptibility assessment model for extreme precipitation-induced geohazard chains,and the optimal model was selected to assess the susceptibility of extreme precipitation-collapse,extreme precipitation-landslide,extreme precipitation-debris flow and the extreme precipitation-induced integrated geohazard chains,respectively.Combined with the simulated extreme precipitation scenarios,the hazard assessment studies under different extreme precipitation scenarios were conducted.Based on the hazard assessment,the exposure,vulnerability of the disaster-bearing body and disaster prevention and mitigation capacity were assessed based on the risk assessment model,and the risk of extreme precipitation-collapse,extreme precipitation-landslide,extreme precipitationdebris flow as well as the extreme precipitation-induced integrated geohazard chains were dynamically assessed.Based on the results of the integrated susceptibility assessment of extreme precipitation-induced geohazard chains,the population risks of extreme precipitationinduced geohazard chains under different future development scenarios were predicted by coupling the changes of extreme precipitation in different recurrence periods and different population exposure levels.The main research results are as follows:(1)Analysis of influencing factors and causal patterns of extreme precipitation-induced geohazard chains in Shuicheng countyThere are 240 historical hazard sites in the study area,including 45 landslides,155 landslides,and 40 debris flows,and these geohazard points have different degrees of relationship with each environmental factors.When extreme precipitation occurs,initially the soil water content increases,the groundwater level is raised,the slip resistance of fissures and other separating surfaces decreases while the gap width continues to expand,the slope center of gravity keeps moving outward,and when the slip resistance is less than the sliding force,the center of gravity of the avalanche slide migrates to the outside of the slope body,it will trigger a collapse or landslide.And if there is a continuous rainstorm,it will also drive the loose rock and soil to the bottom of the slope,forming a debris flow.(2)Assessment of the susceptibility of extreme precipitation-induced geohazard chainsLogistic regression(LR),random forest(RF),gradient boosting decision tree(GBDT),deep neural network(DNN),recurrent neural network(RNN),and convolutional neural network(CNN)multiple machine learning models and Bayesian optimized models of other 5models except LR(RF＿B,GBDT＿B,DNN＿B,RNN＿B,CNN＿B)were selected to construct 11 single machine learning models,and the individual models were integrated according to Stacking combination strategy to construct 11 integrated models.The accuracy of each model was compared based on multiple validation methods.The optimal single machine learning model was the GBDT＿B model(accuracy of 0.788)and the optimal integrated model was the integrated model with RF＿B model as the combination strategy(accuracy of 0.795).Comparing the two models,the integrated model has better performance.The integrated multivariate machine learning model with RF＿B as the combined strategy was used to assess the susceptibility of extreme precipitation-collapse,extreme precipitation-landslide and extreme precipitation-debris flow hazard chains and extreme precipitation-induced integrated geohazard chains,respectively.The results show that the high susceptibility areas of extreme precipitationcollapse and extreme precipitation-landslide hazard chains are distributed near the faults throughout the county,while the high susceptibility area of extreme precipitation-debris flow hazard chains is more concentrated in the southern part of Shuicheng County.The lithology of the high susceptibility area of extreme precipitation-induced geohazard chains in Shuicheng County is mostly claystone,sandstone and basalt,and most of the high susceptibility areas are located in the middle and high elevation areas between 1000 m and 2000 m.(3)Hazard assessment of extreme precipitation-induced geohazard chains under different precipitation scenariosThe overall distribution of the hazard zones of each single hazard chain and the integrated extreme precipitation-induced geohazard chain under different precipitation return period scenarios in the study area are similar with susceptibility.With the increase of precipitation return period,the low-level hazard zones decrease,and the high-level hazard zones increase significantly,and the proportion of medium-level hazard zones does not change.The very high hazard zone of extreme precipitation-induced geohazard chains accounted for only 2.22% in the 5-year precipitation scenario,but increased to 11.27% in the 20-year precipitation scenario.(4)Risk assessment of extreme precipitation-induced geohazard chains under different precipitation scenariosThe overall exposure level of disaster-bearing bodies in the study area is low,and only less than 1% of the high-class exposure areas are mainly concentrated in the county town of Shuicheng County and the residential areas of the more populated townships.Very low vulnerability areas accounted for 61.89%,while very high vulnerability areas were only 2.83%.The very high and high disaster prevention and mitigation capability areas in Shuicheng County are mainly concentrated in the highways,national highways and county city roads.The risk of extreme precipitation-collapse,extreme precipitation-landslide and extreme precipitationmudslide hazard chains and extreme precipitation-induced integrated geohazard chains in Shuicheng County are mainly low risk,while the very high-risk areas are mainly concentrated in most areas of Mi’luo Town and some areas of Shuicheng county town,Panlong Town,Douqing Town and Fa’er Town.With the increase of precipitation recurrence period,the very low risk areas of each hazard chain and integrated geohazard chain decreased significantly,while the other four risk level areas increased.(5)Population risk prediction of extreme precipitation-induced geohazard chains under different future development scenariosThe high population risk areas of the extreme precipitation-induced geohazard chain for different future development scenarios in Shuicheng County are mainly concentrated in Fa’er Town,Yingpan Town,and Shunchang Town in the southwest,and A’ga Town and Mi’luo Town in the central part,while Baohua Town and Qinglin Town in the north are at a lower risk level.The population risk increases with the increase of extreme precipitation.Since the hazard of extreme precipitation-induced geohazard chains in the 2018 base year is much higher than that of different future development scenarios,its population risk is higher than that of different future development scenarios.Among the future development scenarios,the fossil-fueled development path-representative concentration pathway 8.5 emission(SSP585)scenario has the lowest population risk due to the lower exposed population level,followed by the sustainability path-representative concentration pathway 2.6 emission(SSP126)scenario with slow population growth and the middle of the road path-representative concentration pathway4.5 emission(SSP245)scenario with steady population growth,and the highest population risk is the regional rivalry path-representative concentration pathway 7.0 emission(SSP370)scenario with rapid population growth,in this scenario,if an extreme precipitation of one in 20 years occurs,there will be four towns including Fa’er Town,A’ga Town,Yingpan Town and Shunchang Town with a population risk of more than 500 people,and in the future these towns need to strengthen the risk warning and risk management capabilities of extreme precipitationinduced geohazard chain to reduce the casualties.Based on the scenario simulation technique,this study dynamically assesses the risk of extreme precipitation-induced geohazard chains and the population risk of extreme precipitation-induced geohazard chains under different development scenarios in the future,which provides a new idea for the risk assessment study of extreme precipitation-induced geohazard chains.The constructed assessment model can be extended and applied to other regions with similar geological conditions.The results of the study can provide a scientific basis for relevant departments to set the corresponding warning levels,develop emergency plans and response measures by establishing the risk levels of extreme precipitation-induced collapse,landslide,debris flow and integrated geohazard chains under different rainfall conditions.Meanwhile,the prediction of population risk of extreme precipitation-induced geohazard chain under different development scenarios in the future can provide theoretical guidance for relevant departments to develop corresponding disaster prevention and mitigation plans in response to the risk of extreme precipitation-induced geohazard chains under different development scenarios in the future.