Quantitative Hazard Assessment Model And Method Of Debris Flows Considering Statistical Uncertainty

Debris flows,which often cause casualties and economic loss,are common geological hazards in mountainous regions.In order to mitigate the debris flow hazard,it is necessary to construct debris flow hazard prevention project.Debris flow hazard assessment provides important references for the design of hazard prevention project.In hazard assessment,the joint probabilistic model of debris flow properties(e.g.,total discharge Qtotal and maximum impact pressure Pmax)should be developed using observation data.Then,the exceedance probability of debris flow is calculated based on the probabilistic model,hence quantitatively assessing the debris flow hazard and providing reference for the design of hazard prevention project.The identification of probabilistic model and model parameters rely on observation data,which is affected by various uncertainties(e.g.,observational error,stochastic fluctuation).These uncertainties lead to the statistical uncertainty in model parameters,and eventually cause the uncertainty in the exceedance probability.Although many hazard assessment studies have been done in literature,the influence of statistical uncertainty in model parameters on the exceedance probability is not taking into account.How to calculate the exceedance probability considering statistical uncertainty and assess the debris flow hazard in a proper and robust manner remains a key issue to be solved in hazard assessment of debris flows.In order to tackle the problem mentioned above,this study proposed a set of quantitative hazard assessment model and method of debris flows considering statistical uncertainty.The proposed method is comprised of three components:construction of candidate probabilistic models,Bayesian model selection and parameters identification,evaluation of exceedance probability and its uncertainty.In this study,the joint probabilistic models of total discharge and maximum impact pressure are constructed using copulas and Gaussian mixture model,respectively.Bayesian functions based on these two types of model are formulated.To tackle the difficulty in computation of Bayesian function,a structural reliability-based Bayesian updating algorithm is employed.The proposed method is able to identified the most probable model and the posterior distribution of its parameters,which are then used to evaluate the exceedance probability of debris flows and quantify its uncertainty,hence yielding a reliable and robust hazard assessment of debris flows.This study briefly introduces the backgournd and aim of the research.Then the exceedance probability of debris flows is defined and its significance in practice is illustrated with examples.Bayesian model selection and parameter identification theory and methods for solving Bayesian function are introduced.The probabilistic models are constructed based on Gaussian mixture model and copulas.The effectiveness of the proposed model and method is validated by simulated data.Finally,the proposed method is applied to Jiangjia Ravine for debris flow hazard assessment.Results show that:ignoring the statistical uncertainty in model parameters may underestimate the exceedance probability of debris flow,leading to an unconservative hazard assessment result,therefore it is necessary to take statistical uncertainty into consideration;the uncertainty of exceedance probability increases as the exceedance probability decreases,and the uncertainty of exceedance probability of extreme event should be considered in the design of debris flow hazard prevention project;the hazard assessment for Jiangjia Ravine indicates that the exceedance probability of extreme event for continuous flow is larger than that of surge flow,and special attention should be given to continuous flow in the design of mitigation strategies.