【摘要】 在泥石流风险评估与防治中,需要根据既有观测数据建立泥石流参数的概率模型(如总流量Qtotal和最大冲击力Pmax的概率分布),计算泥石流超越概率,从而为泥石流灾害评估和防治措施设计提供参考依据.然而,由于观测数据受到多种不确定性因素(比如降雨强度不确定性、岩土体参数变异性以及测量误差)的影响,导致所计算的超越概率具有波动性或不确定性,传统方法中超越概率的点估计值无法合理地考虑其波动性的影响.基于此提出了一种基于Qtotal和Pmax观测数据识别泥石流参数概率模型的贝叶斯方法.根据所提方法识别的泥石流参数概率模型不仅可以计算泥石流的超越概率,还能够合理地考虑超越概率的波动性对泥石流灾害风险水平的影响.以蒋家沟泥石流观测数据为例验证了所提出的方法.结果表明:忽略基于观测数据所计算的超越概率的波动性会导致不保守的风险评估结果.更多还原

【Abstract】 In quantitative risk assessment and management of debris flow,it is necessary to develop probabilistic model for debris flow quantities,e.g.total discharge Qtotaland maximum impact pressure Pmax,and calculate the exceedance probability of debris flow to provide reliable references for hazard assessment and/or the design of mitigation strategies.However,this is a nontrivial task because observation data is affected by various uncertainties e.g.rainfall uncertainties,inherent variability of geotechnical properties and measurement errors,which will cause fluctuation or uncertainty in the estimated exceedance probability;while the point estimation of exceedance probability based on conventional method cannot consider the uncertainty.This paper proposes a Bayesian method to identify the probabilistic model for quantitative risk assessment of debris flow based on observation data of Qtotaland Pmax.The probabilistic model obtained from the proposed method provides not only the exceedance probability but also its associated uncertainty.For illustration,the proposed method is applied to developing the probabilistic model of Qtotaland Pmaxfor quantitative risk assessment at Jiangjiagou Ravine,China.The results show that ignoring the uncertainty in the exceedance probability estimation based on observation data leads to unconservative risk assessment results.更多还原