汶川震区暴雨泥石流活动规模预测模型研究

The magnitudes of rainfall-induced debris flows in the Wenchuan earthquake area that are calculated by the existing method are lower than the actual values. Solving this issue has a very important theory and practical value in debris flow control and prevention..In order to predicting of the magnitude of rainfall-induced debris flows in the Wenchuan earthquake area, the field investigation, experiment and theoretical analysis are used. The main content of this study include the dynamic recovery mechanism of the soil density for the loose materials, the interaction mechanism of soil density recovery and debris flow magnitude, and the magnitude-frequency relationship of the rainfall-induced debris flows. The research objectives is the following: (1) to understand spatial-temporal evolution characteristics of the loose soil materials; (2) to clear the main impact factors on the magnitude of rainfall-induced debris flows in the Wenchuan earthquake area; (3) to construct an estimation model of magnitude of rainfall-induced debris flows in the Wenchuan earthquake area. The main purposes of this study are: (1) understanding the calculation of key parameters in the Wenchuan earthquake area, such as frequency of debris flow occurrence, debris flow discharge and debris flow volume; (2) providing the basis and technical support for the debris flow prevention in the strong earthquake area; (3) providing scientific evidence for the prevention and reduction of debris flow hazards.

采用现有计算方法估算的震后暴雨泥石流活动规模远远小于震后的实际值。解决震后暴雨泥石流活动规模估算值严重偏低的问题，对泥石流防治工程设计和提高治理效应具有十分重要的理论意义和现实价值。本项目密切结合汶川震区暴雨泥石流活动规模预测的问题，采用现场调查、理论分析和模拟实验的方法，开展震后暴雨泥石流物源土体密度的动态变化机制分析，探索物源土体密度动态变化与泥石流活动规模变化的相互作用机理，研究震后暴雨泥石流活动规模与频率的关系。研究目标是获得汶川震区暴雨泥石流物源的时空演化特征，揭示震后暴雨泥石流活动规模的主控因素及作用机理，构建震后暴雨泥石流活动规模的预测模型。通过上述研究，可进一步认识汶川震区暴雨泥石流的发生频率、峰值流量及规模等关键参数，也可为强震区的泥石流防治工程设计提供技术支撑，为促进和深化泥石流灾害防灾减灾提供重要的科学依据。

通过对震后暴雨泥石流灾害进行现场调查，结合室内的科学实验、理论分析和数值模拟，探讨了暴雨泥石流峰值流量-频率关系，分析了震后暴雨泥石流活动规模的控制因素，构建了基于经验统计关系的汶川震区暴雨泥石流活动规模预测模型，提出了基于数值模拟的暴雨泥石流活动规模预测模型。搜集了典型暴雨泥石流的地形地貌、地层岩性和泥石流活动规模等数据，构建了汶川震区暴雨泥石流数据库。将数据库数据分为训练集（占总数的80%）和验证集（占总数的20%）。基于Boruta算法计算的因子重要性，确定震后暴雨泥石流活动规模的控制因素。泥石流一次冲出量的控制因素则为流域高差、流域内松散物源面积和流域平均坡度，而泥石流最大冲出长度的主要控制因素为泥石流冲出量和流域高差。基于多重共线性分析、赤池信息准则和稳健回归分析，构建了基于经验关系的震后暴雨泥石流最大冲出距离预测模型和暴雨泥石流一次冲出量预测模型。对预测模型的性能和不确定性进行了评价。数据量的多少在一定程度上对不确定性评价有较大的影响。此外，基于滑坡水动力触发模型和两相流模型，对典型泥石流的冲出量进行了数值模拟。研究结果可为强震区的泥石流防治工程设计提供泥石流冲出规模和淤积范围等关键参数。