耦合分布式水文模型的降水诱发型滑坡集合预报关键技术研究

Rainfall-induced landslides is a frequent and heavily damaging natural geological hazard, and landslides forecasting is one of the most effective non-engineering measures for landslides disaster control. Compared with statistical model, physical dynamic model, which can take into consideration the physical understanding of how landslides respond to rainfall infiltration and runoff yield and concentration hydrological processes, and simulate precipitation-induced slope-failure processes, has become a new direction for landslides forecasting. However, hydrological processes simulation Incompleteness, precipitation forecasting uncertainty, scale match in atmospheric-hydrological-landslides model and its forecasting uncertainty, become a bottleneck to improve the precision and lead time of Rainfall-induced landslides forecasting. Based on the WRF mesoscale NWP model, slide and TRIGRS models, three key technologies of Rainfall-induced landslides forecasting will be studied. (1) A distributed conceptual hydrological model based on the mixed runoff generation model for landslides forecasting will be studied, and the model parameterization scheme is based on prior estimation method. A rainfall-induced landslides forecasting model by coupling with a distributed hydrological model will be developed.(2) With analyzing WRF model's uncertainty, WRF ensemble prediction model with high resolution will be developed.(3) Ensemble rainfall-induced landslides forecasting model by coupling with a distributed hydrological model, which driven by WRF ensemble prediction model , will be evaluated and its uncertainty will be analyzed in the rainfall-induced landslides prone areas. This project aims to improve the precision and lead time of rainfall-induced landslides forecasting, and provide scientific basis for flood control decision-making.

降水诱发型滑坡是我国常见与危害性很大的地质灾害类型，滑坡预报是滑坡灾害防治最有效的非工程措施之一。相比滑坡统计模型，动力物理模型能够考虑降水入渗与产汇流等流域水文过程对滑坡的响应，实现边坡失稳过程模拟，成为滑坡预报新的发展方向。然而，水文过程模拟不完整性、降水预报不确定性、气象-水文-滑坡尺度匹配及预报不确定性，成为制约提高滑坡预报精度与预见期的主要瓶颈。本项目基于WRF中尺度模式、SLIDE和TRIGRS模型，对滑坡预报三个关键方面进行研究：（1）研究面向滑坡预报的分布式混合产流水文模型及参数先验估计技术，建立耦合分布式水文模型的滑坡物理模型；（2）研究WRF模式不确定性，建立高分辨率WRF集合预报模式；（3）建立WRF集合预报驱动，耦合分布式混合产流水文模型的滑坡集合预报模型，在滑坡灾害易发区域滑坡预报中评估预报性能与不确定性。项目旨在提高滑坡预报精度与预见期，为防灾减灾提供科学依据。

降水诱发型滑坡是我国常见与危害性很大的地质灾害类型，滑坡预报是滑坡灾害防治最有效的非工程措施之一。相比滑坡统计模型，动力物理模型能够考虑降水入渗与产汇流等流域水文过程对滑坡的响应，实现边坡失稳过程模拟，成为滑坡预报新的发展方向。然而，水文过程模拟不完整性、降水预报不确定性、气象-水文-滑坡尺度匹配及预报不确定性，成为制约提高滑坡预报精度与预见期的主要瓶颈。本项目基于中尺度数值天气模式、分布式水文模型和滑坡物理模型，对滑坡预报三个关键方面进行研究：.（1）研发了面向滑坡预报的分布式混合产流水文模型及参数先验估计技术，发展了基于Muskingum-Cunge 法的河道水位流量预报模型，构建了GMKHM分布式混合产流水文模型，建立了耦合分布式混合产流水文模型和SLIDE模型的滑坡物理模型，发展了基于参数先验估计的模型参数化方案与参数敏感性分析方案，形成了全国地质灾害气象风险精细化网格预报技术，在业务中应用。.（2）提出了基于中尺度模式的精细化降水集合预报技术。提出了引发区域地质灾害的降水分型及阈值分析，发展了基于集合卡尔曼滤波的中尺度模式陆面同化技术，构建了基于中尺度数值天气预报的短时临近降水集合预报模型和精细化网格降水尺度转化与预报技术，形成了0-240h无缝隙快速滚动更新定量降水预报。.（3）构建了降水-分布式混合产流水文模型-滑坡物理模型耦合的滑坡集合预报模型，发展了基于Kalman滤波的模式实时校正技术，在台风诱发区、西部地质环境脆弱区、北京等滑坡灾害易发区域滑坡预报中评估，取得良好的预报精度。部分成果已经在业务中应用，为防灾减灾提供科学依据。