山丘区产汇流机理、模型尺度效应及突发洪水预报研究

Most hilly regions of China are flash flood-prone areas, and thus the research on applicative distributed hydrologic models for flash flood forecasting is a tendency. However, there still exists some bottlenecks currently in both theory and technology levels, e.g., (a) the design of model structure should be improved for fully reflecting the physical mechanism of runoff and meeting the fast computing requirement of flash flood forecasting; (b) the research on the model parameter spatio-temporal scale effect theory is still weak. The calibrated model in data-sufficient basins are hardly transplanted to data-sparse areas. This results in low forecast accuracy of hydrologic models whose parameters are determined only depending on basin topography data; and (c) the Numerical Weather Prediction (NWP) product is almost the only approach to extend the flood forecast leading time due to the short flood concentration time in hilly regions. However, large biases exist in the NWP product and it lacks effective post-processing/correction techniques. Thus, this proposal focusing on the Tunxi Basin (Huangshan, Anhui province), designs the laboratory and field experiments of rainfall-runoff and parameter scale effect simulation. A distributed hydrologic model based on dynamic soil layering mode of moisture movement and evaporation, “three curves” and “cascaded/parallel structure” typed saturation and excess infiltration runoff method on coarse basin grids would be developed. The analytic method of parameter spatio-temporal scale effect will be proposed based on the scale invariance of fractal theory. Moreover, the techniques for numerical precipitation forecast post-processing, flood forecast uncertainty analysis and real-time risk assessment of flood control would also be developed. Research achievements of this study would promote the development of hydrologic forecasting theory.

我国大部分山丘区是洪水突发、易发区，研究适用的分布式水文模型是解决突发洪水预报问题的趋势。但目前在理论与技术层面仍存在瓶颈，如：①模型结构设计需要突破，既能充分反映产汇流物理机制，又能满足突发洪水快速预报需求；②水文模型尺度效应理论研究薄弱。资料充分地区验证的模型难以移置于资料短缺区，致使完全依靠地形地貌信息确定的模型（参数），其精度并不高；③山丘区汇流时间短，利用数值天气预报延长洪水预见期几乎是唯一途径，但数值天气预报亦有较大误差，尚缺乏有效的后处理（校正）技术。本申请以安徽黄山屯溪以上流域为典型区域，设计室内外降雨径流实验，构建以土壤动态分层的水分运动及蒸发模式、网格“三条曲线”与“水箱串并联结构”的蓄满-超渗混合产流机制为构架的分布式水文模型，提出分形“标度不变性”时空耦合的参数尺度效应解析方法，发展数值降水预报后处理、预报不确定性分析及防洪风险实时评估方法。促进水文预报理论发展。

我国大部分山丘区是洪水突发、易发区，研究适用的分布式水文模型及相关技术是解决突发洪水预报问题的关键。但目前在理论与方法层面仍存在瓶颈，如：①模型结构设计仍需要突破，既能充分反映产汇流物理机制，又能满足突发洪水快速预报需求；②水文模型尺度效应理论研究薄弱，资料充分地区验证的模型参数难以移置于资料短缺区；③山丘区汇流时间短，利用数值天气预报延长洪水预见期几乎是唯一途径，但数值天气预报亦有较大误差。针对如此等诸多挑战，本项目通过室内外降雨径流实验研究，揭示山丘区水文连通性及蓄满-超渗产流统一转换机制，建立了山丘区突发洪水预报的分布式水文模型及相关技术方法体系。主要成果包括：.（1）创建了变动产流层分布式水文模型（Variable Runoff Generation Layer VRGL）。揭示了山丘区土壤分层产流规律，提出变动产流层概念，并建立了一种可同时描述蓄满-超渗产流机制的山丘区分布式水文模型。（2）解析了模型参数尺度效应。基于分形理论中的“标度不变性”，提出一种模型参数空间尺度效应解析方法，建立了模型参数在不同空间尺度之间的转换关系。（3）研发了面向洪水预报的数值降水预报后处理技术。构建数值降水预报产品订正、集合预报时空结构重构及距离-高程的块-协克里金雨量空间展布等技术，提高了模型降雨输入数据的精度。（4）提出了洪水预报实时校正新技术。构建以雨量站网密度为链接因子的“面雨量输入误差-模型参数误差-预报终端误差”互馈解析模型，建立了可准确溯源、划分并修正不同来源误差的全过程联合校正新技术。（5）建立了洪水预报不确定性分析及防洪风险评估技术方法。开展不确定性分析及洪水概率预报的深入研究，构建了水库和河道断面防洪风险计算模型，形成以“不确定性分析-洪水概率预报-风险计算”为核心的防洪风险实时评估技术方法。.项目成果可为进一步揭示山丘区产汇流规律、提高突发洪水实时预报精度和增长预见期、降低防洪决策风险，提供科学依据和实际工作指导。