雨量输入对分布式流溪河模型中小流域预报精度的影响

中小流域由于资料欠缺、洪水汇流时间短、水量集中等特点，导致其洪水预报难度大，预报精度低。针对我国中小流域的洪水预报精度有待进一步提高的问题，为了让分布式水文模型在全国广泛投入实际应用，选取四川省达州市清溪河流域作为了研究对象。采用流溪河分布式水文模型，从降水量入手，讨论雨量数据对模型结果的影响，探讨满足预报精度所需的雨量站布设密度条件，并将得到的洪水模拟结果与传统预报模式下的洪水模拟结果进行对比分析。结果表明: 12 场洪水中，确定性系数在 80%以上的洪水有 10 场，占比 83. 33%，其中 11 场洪水的洪峰误差小于 20%，合格率为91. 67%，峰现误差合格率达到 100%。因此，流溪河分布式水文模型对于中小流域的洪水预报具有较强的适用性，模拟结果较好。中小流域不同雨量站的降水差异明显，降水总量和雨型的差异对模型输出结果影响较大，当前提高预报精度的重要措施是提高基础数据的质量。清溪河流域雨量站布设密度达到 86 km2 /站时已基本满足该流域流溪河模型预报的要求。现阶段我国许多地区雨量站布设密度已达到较高水平，采用流溪河模型等分布式水文模型进行中小流域洪水预报变为可能。

Influence of rainfall input on flood forecast accuracies from Liuxihe Model for medium and small-sized watersheds

For small and medium river basins，the lack of data，short flood confluence time，and concentrated water volume have made the flood forecasting difficult and the forecast accuracy low． Aiming at the problem that the flood forecast accuracies of the medium and small-sized watersheds in China are necessary to be further enhanced，Qingxihe River Watershed in Dazhou Municipality of Sichuan Province is selected as the study case，so as to widely put the distributed hydrological model into the actual application in China． By means of Xiliuhe Model，the influence from rainfall data on the model is discussed from the aspect of the amount of precipitation，and then the conditions for the density of the layout of rain-gauge station necessary for meeting the requirement of the forecast accuracy are explored，while a comparative analysis is made between the obtained flood simulating result and the flood simulating result got under the conventional forecasting mode． The study result shows that the 10 flood events with the certainty coefficient above 80% are there among 12 flood events with the proportion of 83. 33%，in which the flood peak errors are less than 20% for 11 flood events with the qualified rate of 91. 67% and the qualified rate of the peak occurrence error of 100%． Therefore，Xiliuhe Model has stronger applicability to flood forecast for those medium and small-sized watersheds with better simulating result． The precipitation differences among different rain-gauge stations of the medium and small-sized watersheds are obvious，while the impacts from the total amount of precipitation and the rainfall pattern on the result of the model output are larger，thus the important measure to enhance the forecast accuracy at present is to improve the quality of the basic data concerned． The requirement of the flood forecast for Qingxihe River Watershed with Liuxihe Model is basically satisfied，when the density of the layout of rain-gauge stations within the watershed reaches 86 km2 /station． Currently，the rain-gauge station layout densities in many regions in China reach to a higher level，and then it is possible to adopt the distributed hydrological models such as Liuxihe Model for the flood forecast for the medium and small-sized watersheds．