Based on numerical weather prediction model Weather Research and Forecasting (WRF) and Hydrologic Modeling System (HEC-HMS), a coupling model is constructed in Taihang Piedmont basin. The WRF model parameter scheme combinations composed of microphysics, planetary boundary layers, and cumulus parameterizations suitable for the study area are optimized. In both time and space, we tested the effects of the WRF model by a multi-index evaluation system composed of relative error, root meantime square error, probability of detection, false alarm ratio, and critical success index and established this system in two stages. A multi-attribute decision-making model based on Technique for Order Preference by Similarity to an Ideal Solution and grey correlation degree is proposed to optimize each parameter scheme. Among 18 parameter scheme combinations, Mellor-Yamada-Janjic, Grell-Devinji, Purdue-Lin, Betts-Miller-Janjić, and Single-Moment6 are ideal choices according to the simulation performance in both time and space. Using the unidirectional coupling method, the rolling rainfall forecast results of the WRF model in the 24 h and 48 h forecast periods are input to HEC-HMS hydrological model to simulate three typical floods. The coupling simulation results are better than the traditional forecast method, and it prolongs the flood forecast period of the Taihang Piedmont basin.

基于数值天气预报模型Weather Research and Forecasting (WRF)和水文模拟系统(HEC-HMS)，构建了太行山前盆地耦合模型。优化了适合研究区的微物理、行星边界层、积云参数化组成的WRF模型参数方案组合。在时间和空间上，我们通过相对误差、均方根误差、检测概率、误报率和关键成功指标组成的多指标评价体系来检验WRF模型的效果，并分两个阶段建立了该评价体系。提出一种基于理想解相似排序技术和灰色关联度的多属性决策模型来优化各参数方案。在18种参数方案组合中，从时间和空间上的模拟性能来看，Mellor-Yamada-Janjic、Grell-Devinji、Purdue-Lin、Betts-Miller-Janjić和Single-Moment6是理想的选择。采用单向耦合方法，将WRF模型24 h和48 h预报时段的滚动降雨预报结果输入HEC-HMS水文模型，模拟三种典型洪水。耦合模拟结果优于传统预报方法，延长了太行山前盆地洪水预报周期。