以北京副中心北运河生态带、城北、河西和两河片区为研究区，选取典型强降水过程，基于新一代中尺度天气研究和预报模式（weather research and forecasting model,WRF），通过对物理参数化方案的优选，构建适用于北京副中心的数值天气模型。通过区域大气模式可以实现定量降水模拟与预报，为缺乏降水资料地区的相关研究提供数据支撑。研究结果表明：不同参数化方案的模拟结果差异较大，积云参数化方案对研究区强降水模拟效果的影响最大；当云微物理过程取WRF Single-Moment 5-class方案，积云对流过程取Grell-Freitas方案，行星边界层过程取Yonsei University方案，长、短波辐射过程取newer version of the Rapid Radiative Transfer Model方案，表层取Revised MM5 Monin-Obukhov方案，陆地表面取Noah land surface model方案，城市表面取Urban canopy model方案时，模拟结果最优。

At present,urban storms and floods have already been one of the major disasters that threaten the sustainable development of cities.With the rapid development of Beijing′s sub-center,storms,and floods vulnerability has continued to increase.Historical or real-time observed data,interpreted data from weather radar or satellite remote sensing,and simulated data from numerical atmospheric models are currently used to study precipitation characteristics.The regional atmospheric model,which could realize the quantitative precipitation simulation and forecasting and provide data support for related research in ungauged areas,was of great significance for urban flood forecast and control.The new generation of mesoscale weather research and forecast model named Weather Research and Forecasting (WRF)model can provide more complete,efficient,and accurate weather simulation and forecast.The improvement of regional-scale precipitation simulation and forecasting capabilities has made the model gradually applied to smaller-scale storm simulations and forecasts in various countries and regions.In recent years,Beijing′s sub-center has realized the stable and healthy economic and social development of the whole region with the rapid increase of building area and population density,increasing the vulnerability to the effects of storms and floods. Based on WRF,the numerical weather simulation model was built by considering a more comprehensive optimization of the physical parameterization schemes and applied to Beijing′s sub-center.According to the divisions of the Tongzhou District Government concerning water environment management work,the study area included the Chengbei area,Hexi area,Lianghe area,and North canal ecological belt.Three-level two-way nested domains were set according to Lambert Conformal coordinates.Considering cloud microphysical schemes,cumulus convection schemes,planetary boundary layer schemes,and land surface schemes,16 different parameterization scheme combinations were set.To select the optimal combinations,relative error (ER),rootmeansquare error (ERMS),mean deviation error (EMB) and standard deviation (DS) were used to evaluate the simulation effects on temporal and spatial scales. The results showed that different schemes and their combinations had different results,and the simulation results for ′7·20′ were better than those for other rainfalls.The cumulus convective parameterization scheme had the most significant impact on the simulation of heavy rainfall in the study area,and generally,model performance was better when the cumulus parameterization scheme was Grell-Freitas.The combination of the WRF Single-Moment 5-class scheme,the Grell-Freitas scheme,the Yonsei University scheme,the newer version of the Rapid Radiative Transfer Model scheme,the Revised MM5 Monin-Obukhov scheme,the Noah land surface model and the urban canopy model had the best result. In summary,it is feasible to simulate heavy rainfall by the WRF model.The method could extend the forecast period and provide data support for related research in areas lacking precipitation data.Under the current background of increasing short-duration heavy rainfall,the application of above method to urban flood control had important practical significance.Although the simulation accuracy of the WRF model could meet the requirements of actual operations,there were still some shortcomings that required further excavation and analysis.With the solving of the limitations of observed data,the model performance could be improved further in the future.