以2010年8月17—19日华北地区的暴雨事件为研究对象,采用中尺度数值模式WRF,通过8个数值实验测试暴雨事件模拟结果对不同边界层方案的敏感性。结果表明,采用不同边界层方案的模拟结果存在明显差异。分辨率为12 km时,7个边界层方案实验均模拟出类似观测的四处较强降水,但模拟得到的降水强度和强降水中心位置与实况有所差异,NOPBL实验模拟的雨带收缩,降水相对其他模拟实验较少且更为集中。分辨率为4 km时,采用边界层方案的7个实验均可见小尺度降水结构,模拟出较多虚假降水中心,而在NOPBL试验中降水的小尺度结构不明显。检验表明:分辨率为12 km时,MYJ试验的TS评分、相关系数和误差分析等整体表现最优,分辨率为4 km时,Bou Lac试验整体表现最优。与NOPBL试验相比,加入MYJ边界层方案后,模拟的水汽输送增加、上升运动及涡散度绝对值增大。
Based on a rainstorm event in North China during 17-19 August,2010,by using WRF model,eight numerical experiments were designed to test the sensitivity of different boundary layer schemes on rainstorm simulation. Results show that the simulated results with different schemes are different. When the domain resolution is 12 km,the seven boundary layer scheme experiments all simulated the four observed strong precipitation locations,but the exact location and rainfall intensity were different from the observation. For NOPBL experiment,the rain belt was narrower,and its precipitation was more concentrated. When the resolution is 4 km,the seven boundary layer scheme experiments all simulated small-scale precipitation structures,however,they were not obvious in NOPBL experiment. MYJ experiment performs best with 12 km resolution,but when the resolution is 4 km,Bou Lac experiment is the best. Furthermore,the results indicate that the increasing water vapor flux convergence,the strengthening upward motion,and the enlargement of absolute value of divergence and vorticity are all depends on the selected boundary layer scheme.