复杂地形下多普勒雷达资料同化的研究

Study on Doppler Radar Data Assimilation over Complex Terrain

针对2010年7月16日四川地区一次强降雨过程,以1°×1°的NCEP再分析资料为背景场,四川923个国家级自动气象站逐时降雨资料为观测场,地形高度采用数字地形高程模型(DEM)的地形高度数据替代了模式地形数据,应用WRF模式及其三维变分同化系统,对成都、绵阳、达州、西昌、南充5部雷达反射率因子进行了直接同化试验。结果表明:同化反射率因子之后,各层嵌套初始场的水汽通量分布改变较小,但数值和范围改变明显,第一重嵌套水汽通量高值中心显著增强,800 hPa改变较500 hPa明显;同化试验、控制试验均能模拟出此次降雨过程;同化试验在降雨量值、面积、落区上较控制试验均有所改进,对四川盆地东北部的达州和巴中两市改进尤其明显;12 h降雨预报改进优于24h降雨预报。

Taking the heavy rainfall process occurred in Sichuan province on 16 July 2010 as an example,a direct assimilation experiment was conducted using WRF model and the 3DVAR assimilation system with data of five radars located in Chengdu,Mianyang,Dazhou,Xichang and Nanchong.The background was from NCEP1 °×1 ° reanalysis and hourly precipitation observation of 923 national automatic observation stations in Sichuan.A replacement of DEM terrain data to the WRF model terrain data was conducted.Results show that:there is an apparent improvement in the value and area of the water vapor flux at initial time on each of the nested domain as the reflectivity data was assimilated.The maximum of the water vapor flux in the first domain is significantly strengthened and more significant on 800 hPa than that on 500 hPa.Both of the assimilation and control experiments can simulate the process.The assimilation experiment has more advantages on quantity and area forecast of rainfall,especially in Dazhou and Bazhong.12 hours precipitation forecast is much better than that of 24 hours.