多普勒雷达估算降水和反演风在不同初值方案下对降水预报影响的数值研究  

NUMERICAL RESEARCH ON EFFECTS UPON PRECIPITATION FORECAST OF DOPPLER-RADAR ESTIMATED PRECIPITATION AND RETRIEVED WIND FIELD UNDER DIFFERENT MODEL INITIAL SCHEMES

作　　者：王叶红[1] 赵玉春[2] 崔春光[1] 

机构地区：[1]中国气象局武汉暴雨研究所,武汉430074 [2]武汉中心气象台,武汉430074

出　　处：《气象学报》2006年第4期485-499,共15页Acta Meteorologica Sinica

基　　金：湖北省自然科学基金项目(2003ABA009);"十五"国家重点科技攻关计划项目(2001BA607B);国家973项目(2004CB418307)

摘　　要：基于有限区域中尺度暴雨预报模式AREM,利用常德、荆州、宜昌3部多普勒雷达及武汉数字化天气雷达联合反演的1h降水量资料和宜昌、荆州多普勒雷达反演的水平风场资料,采用3种不同的初值方案:Grapes-3DVAR同化方案、Barnes客观分析法及Barnes-3DVAR同化方案,对2002年7月22日20:00—23日20:00发生在长江流域的一次大暴雨过程进行了多普勒雷达估算降水和反演风场在不同初值方案下对降水预报影响的数值试验,结果表明:(1)采用相同资料源(探空+T213L31分析场),3种初值方案降水模拟效果差别显著。整体而言,融合了Barnes和3DVAR两种方案的Barnes-3DVAR初值方案降水模拟效果最好:雨带模拟完整,位置与实况完全一致;降水中心中尺度结构清晰,位置、强度与实况接近。(2)Barnes-3DVAR同化方案与Grapes-3DVAR同化方案都能同化雷达降水和风场资料,但两者降水预报结果相差很大。Barnes-3DVAR方案下的降水模拟结果明显好于Grapes-3DVAR同化方案。(3)Grapes-3DVAR同化方案下,在控制试验的基础上增加雷达降水资料的同化,24h降水模拟效果有所改善;增加雷达风场资料的同化,仅部分改善模拟效果,某些地方效果反而变差;同时增加雷达降水和风场资料的同化,降水模拟效果反而明显变差。Barnes-3DVAR同化方案则不然,无论雷达降水资料和风场资料单独同化还是两种资料同时同化,降水模拟效果都有不同程度的改进,特别是对雷达降水和风场资料同时同化的模拟结果是各种试验中效果最好的。(4)融合Barnes客观分析方法和Grapes-3DVAR同化方案的Barnes-3DVAR同化方案,是同化雷达降水和风场资料的一种新的、有效的初值方案。On the basis of the joint estimated 1 h precipitations from Changde, Jingzhou and Yichang Doppler radar and Wuhan digital radar, and the retrieved wind fields from Yichang and Jingzhou Doppler radar, a series of numerical experiments of the advanced regional η-coordinate model （AREM） under different model initial schemes, i.e. Grapes-3DVAR, Barnes and Barnes-3DVAR, are carried out for a torrential rain process occurring along the Yangtze river in the 24 hour period started from 20：00 BST 22 July 2002 to investigate the effects of the Doppler-radar estimated rainfall and retrieved winds on the rainfall forecast. The main results are as follows： （1） the simulations are obviously different under three initial schemes with the same data source （the radiosoundings and T213L31 analysis）. On the whole, Barnes-3DVAR, which combines the advantages of the Barnes objective analysis, and the Grapes-3DVAR method, gives the best simulations： well-simulated rain band and clear meso-scale structures, as well as their location and intensity close to observations. （2） Both Barnes-3DVAR and Grapes-3DVAR schemes are both able to assimilate the Doppler-radar estimated rainfall and retrieved winds, but differences in simulation results are very large, with Barnes-3DVAR＇s simulation much better than Grapes-3DVAR＇s. （3） Under Grapes-3DVAR scheme, the simulation of the 24 hour rainfall is im- proved a lot obviously when assimilating the Doppler-radar estimated precipitation into the model in comparison with the control experiment; but it becomes a little worse, when assimilating the Doppler-radar retrieved winds into the model, and it becomes worse obviously when assimilating the Doppler-radar estimated precipitation as well as retrieved winds into the model. However, the situation is different under Barnes.3DVAR scheme. The simulation is improved to a certain degree no matter assimilating the estimated precipitation or retrieved winds, respectively, or both of them. The result is the best when assimilating

关 键 词：雷达 估算降水量 反演风场 同化 降水预报 

分 类 号：P457.6[天文地球—大气科学及气象学]