河南“75.8”大暴雨的中尺度集合预报试验  

Mesoscale Ensemble Forecasts on "75.8" Heavy Rain in Henan

作　　者：谭燕 陈德辉[2] 

机构地区：[1]中国气象局上海台风研究所,200030 [2]国家气象中心

出　　处：《气象》2008年第9期10-21,共12页Meteorological Monthly

基　　金：国家自然科学基金课题40575050;"973"项目课题2004CB418306的资助

摘　　要：发生在河南的“75．8”特大暴雨是我国历史上影响最大、受灾程度最严重的一次灾害性天气过程，其造成的人员伤亡与唐山地震相当，造成的经济损失近百亿元，我国气象学家对此展开过大量的研究，其可预报性问题一直都受关注。尝试采用近年来在预报理论与实践中提出的中尺度集合预报技术，以GRAPES中尺度有限区模式作为试验模式工具，针对此次过程展开两组试验。试验1主要着眼于对暴雨有较大影响的积云对流参数化方案，分别从对流激发和大尺度环境场相互作用的角度出发，对其中一些经验性参数在合理取值范围内给予调整，使其在一定程度上表征模式的不确定性，来构建集合成员。试验结果表明：对流参数化方案中不同的参数对降水的影响作用各不相同，空间上对于降水落区的影响不是很大，但是对降水强度的改善却很明显，集合平均对强暴雨中心的体现有积极作用，平均后的暴雨区最大降水量预报为90mm／24h，比“确定性”预报（控制试验）值70mm／24h改善了约30％，最佳的单个成员的暴雨区最大降水量预报值（120mm／24h）比控制试验预报（70mm／24h）提高了70％；时间上看，不同的参数会影响到积云对流的激发，使得降水发生的时间有所不同，进而影响后续降水的发展。试验2利用模式不同分辨率和不同物理参数化方案的组合来构造集合成员，进行中尺度集合预报试验，试验结果表明：简单的集合平均能在一定程度上改善降水强度，与高分辨率模式预报结果比较，暴雨中心降水量约提高了20％～30％。可见：集合预报能在一定程度上减弱模式不确定性的影响；对流参数化方案和经验性参数的差异、模式分辨率的差异确实对模拟结果有一定的影响。＂75.8＂ heavy rain is a disaster, which caused a large loss to the society. Many meteo- rologists have done a lot of research works of this event, and paid much attention to the predictability. ＂75.8＂heavy rain in Henan Province was particularly chosen for the study of the meso scale ensemble forecast. Two experiments were implemented. The first one especially focused on the relations between cumulus convective parameterizations and the heavy rain. For this purpose, the Kain-Fristch ETA scheme was chosen. The interaction between the trigger function and the mass flux calculations in large scale environment was taken into consideration. Some empirical factors have been perturbed in a reasonable range to construct the ensemble members to describe the uncertainty of model. The results indicate that different factors in cumulus convective parame- terization have different impacts on the precipitation. The location of the rainfall is less influenced by different factors, but the intensity of the heavy rain is improved, the precipitation forecasted by ensemble mean increased by 30 %. Meanwhile, different factors would have different impacts on the time when the convection happens, the rainy duration as well as the development of con- vection. The second experiment adopted different resolution of model and multl-pararneterization to form ensemble members. The results prove that the ensemble mean would improve the intensi- ty forecast to some extent. Compared with the results coming from model with high resolution, the precipitation forecasted by ensemble increases by 20 % - 30 %. Ensemble forecast would decrease the uncertainties from the model itself. The differences among various cumulus convective parameterizations, empirical factors and model resolutions indeed have a certain impacts on the simulated results of model.

关 键 词：暴雨 中尺度集合预报 不确定性 对流参数化 模式分辨率 

分 类 号：P457.6[天文地球—大气科学及气象学] P457