初值形成方法对降水数值预报的影响

北欧有限区域模式ＨＩＲＬＡＭ被应用于中国的暴雨个例以探讨初值形成方法对有限区域模式定量降水数值预报的影响，对两种初值形成方案进行了对比，一种是由ＨＩＲＬＡＭ自己的数值同化系统提供初值，另一种是直接内插ＥＣＭＷＦ全球模式的相应分析场，与这两种方案对应的数值试验分别是控制试验（ＣＯＮＬ）和对比试验（ＣＯＭＰ），将ＣＯＮＬ和ＣＯＭＰ的降水预报与观测值比较，结果表明：（１）当为ＣＯＭＰ提供初值的ＥＣＭＷＦ

THE EFFECT OF METHOD FOR CONSTRUCTING INITIAL FIELD ON NUMERICAL FORECAST OF RAINFALL

The HIRLAM model (a limited area numerical weather forecasting model developed by the Nordic countries) was applied to the summer heavy rain cases in China to study the effect of the methods for constructing initial fields on numerical rainfall forecasts of a limited area model. Two methods for constructing the initial fields are compared: one initial fields is provided by the HIRLAM's own data assimilation system, another is given by interpolating the ECMWF global model's analysis fields into the HIRLAM grids. The two numerical experiments corresponding to these two methods for constructing the initial fields are the control experiment(CONL) and the comparison experiment (COMP), respectively. The rainfall predicted by CONL and COMP are compared with observation, the results show: (1) when the resolution of ECMWF global model, whose analysis is used to provide initial fields for HIRLAM model in COMP run, is considerably lower than that of HIRLAM model, the rainfall forecast is markedly improved in CONL in which the initial fields are provided by HIRLAM's own data assimilation system, in comparison with that in COMP. (2) when ECMWF model and HIRLAM model approximately equal resolution, the qualities of rainfall forecasts in CONL and COMP turn out to be at nearly the same level in spite of the supposed better consistency between initial fields, model dynamics and model's topography in CONL.