基于LAF思想的一种初始扰动生成方法及集合预报试验

A Method of Creating Initial Perturbation Based on LAF Method and Experiments of Ensemble Forecast

在时间滞后平均法（LAF）思想的基础上，提出了一种改进的LAF方法，这种方法的思路是将不同的初始扰动叠加到模式初始场上进行不同的扰动预报，形成集合预报的成员；而不同的扰动场是通过在滞后不同时次上求取不同资料分析场之差获得。这种扰动不仅体现了不同分析场的差异，而且不同的滞后时间也体现了LAF方法的思路。然后利用经典的LAF方法及改进的LAF方法和AREM模式，分别对2004年6月14～15日安徽入梅首场暴雨过程进行了集合预报试验。结果表明，无论是雨带还是雨团位置和强度的预报，改进的LAF方法都优于经典的LAF方法。进一步研究发现，对于同一个扰动，加扰预报结果比减扰结果要好，若只选择加扰预报作集合平均，其预报结果更接近实况。

An improved LAF method is brought forward based on LAF idea. The idea of this new method is adding different initial perturbations to model initial field for different perturbation forecastings, which generates members of ensemble forecasting. The perturbation field is derived from difference between different analysis fields on lagged various time. Thus, not only the perturbation reflects the difference between different analysis fields but also lagged various time embodies idea of LAF method. Then, ensemble forecasting experiments of the first Meiyu rainstorm in Anhui province from 14 to 15 June 2004 are made respectively with LAF method and advanced LAF method with AREM model used. The results show that advanced LAF method is superior to classical LAF about forecasting rainband position and intensity of the rainfall. After further analyzing, it is found that to the same perturbation the forecasting effect of adding perturbation forecast is better than that of subtracting perturbation forecast. Even the ensemble forecast gets much closer to the observation while only the results of adding perturbation forecast are averaged.