初始场对暴雨数值预报的影响及集合预报试验

The effect of initial condition on numerical precipitation prediction and ensemble forecast

以2003年7月3日至4日发生在淮河流域的暴雨过程为例,利用AREM模式,分析了初始场对暴雨预报的影响,提出了暴雨预报中初始场不确定性包含的两层含义,一是被常规观测遗漏的中小尺度信息误差;另一个则是随着环流变化造成的信息误差的不确定性。并针对着初始场的不确定性,设计了一种初值集合预报的方法,它包含了经典的集合预报方法MCF、LAF、BGM的思想。用这种方法进行了集合预报试验,结果表明:集合平均预报的预报技巧高于24 h控制预报,集合预报还可给出降水概率预报、离散度等产品为暴雨可预报性的评估提供参考。

A particular precipitation event on 3-4 July.2003 is simulated by AREM.Through the analysis of the effect of initial condition on heavy rainfall prediction,it is found that the initial uncertainties include two aspects which can be regarded as the mesoscale feature error due to the coarse observation resolution and the variation of such error because of the evolvement of basic flow.Based on the initial uncertainties,a scheme which reflects the idea of MCF,LAF and BGM is proposed for the ensemble forecast of heavy rainfall.The results indicate that the ensemble average is more skillful than the control simulation.Besides,the ensemble prediction can be used in evaluation of the precipitation predictability such as the probability prediction and dispersion.