短期集合预报中定量降水预报集合方法初探

A Quantitative Precipitation Forecasts Method for Short-range Ensemble Forecasting

摘要:

针对短期集合预报中集合平均等方法对极端天气事件预报能力低下的缺陷,提出一种定量降水预报集合方法,简称两步法。该方法首先采用逐级归并法确定集合成员的总体相似度,进而确定集合成员预报场的延伸半径,对各成员的预报场进行修正,然后根据每个成员与其他成员的相似程度,确定各成员的权重系数。这个方法既考虑了集合成员总体的差异性又考虑了降水的概率分布。试验证明,通过上述两步得到的定量降水预报,对提高极端天气事件的预报确率有一定效果。本方法无需历史资料,能很好适应集合成员的改变,实现方便快捷。

关键词:短期集合预报;定量降水预报;集合方法;相似聚类

Abstract:

A method of quantitative precipitation forecasts for short-range ensemble forecast is presented aiming at inability of ensemble average method to extreme weather event.This method firstly determines ensemble similitude by stepwise aggregation to define the influencing radius of members forecast fields,then determines the weight coefficient of every member according to the similitude degree of one member to other members.It takes into account both the ensemble otherness of members and probability distributing of precipitation.Examination proves that this method improved the forecasting of extreme weather events.This method needn't history data and can accommodate to the change of members so it is prone to operate.

Key words:short-range ensemble forecast;quantitative precipitation forecasts;ensemble method;similitude clustering