热带气旋降水的参考态

浙江省热带气旋过程降水预报的一个基本实践是:事发之后总能找到一个与当前态最相似的历史样本,它的过程雨量及分布特征与当前态未来实况相当接近,基本上可满足预报服务要求,然而要事先找出这个最相似的历史样本,却是十分困难的.文献1]介绍的,在全体样本空间里用算子DE搜索与当前态最相似历史样本的方法,效果仍不理想,本文从历史样本出现的概率出发,构造若干热带气旋降水的参考态,业务使用时,先将当前态用参考态作类型判别,然后在该类型中进行最相似历史样本的筛选。本文给出了对1990年5次热带气旋所作判别和筛选的结果,并作了讨论。

ON REFERENCE STATES OF PRECIPITATION DUE TO TROPICAL CYCLONES

A basic fact prevailing in the forecasting practice in Zhejiang Province of process rainfall due to tropical cyclones is that there is always a historic case that is most similar to the current state in the post event check. Its process rainfall and distribution features are so close to the reality that comes after the current state that they generally meet the requirements of forecasting services. It is, however, very difficult to locate it. A method^1] of using an operator DE within the whole space of samples in search of a historic case to best match the current state proves to be unsatisfactory. In this paper,starting from occurrence probabilities of historic samples, a number of reference states are set up for precipitation due to tropical cyclones. In operational utilization, a pattern dete rmination is done by comparing the current state with the reference state, followed by a screening of the pattern isolated for a historic case that bears the most similarity. Results of such determination and screening are given for five of the tropical cyclones occurring in 1990 and discussions presented.