基于模糊聚类分型的数值产品暴雨预报释用方法

A heavy rainfall forecast method based on fuzzy cluster typing by using application and interpretation of NWP

本文阐述了一种基于数值产品的预报信息场的暴雨预报释用方法,提出了对暴雨历史样本形势场使用模糊聚类的客观分型技术,用条件概率让小概率事件成为相对的大概率,更有利于暴雨指标的提炼;该释用方法包含的信息量全面,以历史暴雨样本同期的包含大气三维空间压、温、湿、风及其释演的物理量,构建出25个数值产品参数指标作为鉴别暴雨的基础,同时也勾划出各暴雨型发生期大气高温、高湿、对流不稳定等机理结构特征,且均有具体物理量数据表述。预报方法的研究过程中探索了某些物理量场预报暴雨落区的规律;建立了暴雨预报工具,充分利用了现有业务中的质量较好的预报产品设立预报工具的起报条件,减少了暴雨空报率。

The paper investigated a heavy rainfall forecast method of application and interpretation NWF in accordance with its forecasting fields. An objective fuzzy cluster typing technique for heavy rainfall history data was proposed, in which a small probability turned into large by using condition probability in favor of trying out the heavy rainfall indexes. The method summed up 25 parameter indexes to diagnose heavy rainfall according to historical samples for same period which include pressure, temperature, humidity, wind and interrelated physical quantities in the 3D space of atmosphere, and developed the mechanism and the atmosphere structure feature such as high temperature, high humidity and convective instability described by idiographic physical factors. In addition, the study explored some rules of fore-casting heavy rainfall area by means of specifically physical quantities. A rainstorm forecast tool was set up, of which the result of NWP operation in existence can be regarded as the start-up condition for reducing rainstorm false rate of forecast.