本文已被:浏览 1617次 下载 2594次
投稿时间:2009-06-29 修订日期:2009-12-01
投稿时间:2009-06-29 修订日期:2009-12-01
中文摘要: 在贝叶斯概率决策理论的基础上,探索了一种提高四川暴雨预报准确率的方法,该方法利用四川境内1951—2004年147站暴雨的气候概率对西南区域中尺度集合预报模式提供的≥50 mm集合降水概率预报产品进行了修正。从2008年6—9月的连续性预报试验结果来看:基于贝叶斯方法修正后的集合概率预报产品在一定程度上消除了空报,尽管相比于区域中尺度集合预报系统直接提供的暴雨概率预报产品实际提高暴雨准确率的效果不明显,但在如何有效地利用数值集合预报产品提高四川暴雨预报的准确率以及如何为预报员提供更有价值的暴雨预警决策方法上值得进一步探索。
中文关键词: 贝叶斯方法, 四川暴雨, 集合概率预报, 产品释用
Abstract:A method of improving the Sichuan heavy rainfall forecasting accuracy based on the Bayesian decision theory is explored. This method uses the heavy rainfall climate probability of Sichuan 147 stations from June to September in 1951 to 2004 to modify the ensemble prediction probability products of precipitation more than 50 mm produced by southwest regional ensemble forecasting system. The continuous forecasting test results from June to September in 2008 show that the posterior probability, i.e. the probability modified by the Bayesian mathod, can eliminate the false prediction to some extent. Although the effect of improving the heavy rainfall forecasting accuracy is not obvious compared with the prior probability, it still provides an exploring method-how to use the numerical ensemble prediction products to improve the Sichuan heavy rainfall forecasting accuracy and how to provide more valuable heavy rainfall information for the early warning decision method.
keywords: Bayesian method, Sichuan heavy rainfall, ensemble prediction probability, product explanation
文章编号: 中图分类号: 文献标志码:
基金项目:本文得到中国气象局公益性行业专项(气象)科研专项资助项目“面向TIGGE的集合预报关键应用技术研究”和国家自然科学基金项目No.40675061的资助
作者 | 单位 |
陈朝平 | 四川省气象台, 成都 610072 |
冯汉中 | 四川省气象台, 成都 610072 |
陈静 | 国家气象中心, 北京 100081 |
引用文本:
陈朝平,冯汉中,陈静,2010.基于贝叶斯方法的四川暴雨集合概率预报产品释用[J].气象,36(5):32-39.
CHEN Chaoping,FENG Hanzhong,CHEN Jing,2010.Application of Sichuan Heavy Rainfall Ensemble Prediction Probability Products Based on Bayesian Method[J].Meteor Mon,36(5):32-39.
陈朝平,冯汉中,陈静,2010.基于贝叶斯方法的四川暴雨集合概率预报产品释用[J].气象,36(5):32-39.
CHEN Chaoping,FENG Hanzhong,CHEN Jing,2010.Application of Sichuan Heavy Rainfall Ensemble Prediction Probability Products Based on Bayesian Method[J].Meteor Mon,36(5):32-39.