基于模式先验信息的贝叶斯集合降水概率预报试验  

Bayesian Ensemble Probabilistic Forecasting Model Experiment of Precipitation Based on Model Priori Information

作　　者：张宇彤[1,2] 矫梅燕[3] 陈静[4] 

机构地区：[1]南京信息工程大学,南京210044 [2]大连市气象局,大连116001 [3]中国气象局,北京100081 [4]中国气象局数值预报中心,北京100081

出　　处：《气象》2013年第10期1233-1246,共14页Meteorological Monthly

基　　金：国家自然科学基金面上项目(41075035);公益性行业(气象)科研专项(GYHY200906007);国家科技支撑计划项目(2009BAC51B00);国家重点基础研究发展计划项目(2012CB417204)共同资助

摘　　要：为了更好地利用降水预报历史先验概率分布函数信息修订集合概率预报效果，基于贝叶斯原理和贝叶斯降水概率预报模型，分别使用1952—2007年历史观测资料和2009—2011年6月24-120 h中国T213全球集合预报历史资料作为先验信息，对中国不同气候区代表站（广州、南京、武汉和成都）建立贝叶斯降水概率预报模型，对比不同先验信息下集合成员与集成贝叶斯降水概率预报拟合结果差异，分析先验信息对贝叶斯降水概率预报模型的影响，在此基础上，采用模式先验信息的贝叶斯降水概率预报模型，进行2008年6月降水概率预报试验。试验结果表明，由T213集合预报产生的先验信息较历史观测资料产生的先验信息更优，当先验信息的降水概率分布函数曲率最大处偏向降水大值区时，贝叶斯模型的降水预报结果也偏向降水大值区，反之亦然。结果还显示，先验信息对贝叶斯降水概率预报模型有重要影响，若先验信息偏向更多更大降水量时，贝叶斯降水概率预报对有降水的预报更优，若降水先验信息偏向更少更小降水量时，对无雨或微量降水预报效果越好。In the rainfall probabilistic forecasting, information from the end of the probability distribution function should be applied. Then better forecast about rain and the ability of probabilistic forecasting will be developed. Based on the Bayesian theory and probabilistic forecasting model of precipitation, observational and T213 ensemble prediction data are used as the different priori information sources and experiments about precipitation in Guangzhou, Nanjing, Wuhan and Chengdu are carried out. Differences on the forecasts that are dependent on different priori informations are compared. Also influence of priori information on the Bayesian precipitation probabilistic forecast model is analyzed. And then the model is developed as well. Furthermore, experiment based on Bayesian precipitation probabilistic forecasting which is dependent on model priori information is conducted. Results show that the priori information got from the ensemble prediction data is better. While the priori precipitation amount is more, the precipitation from the forecasting of model is also more and vice versa. Meanwhile, priori information has a strong impact on the Bayesian precipitation probabilistic forecast model. If the precipitation amount is much more, Bayesian ensemble probabilistic forecasting model can produce much more accurate prediction of rainy days. If the precipitation amount is little, prediction of no rain or light rain from the model is better.

关 键 词：贝叶斯方法 先验信息 模式气候 降水概率预报 

分 类 号：P457[天文地球—大气科学及气象学]