气候集成预报的初步探讨
投稿时间:1998-04-20  修订日期:1998-06-26  点此下载全文
引用本文:
摘要点击次数: 553
全文下载次数: 425
作者单位
余锦华 南京气象学院应用气象学系, 南京 210044 
丁裕国 南京气象学院应用气象学系, 南京 210044 
基金项目:南京气象学院科研基金
中文摘要:将气候集成预报的MonteCarlo模拟方法应用到带有门限的海气耦合随机动力模式中。利用相空间概念确定预报误差,对不同相变量初值随机扰动的概率密度分布函数以及MonteCarlo模拟的集成样本数进行了最佳选择试验。结果表明,初值随机扰动的概率密度分布以均值为零的高斯分布为最好,集成样本数以N=32为宜。用气候持续预报来检测不同预报方法的预报技巧。对1982~1983年,1986~1987年的ENSO事件进行了个例预报试验,表明,MonteCarlo集成预报方法在一定程度上优于单个的随机动力模式预报,它们的预报效果都比气候持续预报的好。
中文关键词:气候集成预报  MonteCarlo模拟  短期气候预测  预报技巧
 
PROBE INTO CLIMATIC ENSEMBLE PREDICTION
Abstract:The Monte Carlo simulation is applied to an air-sea coupling stochastic dynamic model with threshold in this paper. The prediction error is defined by means of the concept of phase space,the best choice tests are made for probability density functions of initial stochastic perturbations of different phase space variables as well as for the ensemble sample size of Monte Carlo simulation . The results show that the zero mean Grass distribution is the best for the probability density functions of initial stochastic perturbations and the ensembe sample size N =32 is optimal .The forecast skill for different forecast methods is examined by the climatic persistence forecast.The test results of 1982~1983 and 1986~1987 ENSO events show that the Monte Carlo ensemble prediction is superior to any single stochastic dynamical models in smoe extent,and its forecast effect is totally superior to the climatic persistence forecasts.
keywords:climatic ensemble prediction  Monte Carlo simulation  short-term climate prediction  forecast skill
查看全文  查看/发表评论  下载PDF阅读器