摘要
本文介绍一个跨季度汛期降水预测的集会预测模拟试验，以１９８７年为例，用２月１日～２６日每隔５天的观测资料作初值进行了６个积分，取其平均作为集合预测，共有７个预测结果，计算了７个预测与观测场的同号率，计算了６个预测的标准差，结果表明：集会预测比个别试验更为合理，这６个试验的结果在大尺度分布上是相近的，其差别也是不大的。集合预测值得继续进行试验。文中也讨论了一些由于大气运动固有的动力学特性，其集合预测与经典的数学考虑有所区别，对集合预测提出了一些新的看法，许多问题还需要继续深入探讨，以期能对该方法作出恰当的评价。
An experiment in ensemble prediction of the summer precipitation over East China is described in this paper. Six predictions start from February 1, 6, 11, 21 and 26, 1987. Taking the average of the six cases as the ensemble mean, thus we have seven predictions. The rate of the same signs between simulated rainfall anomaly and observed one are calculated. We also calculate the standard deviation of simulated rainfall for the six experiments. The results show that the difference between the six experiments are small in the distributions of rainfall on a large scale, the ensemble mean is more reasonable.The features of atmospheric dynamics are also discussed in this paper. It shows that the ensemble prediction of short-range climate differs from the short time and medium range numerical forecasting by physical base.
出处
《大气科学》
CSCD
北大核心
1996年第5期593-598,共6页
Chinese Journal of Atmospheric Sciences
基金
国家攀登计划"气候动力学及气候预测理论的研究"
气候与环境预测研究中心
LASG及中国科学院KY85-10项目
关键词
降水预报
汛期预测
集合预告模拟
short range climate anomaly rainfall anomaly percentage standard deviation ensemble precipitation
