论文部分内容阅读
提出一种基于主分量典型相关分析（ＰＣ－ＣＣＡ）的广义典型混合回归模式，用于建立ＮＩＮＯ海区ＳＳＴ预报方案．该模式引入ＥＥＯＦ、ＰＲＥＳＳ准则和集成预报等技术思想，在优选物理因子，确定最佳模式参数的基础上，对ＮＩＮＯ海区海温指数所作的超前１—４季度预报试验取得优良效果．试验表明，该模式方案性能稳定，其总体预报技术水平已达到美国ＮＯＡＡ／ＮＷＳ／ＮＣＥＰ／气候诊断公报（ＣＰＣ）所用同类模式水平．而本模式方案预报同类产品所需因子数远少于ＣＰＣ方法。这就有可能为建立我国的ＥＮＳＯ业务监测系统提供有益的基础。
A generalized typical mixed regression model based on Principal Component Canonical Correlation Analysis (PC-CCA) is proposed to establish the SST prediction scheme in NINO sea area. This model introduces the technical ideas of EEOF, PRESS and integrated forecasting. Based on the optimization of the physical factors and the determination of the best mode parameters, this model achieves excellent results in the first 1-4 quarters of forecast tests conducted by the SSTA of NINO. Experiments show that the performance of this scheme is stable and its overall forecasting technology has reached the level of similar models used in the NOAA / NWS / NCEP / Climate Diagnostic Bulletin (CPC) in the United States. However, the number of factors needed for forecasting similar products in this model is far less than that of CPC. This may provide a useful basis for the establishment of our country’s ENSO operational monitoring system.