复经验正交函数在水温预报中的应用

An Application of Complex Empirical Orthogonal Function in the Prediction of Sea Temperature

复经验正交函数（ＣＥＯＦ）是一种新的物理场的分解方法，它除了可以揭示物理场的时、空分布和变化规律外，还可以反映波动或平流效应。故本文首次将该方法移植到预报方法研究上来，建立了渤海、黄海的表、底层水温预报方案。为了检验该预报方案的可靠性，试报的结果除与实测值进行比较外，同时还与依据经验正交函数（ＥＯＦ）建立预报方法的预报结果进行了对比。结果表明，对于南、北黄海表层水温的试报结果，ＣＥＯＦ方法试报的平均绝对误差比ＥＯＦ方法预报的误差平均小０．２℃；而对于底层水温，两种方法的预报误差相近。

The complex empirical orthogonal function(CEOF) is one of the new decomposition methods for physical fields. It reveals not only the spatial and temporal distributions and characteristics but also the fluctuations or advection effects of physical fields. Therefore, the paper applies for the first time this method to the study of predictive method and establishes predictive programme for the prediction of sea surface and bottom temperatures in the Bobal and Huanghai Seas.In order to verify the reliability of the predictive programme,a comparison between the results of trial prediction and observational values is made, as well as that between the former from CEOF and those from EOF. The results show that the average absolute error from the results of trial prediction of sea surface temperature in the southern and northern Huanghai Sea using CEOF method is 0. 2℃smaller than that using EOF method; but such an error is quite close in prediction of sea bottom temperature.