多维时间序列交叉谱分析在降水场诊断中的应用

An Application of Multi-Dimensional Time Series Cross Spectrum Analysis to the Diagnosis of Yunnan's Precipitation Field

在经典时间序列的交叉谱分析基础上 ,建立多维交叉谱矩阵 ,对某一感兴趣的频率段交叉谱矩阵作特征值和特征向量分析。作为初步应用 ,在获得云南 5月雨量存在准4 5年的显著耦合振荡周期的前提条件下 ,对相应的交叉谱矩阵进行特征值和特征向量分析后发现 ,用前 3个相应协谱和正交谱的特征向量就可反映云南 5月雨量场在 4 5年耦合振荡周期上的大部分空间分布信息 ,并得到了一些具有天气气候学意义的结论。结果表明应用多维时间序列的交叉谱分析来研究云南 5月雨量场的演化特征是可行的。

After the multi-dimensional cross spectrum matrix introducing on the basis of classical time series cross spectrum, we analyzed the eigenvalue and characteristic vectors of cross spectrum matrix on a interested frequency range. As an application example, on the condition of having obtained quasi-4.5-year significant coupling oscillation period of Yunnan's May precipitation, the eigenvalue and characteristic vector of corresponding cross spectrum matrixes were analyzed. The leading 3 characteristic vectors of the covariant and orthogonal spectrums can reflect the most spatial distribution of Yunnan's May precipitation on quasi-4.5-year significant coupling oscillation period. Some results with synoptic-climatologic characteristics were also obtained. It was feasible that the multi-dimensional time series cross spectrum analysis was applied to studying the evolution characteristics of Yunnan’s May precipitation field.