中国冬夏季气温和降水异常耦合关系的SVD与MEOF分析对比

以揭示中国冬、夏季气温、降水之间的同期耦合时空特征为例,比较了SVD与MEOF分析所得主要正交模的相关关系及均衡性,论证了SVD方法在揭示两个要素场序列互相关关系上的优越性。结果表明:1) SVD的分解对象仅包含反映两种要素场相关关系的信息,而MEOF还包含了两个要素场各自的自相关信息,前者分解对象更纯粹。2) MEOF和SVD揭示的气温和降水空间耦合模态正负异常分布大致相似,但相较SVD所揭示的耦合模态,MEOF分析的冬、夏季气温、降水主要正交模存在不均衡现象,即一个主要MEOF正交模往往侧重描述两种异常要素场中的一种,这种差异在冬季尤为突出。SVD方法揭示了中国冬、夏季同期气温、降水间局域高温少降水或低温多降水是二者主要异常配置,且相关性夏强冬弱。3)比较SVD、MEOF方法同序正交模的相关系数和均衡度发现,SVD多数正交模的相关系数较MEOF的更大、更稳定,且均通过显著性检验,MEOF的冬季部分正交模相关系数未通过显著性检验; SVD主要正交模的均衡性冬季优于MEOF分析、夏季优于(第一模态)或者接近于(第二、三模态) MEOF分析。故SVD方法是揭示两要素之间耦合关系的更好方法。

Comparison of SVD and MEOF analysis on anomalous coupling relationship between temperature and precipitation in winter and summer in China

Taking study on the simultaneous coupling temporal and spatial characteristics between temperature and precipitation of China as an example in winter and summer respectively, the correlation relationship and equilibrium of the main orthogonal modes of SVD and MEOF analysis are compared, and the superiority of SVD method in revealing correlation relationship between two element fields is demonstrated. The main conclusions are as follows:(1) Comparing the principle of SVD with that of MEOF, the direct decomposition object of SVD is more pure and only contains information reflecting correlation relationship between two element fields, while that of MEOF also includes the autocorrelation information of two element fields itself, respectively. (2) The positive and negative anomalous distributions of the spatial coupled modes between air temperature and precipitation revealed by MEOF and SVD are approximately similar, but the main MEOF orthogonal modes between air temperature and precipitation in winter and summer are unbalanced compared with those revealed by SVD. That is, one of the main MEOF orthogonal modes often focuses on describing one of the two anomalous element fields. This difference is especially pronounced in winter. The SVD method reveals that the local high temperature and less precipitation or low temperature and more precipitation are the main abnormal configurations between temperature and precipitation in winter and summer over the same period in China, and the correlation is strong in summer and weak in winter. (3) By comparing the correlation coefficients and equalization of the same order orthogonal modes of SVD with those of MEOF, it is found that the correlation coefficients of most orthogonal modes of SVD are larger and more stable than those of MEOF, and all pass the significance test, and part of correlation coefficients of MEOF orthogonal modes in winter do not pass the significance test. The equalization of SVD main orthogonal modes is superior to that of MEOF analysis in winter, and is superior to (for the first mode) or close to (for the second and third modes) that of MEOF in summer. So the SVD method is a better way to reveal the coupling relationship between the two elements.