Identification of the Madden–Julian Oscillation With Data‐Driven Koopman Spectral Analysis
Identification of the Madden–Julian Oscillation With Data‐Driven Koopman Spectral Analysis
复制标题
DOI：
10.1029/2023gl102743
复制
发表时间：
2023-05
影响因子：
5.2
通讯作者：
B. Lintner;D. Giannakis;M. Pike;J. Slawinska
中科院分区：
文献类型：
--
作者：
B. Lintner;D. Giannakis;M. Pike;J. Slawinska
马登 - 朱利安振荡（MJO）是热带季节内变率的主要模态，通常使用实时多变量MJO（RMM）指数来识别，该指数基于对近赤道高低层纬向风及向外长波辐射的联合经验正交函数（EOF）分析。在此，我们不采用传统的EOF方法，而是应用基于动力系统理论（库普曼算子）的算子理论形式体系，提取出一个类似于RMM的指标，该指标展现出一些特性，完善了对MJO的特征描述及可预测性。具体而言，库普曼算子特征函数的频谱（其特征值对应模态周期）包含一个主导的季节内模态，周期约为50天。此外，这一主导季节内特征函数的振幅呈现出季节性调制，在北半球冬季明显达到峰值。最后，由复数形式的库普曼MJO特征函数构成的相空间，相较于RMM，呈现出更平滑的时间演变和更高程度的自相关性，这可能有助于提升预测技巧。
The Madden‐Julian Oscillation (MJO), the dominant mode of tropical intraseasonal variability, is commonly identified using the realtime multivariate MJO (RMM) index based on joint empirical orthogonal function (EOF) analysis of near‐equatorial upper and lower level zonal winds and outgoing longwave radiation. Here, in place of conventional EOFs, we apply an operator‐theoretic formalism based on dynamic systems theory (the Koopman operator) to extract an analog to RMM that exhibits certain features that refine the characterization and predictability of the MJO. In particular, the spectrum of Koopman operator eigenfunctions, with eigenvalues corresponding to mode periods, contains a leading intraseasonal mode with period of ∼50 days. Moreover, the amplitude of this leading intraseasonal eigenfunction exhibits a seasonal modulation clearly peaked in boreal winter. Finally, the phase space formed by the complex Koopman MJO eigenfunction exhibits a smoother temporal evolution and higher degree of autocorrelation than RMM, which may contribute to enhanced predictive skill.