A Multivariate Index for Tropical Intraseasonal Oscillations Based on the Seasonally‐Varying Modal Structures

Abstract The spatial structure and propagation characteristics of tropical intraseasonal oscillations vary substantially by season. In this study, these seasonal variations are identified using a multivariate sliding‐window Empirical Orthogonal Function (EOF) analysis. The two modes comprising the leading EOF pair have equal variances and depict the propagation of intraseasonal oscillations in convection and low‐level circulation over the Indian Ocean, the Maritime Continent, and the western Pacific region in the equatorial summer hemisphere. In contrast, the upper tropospheric circulation shows more structure in the winter hemisphere. It is suggested that this variation in seasonality with height is an inherent feature of intraseasonal oscillations. A new multivariate index for tropical intraseasonal oscillations (MII) is developed based on the leading EOFs and represents the three‐dimensional structure of intraseasonal variability in all seasons. The MII is computed by projecting intraseasonal anomalies onto the leading EOFs pair, and it exhibits clearly delineated but smooth seasonal transitions and rich meridional structure. The real‐time version of this new index, rMII, is shown to be similar to MII, with a correlation of 0.9. Compared to the widely used Real‐time Multivariate MJO (RMM) index, the power spectrum of rMII represents substantially greater intraseasonal variance, and the application of rMII in dynamical forecast models indicates rMII is skillfully predicted for an additional week compared to RMM.