Rotated spectral principal component analysis (rsPCA) for identifying dynamical modes of variability in climate systems

Abstract: Spectral PCA (sPCA), in contrast to classical PCA, offers the advantage of identifying organized spatio-temporal patterns within specific frequency bands and extracting dynamical modes. However, the unavoidable tradeoff between frequency resolution and robustness of the PCs leads to high sensitivity to noise and overfitting, which limits the interpretation of the sPCA results. We propose herein a simple non-parametric implementation of the sPCA using the continuous analytic Morlet wavelet as a robust estimator of the cross-spectral matrices with good frequency resolution. To improve the interpretability of the results when several modes of similar amplitude exist within the same frequency band, we propose a rotation of eigenvectors that optimizes the spatial smoothness in the phase domain. The developed method, called rotated spectral PCA (rsPCA), is tested on synthetic data simulating propagating waves and shows impressive performance even with high levels of noise in the data. Applied to historical sea surface temperature (SST) time series over the Pacific Ocean, the method accurately captures the El Ni~{n}o-Southern Oscillation (ENSO) at low frequency (2 to 7 years periodicity). At high frequencies (sub-annual periodicity), at which several extratropical patterns of similar amplitude are identified, the rsPCA successfully unmixes the underlying modes, revealing spatially coherent patterns with robust propagation dynamics. Identification of higher frequency space-time climate modes holds promise for seasonal to subseasonal prediction and for diagnostic analysis of climate models.