Extraction and Predictability of Intraseasonal Signals in Infrared Brightness Temperature Data

We investigate here both the dominant modes of variability and the large-scale regimes associated to tropical convection that can be recovered from infrared brightness temperature data using data mining and machine learning approaches. A hierarchy of spatiotemporal patterns at different timescales is extracted using a nonlinear dimension reduction method, namely Nonlinear Laplacian Spectral Analysis (NLSA). Among other signals of interest, the method also separates very clearly the boreal winter and boreal summer intraseasonal oscillations. Predictability of the Madden-Julian oscillation (MJO) is then quantified using a cluster-based information-theoretic framework applied in a cyclostationary setting. The clustering is performed in the space of the NLSA temporal patterns and the results show a strong influence of ENSO in the early MJO season, while in the active phase the current state of MJO dominates the predictability results.