Predictable Patterns of the Atmospheric Low-Level Circulation over the Indo-Pacific Region in Project Minerva: Seasonal Dependence and Intra-Ensemble Variability

The predictable patterns and the intra-ensemble variability of the 850-hPa monthly zonal wind over the tropical Indo-Pacific domain are investigated using 7-month hindcasts for 1983-2009 from Project Minerva. Applied to the ensemble hindcasts initialized on the 1st of May and November, a maximum signal-to-noise empirical orthogonal function analysis identifies the patterns of high predictability as the hindcasts progress. For both initial months, the most predictable patterns are associated with El Niño-Southern Oscillation (ENSO). The second predictable patterns with May initialization reflect the anomalous evolution of the western North Pacific (WNP) monsoon, characterized by a northward shift of the WNP anomalous anticyclone/cyclone in summer and a southward shift in fall. The intra-ensemble variability shows a strong seasonality that affects different predictable patterns in different seasons. For May initialization, the dominant patterns of the ensemble spread bear some resemblance to the predictable WNP patterns in summer and ENSO patterns in fall, which reflect the noise-induced differences in the evolution of the predictable signals among ensemble members. On the other hand, the noise patterns with November initialization are dominated by the northern extratropical atmospheric perturbations from winter to early spring, which expand southward through the coupled footprinting mechanism to perturb the ENSO evolution in different ensemble members. In comparison, extratropical perturbations in the Southern Hemisphere, most significant in early months with May initialized predictions, is less effective in affecting the tropical circulation.