Long-lead prediction of Pacific SSTs via Bayesian dynamic modeling

Tropical Pacific sea surface temperatures (SSTs) and the accompanying El Ni no-Southern Oscillation phenomenon are recognized as significant components of climate behavior. The atmospheric and oceanic processes involved displa y highly complicated variability over both space and time. Researchers have applied both physically derived modeling and statistical approaches to dev elop long-lead predictions of tropical Pacific SSTs. The comparative succes ses of these two approaches are a subject of substantial inquiry and some c ontroversy. Presented in this article is a new procedure for long-lead fore casting of tropical Pacific SST fields that expresses qualitative aspects o f scientific paradigms for SST dynamics in a statistical manner. Through th is combining of substantial physical understanding and statistical modeling and learning, this procedure acquires considerable predictive skill. Speci fically, a Markov model, applied to a low-order (empirical orthogonal funct ion-based) dynamical system of tropical Pacific SST, with stochastic regime transition, is considered. The approach accounts explicitly for uncertaint y in the formulation of the model, which leads to realistic error bounds on forecasts. The methodology that makes this possible is hierarchical Bayesi an dynamical modeling.