Bayesian Hierarchical Models for Atmospheric and Oceanic Processes.

Christopher K. Wikle

Department of Statistics
University of Missouri

Abstract

A modeling paradigm, relatively new to geophysical applications, is introduced in the context of atmospheric and oceanic processes in the presence of uncertainty in data, model, and parameters. The approach, Bayesian hierarchical modeling (BHM), relies on both physical reasoning and statistical techniques for data processing and uncertainty management. Critical to this approach is that physical and dynamical constraints are easily incorporated into a conditional formulation, so that the series of relatively simple, yet physically realistic, conditional models leads to a much more complicated joint space-time covariance structure than can be specified directly. One of the benefits of this approach is that relatively simple dynamical models, with random parameters, are able to model more complicated processes in the presence of data. In addition, this approach easily accommodates different data sources. The theory and current practice of BHM of air-sea interaction physics will be introduced and demonstrated in this talk.