Bayesian Prediction and Projection of Sea Levels

Tuesday, 16 December 2014
Mark Berliner, Ohio State University, Columbus, OH, United States
I will begin with a brief review of Bayesian hierarchical modeling and then turn to a model for sea levels. It is well-accepted that global sea levels have been rising in response to rising global temperatures. The strategy is the development of a Bayesian hierarchical model of sea levels. The hierarchical nature of the model is formulated to enable inference at various spatial scales. Further, temperature is incorporated in the model as a predictor or explanatory variable. Hence, information regarding future sea levels provided by the model rely on information regarding future temperatures. Forming predictions of future temperatures can be done in several
ways, depending on the goals of the analysis. I consider two classes of goals. In the first we seek short-term or medium-range forecasts as in weather-like forecasting. In the second we seek projections of sea levels under various emissions scenarios as in studies of the impacts of climate change. I illustrate methods and results for each class and suggest how results can contribute to decision support.