Title: Physical Statistical Environmental Modeling
Speaker: Mark Berliner, Ohio State University
Abstract: Combining scientific reasoning, often reflected in numerical
models, and observational data is a crucial challenge in environmental
analysis and prediction. The hierarchical Bayesian framework provides
opportunities for merging diverse information sources in a coherent
framework, and in a fashion which manages the uncertainty. A key to
the hierarchical viewpoint is that separate statistical models are
developed for the process variables studied and the observations
conditional on those variables. Modeling process variables in this
way enables incorporation of scientific models across a spectrum of
levels of intensity ranging from qualitative use of physical reasoning
to strong reliance on numerical models. I will review the basic
notions and present two examples. The first involves understanding
controls on glacial dynamics. The second is a climate forecasting
analysis assessing impacts of different scenarios for future carbon
dioxide concentrations, incorporating model output from large-scale
climate system models.