Efficiently Parameterized Bayesian Hierarchical Models for Spatio-Temporal Processes

Efficiently Parameterized Bayesian Hierarchical Models for Spatio-Temporal Processes

03-13-2014

Abstract
The processes governing physical and environmental systems are often complex, involving dierent interacting scales of variability in space and time. The complexities, and often high dimensionality, of such spatio-temporal processes can be eectively addressed using a hierarchical modeling framework where a complex problem is decomposed into a series of simpler problems that are linked through rules of probability. The Bayesian hierarchical modeling framework allows for accounting for uncertainty in dierent levels of the model (processes and parameters) as well as utilizing potential a priori information based on scient c knowledge and expert opinion. In this talk, I will review Bayesian hierarchical modeling for spatio-temporal processes, eciently parameterized and science-based dynamical modeling, and dimension-reduced (low-rank) dynamical models. I will provide examples from environmental, ecological and atmospheric science problems.

Biography:
Ali Arab is an Assistant Professor of Statistics in the Department of Math-ematics and Statistics at Georgetown University. He received a BS in Applied Mathematics at the Iran University of Science and Technology (Tehran, Iran; 1999) and an MS in Applied Mathematics and Statistics at Southern Illinois University Edwardsville (2002) and PhD in Statistics at University of Missouri-Columbia (2007). He joined Georgetown University in 2007. His methodological research is in spatio-temporal and spatial statistics, and hierarchical Bayesian modeling. He is interested in applications of statistics in the environment and climate change, ecology, epidemiology, meteorology, and risk and reliability analysis. He has published in peer-reviewed journals in statistics, ecology, environmental sciences,sheries, reliability engineering, and epidemiology. 5:45pm reception (with refreshments) followed by the seminar at 6:00 in caputo 210.