Title
Statistical Models for Processes Varying in Space and Time
Speaker
Mike Stein, University of Chicago
Abstract
Modeling processes that vary in space and time requires more than a simple combination of approaches from time series and spatial statistics. In particular, one needs models and methods of analysis for capturing spatial-temporal interactions. This talk describes exploratory methods suitable for spatial-temporal processes observed at a set of fixed monitoring locations, which is common for meteorological and environmental data. An important and frequent form of a spatial-temporal interaction is a space-time asymmetry in the correlation structure: the correlation of the process at site A today and site B tomorrow is different from the correlation at site A tomorrow and B today. Some methods for modeling space-time asymmetries will be described along with extensions of these methods to the case where the region of space of interest is the surface of a sphere. Application of these methods and models to wind data in Ireland will be explored.
