A Class of Covariate-Dependent Covariance Functions for the Analysis of Spatio-Temporal Data

• Activity Number: 591
• Type: Contributed
• Date/Time: Wednesday, August 4, 2010 : 2:00 PM to 3:50 PM
• Sponsor: Section on Statistics and the Environment
• Author(s): Michele Guindani*+ and Brian Reich and Jo Eidsvik and Alexandra Schmidt and Amy Nail
• Companies: University of New Mexico and North Carolina State University and Norwegian University of Science and Technology and Universidade Federal do Rio de Janeiro and North Carolina State University
• Keywords: Spatial Statistics ; Stationarity ; Covariance function ; Ozone modeling ; SAMSI

Abstract

The customary assumption of an underlying stationary and isotropic spatial process is untenable in many practical environmental problems, due to the influence of local effects in the correlation structure. In fact, the correlation structure might be highly dependent on local characteristics of the domain of study (e.g. latitude, longitude and altitude), as well as other locally defined covariate information. We provide a flexible and computationally feasible way for allowing the correlation structure of the underlying processes to depend on local covariate information.This is obtained by means of a convenient specification of the mean spatial effects, and their reinterpretation in the covariance domain. We discuss the properties of the induced covariance functions by means of a simulation study and the analysis of data observed at ozone-monitoring stations.