Compactly Supported Multivariate Covariance Matrix Functions

• Activity Number: 313
• Type: Contributed
• Date/Time: Tuesday, August 2, 2011 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistics and the Environment
• Author(s): Juan Du*+ and Chunsheng Ma
• Companies: Kansas State University and Wichita State University
• Keywords: Covariance matrix function ; Covariance tapering ; Cross covariance ; Direct covariance ; Multivariate random field ; Variogram matrix function

Abstract

Covariance tapering is a useful technique to mitigate the numerical burdens in dealing with the large spatial data sets. This technique is applied to multivariate case and compactly supported multivariate covariance functions are needed for multivariate tapering functions. To meet this need, we construct a class of multivariate random fields in R^d whose direct and cross covariance functions are compactly supported by using the convolution approach. In addition, a class of second-order stochastic processes whose direct and cross covariance functions are of Pólya type is also derived. Simulation study is conducted to show the computational gain and application in cokrigging by using proposed multivariate tapering function.