Nonparametric Covariance Functions for Multivariate Geostatistics

• Activity Number: 601
• Type: Topic Contributed
• Date/Time: Thursday, August 2, 2012 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistics and the Environment
• Author(s): Juan Hu*+ and Hao Zhang
• Companies: Purdue University and Purdue University
• Keywords: auxiliary variables ; cokriging ; interpolation ; multivaraite process ; spatial statistics

Abstract

The cross covariance function is employed in cokriging, i.e., the best linear unbiased prediction of the primary variable using some auxiliary variables. There have been recent developments in the parametric cross covariance functions for multivariate spatial processes. However, estimation of parameters is generally complex due to the increased number of parameters and the fact that the parameters have to satisfy some constraints. In addition, the existing parametric models are not sufficiently flexible in the sense that they cannot accommodate any given marginal covariance models. We intend to overcome these shortcomings by proposing a non-parametric alternative.