Non-Stationary Spatial Modeling

TL;DR

This paper introduces a hierarchical non-stationary spatial model that accounts for spatial dependence variability across locations, improving upon traditional stationary models by incorporating uncertainty in the non-stationarity specification.

Contribution

It develops a hierarchical process-convolution model for non-stationary spatial dependence that explicitly includes uncertainty in the non-stationarity structure.

Findings

Successfully applied to toxic waste remediation data

Provides a valid covariance structure for non-stationary processes

Enhances spatial modeling flexibility and accuracy

Abstract

Standard geostatistical models assume stationarity and rely on a variogram model to account for the spatial dependence in the observed data. In some instances, this assumption that the spatial dependence structure is constant throughout the sampling region is clearly violated. We present a spatial model which allows the spatial dependence structure to vary as a function of location. Unlike previous formulations which do not account for uncertainty in the specification of this non-stationarity (eg. Sampson and Guttorp (1992)), we develop a hierarchical model which can incorporate this uncertainty in the resulting inference. The non-stationary spatial dependence is explained through a constructive "process-convolution" approach, which ensures that the resulting covariance structure is valid. We apply this method to an example in toxic waste remediation.