Review: Nonstationary Spatial Modeling, with Emphasis on Process Convolution and Covariate-Driven Approaches

TL;DR

This review discusses nonstationary spatial modeling techniques, especially process convolution and covariate-driven methods, highlighting their advantages over traditional stationary models in environmental data analysis.

Contribution

It provides a comprehensive overview of nonstationary spatial modeling approaches, emphasizing recent covariate-based methods and their interpretability and computational benefits.

Findings

Process convolution approaches effectively model nonstationarity.

Covariate-driven methods enhance interpretability and efficiency.

The literature review clarifies methodological developments in nonstationary spatial modeling.

Abstract

In many environmental applications involving spatially-referenced data, limitations on the number and locations of observations motivate the need for practical and efficient models for spatial interpolation, or kriging. A key component of models for continuously-indexed spatial data is the covariance function, which is traditionally assumed to belong to a parametric class of stationary models. While convenient, the assumption of stationarity is rarely realistic; as a result, there is a rich literature on alternative methodologies which capture and model the nonstationarity present in most environmental processes. This review document provides a rigorous and concise description of the existing literature on nonstationary methods, paying particular attention to process convolution (also called kernel smoothing or moving average) approaches. A summary is also provided of more recent methods which leverage covariate information and yield both interpretational and computational benefits. Note: the article is borrowed from Chapters 1 and 2 of the author's Ph.D. dissertation, joint with Catherine A. Calder.