P-spline smoothing for spatial data collected worldwide

TL;DR

This paper introduces a computationally efficient non-parametric method using P-splines and GMRF priors on a geodesic grid to model global spatial data, demonstrated through climate data analysis.

Contribution

It presents a novel framework for spatial modeling on a spherical surface using sparse basis functions and priors, improving computational efficiency for large-scale data.

Findings

Effective modeling of global spatial data using P-splines on a geodesic grid.

Successful identification of climate features like the Intertropical Convergence Zone.

Enhanced computational efficiency due to sparse matrix structures.

Abstract

Spatial data collected worldwide at a huge number of locations are frequently used in environmental and climate studies. Spatial modelling for this type of data presents both methodological and computational challenges. In this work we illustrate a computationally efficient non parametric framework to model and estimate the spatial field while accounting for geodesic distances between locations. The spatial field is modelled via penalized splines (P-splines) using intrinsic Gaussian Markov Random Field (GMRF) priors for the spline coefficients. The key idea is to use the sphere as a surrogate for the Globe, then build the basis of B-spline functions on a geodesic grid system. The basis matrix is sparse and so is the precision matrix of the GMRF prior, thus computational efficiency is gained by construction. We illustrate the approach on a real climate study, where the goal is to identify the Intertropical Convergence Zone using high-resolution remote sensing data.