Abstract:Spatial data are often derived from multiple sources (e.g. satellites, in-situ sensors, survey samples) with different supports, but associated with the same properties of a spatial phenomenon of interest. It is common for predictors to also be measured on different spatial supports than the response variables. Although there is no standard way to work with spatial data with different supports, a prevalent approach used by practitioners has been to use downscaling or interpolation to project all the variables of analysis towards a common support, and then using standard spatial models. The main disadvantage with this approach is that simple interpolation can introduce biases and, more importantly, the uncertainty associated with the change of support is not taken into account in parameter estimation. In this article, we propose a Bayesian spatial latent Gaussian model that can handle data with different rectilinear supports in both the response variable and predictors. Our approach allows to handle changes of support more naturally according to the properties of the spatial stochastic process being used, and to take into account the uncertainty from the change of support in parameter estimation and prediction. We use spatial stochastic processes as linear combinations of basis functions where Gaussian Markov random fields define the weights. Our hierarchical modelling approach can be described by the following steps: (i) define a latent model where response variables and predictors are considered as latent stochastic processes with continuous support, (ii) link the continuous-index set stochastic processes with its projection to the support of the observed data, (iii) link the projected process with the observed data. We show the applicability of our approach by simulation studies and modelling land suitability for improved grassland in Rhondda Cynon Taf, a county borough in Wales.

Gaussian Markov random field spatial models in GAMLSS

Spatial Gaussian Markov Random Fields: Modelling, Applications and Efficient Computations

Spatial Latent Gaussian Modelling with Change of Support

Comparison between Highly Complex Location Models and GAMLSS

Distribution-Free Location-Scale Regression

A multivariate Gaussian random field prior against spatial confounding

Flexible Basis Representations for Modeling Large Non-Gaussian Spatial Data

An explicit link between Gaussian fields and Gaussian Markov random fields: the stochastic partial differential equation approach

A Bivariate Copula Additive Model for Location, Scale and Shape

Nonstationary Spatial Gaussian Markov Random Fields

sdmTMB: an R package for fast, flexible, and user-friendly generalized linear mixed effects models with spatial and spatiotemporal random fields

Exploring the Distributional Properties of the Non-Gaussian Random Field Models

Spatial GARCH models for unknown spatial locations – an application to financial stock returns

Estimation and prediction for spatial generalized linear mixed models with parametric links via reparameterized importance sampling

GAMLSS for Longitudinal Multivariate Claim Count Models

A Variational Approach for Modeling High-dimensional Spatial Generalized Linear Mixed Models

Statistical Modeling of Right-Censored Spatial Data Using Gaussian Random Fields

Modeling Big, Heterogeneous, Non-Gaussian Spatial and Spatio-Temporal Data Using FRK

Dimension Reduction and Alleviation of Confounding for Spatial Generalized Linear Mixed Models

GEE-assisted Forward Regression for Spatial Latent Variable Models

Estimation and Inference for Generalized Geoadditive Models