Constructing Priors that Penalize the Complexity of Gaussian Random Fields

DOI10.1080/01621459.2017.1415907zbMATH Open1478.62279arXiv1503.00256OpenAlexW2963282124MaRDI QIDQ5229926FDOQ5229926

Authors: Geir-Arne Fuglstad, Finn Lindgren, Håvard Rue, D. Simpson 

Publication date: 19 August 2019

Published in: Journal of the American Statistical Association (Search for Journal in Brave)

Abstract: Priors are important for achieving proper posteriors with physically meaningful covariance structures for Gaussian random fields (GRFs) since the likelihood typically only provides limited information about the covariance structure under in-fill asymptotics. We extend the recent Penalised Complexity prior framework and develop a principled joint prior for the range and the marginal variance of one-dimensional, two-dimensional and three-dimensional Mat'ern GRFs with fixed smoothness. The prior is weakly informative and penalises complexity by shrinking the range towards infinity and the marginal variance towards zero. We propose guidelines for selecting the hyperparameters, and a simulation study shows that the new prior provides a principled alternative to reference priors that can leverage prior knowledge to achieve shorter credible intervals while maintaining good coverage. We extend the prior to a non-stationary GRF parametrized through local ranges and marginal standard deviations, and introduce a scheme for selecting the hyperparameters based on the coverage of the parameters when fitting simulated stationary data. The approach is applied to a dataset of annual precipitation in southern Norway and the scheme for selecting the hyperparameters leads to concervative estimates of non-stationarity and improved predictive performance over the stationary model.

Full work available at URL: https://arxiv.org/abs/1503.00256

Recommendations

• Random field priors for spectral density functions
• Hyperpriors for Matérn fields with applications in Bayesian inversion
• Objective Bayesian Analysis of Spatially Correlated Data
• Gaussian random fields: with and without covariances
• Default priors for Gaussioan processes

zbMATH Keywords

Bayesiannonstationaryspatial modelspriorsrangepenalized complexity

Mathematics Subject Classification ID

Bayesian inference (62F15) Random fields; image analysis (62M40)

Cites Work

• Interpolation of spatial data. Some theory for kriging
• Approximate Bayesian inference for latent Gaussian models by using integrated nested Laplace approximations (with discussion)
• Objective Bayesian Analysis of Spatially Correlated Data
• Inconsistent Estimation and Asymptotically Equal Interpolations in Model-Based Geostatistics
• An explicit link between Gaussian fields and Gaussian Markov random fields: the stochastic partial differential equation approach
• Penalising model component complexity: a principled, practical approach to constructing priors
• Default priors for Gaussioan processes
• Scale-dependent priors for variance parameters in structured additive distributional regression
• Objective Bayesian analysis of spatial data with measurement error
• Adaptive Bayesian estimation using a Gaussian random field with inverse gamma bandwidth
• The formal definition of reference priors
• On the consistent separation of scale and variance for Gaussian random fields
• Problems with likelihood estimation of covariance functions of spatial Gaussian processes
• Asymptotic properties of a maximum likelihood estimator with data from a Gaussian process
• Objective Bayesian analysis of spatial data with uncertain nugget and range parameters
• Objective Bayesian analysis of geometrically anisotropic spatial data

Cited In (48)

• Bayesian inference with subset simulation in varying dimensions applied to the Karhunen-Loève expansion
• A spatial modeling framework for monitoring surveys with different sampling protocols with a case study for bird abundance in Mid-Scandinavia
• Probabilistic forecasts of arctic sea ice thickness
• Fast, Approximate Maximum Likelihood Estimation of Log-Gaussian Cox Processes
• High-dimensional modeling of spatial and spatio-temporal conditional extremes using INLA and Gaussian Markov random fields
• PC priors for residual correlation parameters in one-factor mixed models
• Improving Bayesian Local Spatial Models in Large Datasets
• A joint Bayesian space-time model to integrate spatially misaligned air pollution data in R-INLA
• Stochastic Convergence Rates and Applications of Adaptive Quadrature in Bayesian Inference
• Additive multivariate Gaussian processes for joint species distribution modeling with heterogeneous data
• An efficient workflow for modelling high-dimensional spatial extremes
• Joint modeling and prediction of massive spatio-temporal wildfire count and burnt area data with the INLA-SPDE approach
• Combining heterogeneous spatial datasets with process-based spatial fusion models: a unifying framework
• Convergence of Gaussian process regression with estimated hyper-parameters and applications in Bayesian inverse problems
• Spatial 3D Matérn priors for fast whole-brain fMRI analysis
• Intuitive joint priors for Bayesian linear multilevel models: the R2D2M2 prior
• District-level estimation of vaccination coverage: discrete vs continuous spatial models
• Bayesian multiresolution modeling of georeferenced data: an extension of `LatticeKrig'
• Small Area Estimation for Disease Prevalence Mapping
• Some models are useful, but how do we know which ones? Towards a unified Bayesian model taxonomy
• A spliced gamma-generalized Pareto model for short-term extreme wind speed probabilistic forecasting
• Bayesian fixed-domain asymptotics for covariance parameters in a Gaussian process model
• A simulation study of disaggregation regression for spatial disease mapping
• A spatiotemporal analysis of NO\(_2\) concentrations during the Italian 2020 COVID-19 lockdown
• Generalization of the power-law rating curve using hydrodynamic theory and Bayesian hierarchical modeling
• Spatio-temporal downscaling emulator for regional climate models
• Spatial regression modeling via the R2D2 framework
• Modeling and simulating depositional sequences using latent Gaussian random fields
• Spatiotemporal probabilistic wind vector forecasting over Saudi Arabia
• Conditional probability and ratio-based approaches for mapping the coverage of multi-dose vaccines
• A unified view on Bayesian varying coefficient models
• Intuitive joint priors for variance parameters
• Modelling sub-daily precipitation extremes with the blended generalised extreme value distribution
• The SPDE approach to Matérn fields: graph representations
• Gradient boosting with extreme-value theory for wildfire prediction
• A flexible generalized Poisson likelihood for spatial counts constructed by renewal theory, motivated by groundwater quality assessment
• Approximate Bayesian inference for analysis of spatiotemporal flood frequency data
• Impact of jittering on raster- and distance-based geostatistical analyses of DHS data
• The integrated nested Laplace approximation applied to spatial log-Gaussian Cox process models
• A non-stationary model for spatially dependent circular response data based on wrapped Gaussian processes
• Compression of climate simulations with a nonstationary global spatiotemporal SPDE model
• Spatiotemporal wildfire modeling through point processes with moderate and extreme marks
• Spatial modeling with R-INLA: a review
• Covariance–Based Rational Approximations of Fractional SPDEs for Computationally Efficient Bayesian Inference
• A general theory for preferential sampling in environmental networks
• Disciplinary proper orthogonal decomposition and interpolation for the resolution of parameterized multidisciplinary analysis
• A stochastic locally diffusive model with neural network-based deformations for global sea surface temperature
• Bayesian prediction of spatial data with non-ignorable missingness

This page was built for publication: Constructing Priors that Penalize the Complexity of Gaussian Random Fields

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5229926)