Abstract:A non-stationary Gaussian random field model is developed based on a combination of the stochastic partial differential equation (SPDE) approach and the classical deformation method. With the deformation method, a stationary field is defined on a domain which is deformed so that the field becomes non-stationary. We show that if the stationary field is a Mat'ern field defined as a solution to a fractional SPDE, the resulting non-stationary model can be represented as the solution to another fractional SPDE on the deformed domain. By defining the model in this way, the computational advantages of the SPDE approach can be combined with the deformation method's more intuitive parameterisation of non-stationarity. In particular it allows for independent control over the non-stationary practical correlation range and the variance, which has not been possible with previously proposed non-stationary SPDE models. The model is tested on spatial data of significant wave height, a characteristic of ocean surface conditions which is important when estimating the wear and risks associated with a planned journey of a ship. The model parameters are estimated to data from the north Atlantic using a maximum likelihood approach. The fitted model is used to compute wave height exceedance probabilities and the distribution of accumulated fatigue damage for ships traveling a popular shipping route. The model results agree well with the data, indicating that the model could be used for route optimization in naval logistics.

Spatial analysis of wave direction data using wrapped Gaussian processes

The wrapped skew Gaussian process for analyzing spatio-temporal data

A multivariate pseudo-likelihood approach to estimating directional ocean wave models

Spatio-temporal circular models with non-separable covariance structure

Spatial Hyperspheric Models for Compositional Data

Directional Gaussian spatial processes for South African wind data

Spatial Process Gradients and Their Use in Sensitivity Analysis for Environmental Processes

Nonparametric estimation of circular trend surfaces with application to wave directions

Probabilistic Spatiotemporal Modeling of Day-Ahead Wind Power Generation with Input-Warped Gaussian Processes

Inference for spatial processes using imperfect data from measurements and numerical simulations

Bayesian Modeling with Spatial Curvature Processes

Warped multifidelity Gaussian processes for data fusion of skewed environmental data

On modelling positive continuous data with spatio-temporal dependence

On the Reliability and Limitations of the SPAC Method with a Directional Wavefield

Nearest-Neighbor Mixture Models for Non-Gaussian Spatial Processes

Species distribution modelling with spatio-temporal nearest neighbour Gaussian processes

Modal Analysis of Spatiotemporal Data via Multivariate Gaussian Process Regression

Deformed SPDE models with an application to spatial modeling of significant wave height

Stochastic Asymmetry Properties of 3D Gauss-Lagrange Ocean Waves with Directional Spreading

Space-Time Geostatistical Models with both Linear and Seasonal Structures in the Temporal Components

Bag of DAGs: Inferring Directional Dependence in Spatiotemporal Processes