A heterogeneous Bayesian regression model for skewed spatial data

After displaying skewness, spatial data can cause nonstationarity. This paper develops a hierarchical skew-Gaussian process capable of simultaneously handling skewness and nonstationarity. At the first level of the hierarchy, we specify a multivariate skew-normal distribution for the realizations of the process over a fixed finite set. At the second level, it is extended to a process across the domain based on the referenced locations. We applied Markov chain Monte Carlo algorithms for Bayesian inference. The proposed methodology is illustrated by some simulation experiments and by applying it to rainfall data modelling.