Pivotal discrepancy measures for Bayesian modelling of spatio-temporal data

Within the field of geostatistics, Gaussian processes are a staple for modelling spatial and spatio-temporal data. Statistical literature is rich with estimation methods for the mean and covariance of such processes (in both frequentist and Bayesian contexts). Considerably less attention has been paid to developing goodness-of-fit tests for assessment of model adequacy. Jun et al. (Environmetrics 25(8):584-595, 2014) introduced a statistical test that uses pivotal discrepancy measures to assess goodness-of-fit in the Bayesian context. We present a modification and generalization of their statistical test. The initial method involves spatial partitioning of the data, followed by evaluation of a pivotal discrepancy measure at each posterior draw to obtain a posterior distribution of pivotal statistics. Order statistics from this distribution are used to obtain approximate p-values. Jun et al. (Environmetrics 25(8):584-595, 2014) use arbitrary partitions based on pre-existing spatial boundaries. The partitions are made to be of equal size. Our contribution is two-fold. We use K-means clustering to create the spatial partitions and we generalise Jun et al.'s approach to incorporate unequal partition sizes. Observations from a spatial or spatio-temporal process are partitioned using an appropriate feature vector that incorporates the geographic location of the observations into subsets (not necessarily of the same size). The method's viability is illustrated in a simulation study, and in an application to hoki (Macruronus novaezelandiae) catch data from a survey of the sub-Antarctic region.

Automated summary

Gaussian processes are widely used in geostatistics for modelling spatial and spatio-temporal data. While there are many estimation methods for the mean and covariance of such processes, there has been less focus on developing goodness-of-fit tests for model adequacy. This study presents a modified and generalized statistical test, using pivotal discrepancy measures, to assess goodness-of-fit in a Bayesian context. The proposed method involves spatial partitioning of the data and incorporates unequal partition sizes using K-means clustering. The viability of the method is demonstrated through simulation studies and an application to hoki catch data.