Modelling environmental monitoring data coming from different surveys

Abstract

Environmental monitoring networks are providing large amounts of spatio-temporal data. Air pollution data, as other environmental data, exhibit a spatial and a temporal correlated nature. To improve the accuracy of predictions at unmonitored locations, there is a growing need for models capturing those spatio-temporal correlations. With this work, we propose a spatio-temporal model for gaussian data collected in a few number of surveys. We assume the spatial correlation structure to be the same in all surveys. In an application of this model to real data, concerning heavy metal concentrations in mosses collected from three surveys occurring between 1992 and 2002 in mainland Portugal, the data set is dense in the spatial dimension but sparse in the temporal one, thus our model-based approach corresponds to a saturated correlation model in the time dimension. A novel interpretation for the space-time covariance function is introduced. A simulation study, aiming to validate the model, provided better results in terms of accuracy with the novel covariance function. Prediction maps of the observed variable for the most recent survey, and of the inter- polation error as a measure of accuracy, are presented.