Mixture-based modeling for space-time data

An overview of spatio-temporal covariance functions built through mixtures is presented in this paper. We highlight the potentiality of mixture modeling for the construction of nonseparable space-time covariances. In particular, we make use of mixed forms (MF), copulas, and completely monotone functions as the basic setup representing powerful instruments to build mixture-based covariance functions. We re-analyze, by using a particular model of mixtures, the Indian Ocean wind speed data and compare the results with others previously published in the literature.