Factor copula models for data with spatio-temporal dependence

We propose a new copula model for spatial data that are observed repeatedly in time. The model is based on the assumption that there exists a common factor that affects the measurements of a process in space and in time. Unlike models based on multivariate normality, our model can handle data with tail dependence and asymmetry. The likelihood for the proposed model can be obtained in a simple form and therefore parameter estimation is quite fast. Simulation from this model is straightforward and data can be predicted at any spatial location and time point. We use simulation studies to show different types of dependencies, both in space and in time, that can be generated by this model. We apply the proposed copula model to hourly wind data and compare its performance with some classical models for spatio-temporal data.