Space-time modelling of coupled spatio-temporal environmental variables
Dani Gamerman (UFRJ)

We propose a dynamic factor model for spatio-temporal coupled environmental variables. The model is discussed in a state-space framework which is useful for interpolation and forecast of the variable of interest. The role of the measurement matrix in spatial interpolation is considered and the proposal of its stochastic specification is discussed. Full probabilistic inference for the model parameters is facilitated by Markov chain Monte Carlo (MCMC) algorithms. Standard MCMC for dynamic linear models are adapted to our model specification and predictive results are discussed for two different data sets with variables measured at two different scales.