ABSTRACT
In recent years, there has been an explosion of interest in spatio-temporal modeling and there have been a number of noteworthy publications in this field, including Le and Zidek, Cressie and Wikle and Banerjee et al. There are many ways in which space and time can be incorporated into a statistical model. Time is obviously different from space. Time is regarded as discrete because measurements are made at specified, commonly equally spaced, time points. A spatial-temporal random field is a stochastic process over a region and time period. In most applications, modeling the entire spatio-temporal structure will be impractical because of high dimensionality. Separable models impose a particular type of independence between space and time components. Non-separable processes will often be more difficult to understand than when separation processes can be assumed for space and time, and as a consequence modeling is often complex.
