Hierarchical Spatial and Spatio-Temporal Statistical Modeling for Environmental Applications.

Spatially and serially correlated data are ubiquitous in the environmental sciences.  In general, real-world environmental processes are complex, containing many sources of uncertainty, and thus it is often not feasible to consider these processes from a joint modeling perspective.  Instead, these processes often must be considered as a coherently linked system of conditional models.  As a consequence, in recent years, it has been increasingly recognized by statisticians that when modeling in the presence of uncertainties associated with observations, process and parameters, a hierarchical framework is particularly appealing.  This talk will provide a brief overview of hierarchical approaches to modeling environmental time series. Finally, the effectiveness of hierarchical modeling is illustrated through real environmental applications involving forecasting agricultural yield.