ABSTRACT

Effective analysis of natural processes requires efficient tools for storing, processing and visualizing multi-dimensional data as well as detailed spatial and temporal observations. Many earth science observations, however, are generally sparse and irregularly sampled in space and time and thus potentially compromise an effective analysis of the process. This research is concerned with both the development and implementation of fourdimensional kriging, an extension of regionalized variable theory, and its role as an interpolative component of a four-dimensional GIS. Four-dimensional geostatistical kriging will be used for the analysis of spatio-temporal correlation and the estimation and corresponding variance associated with unsampled observations in space and time. An application of this research is shown by the reconstruction and visualization of a complex non-point source subsurface contaminant flow from point source observations.