Geographical Analysis, Volume 42 Issue 2, Pages 161 – 179, Published Online 13 Apr 2010
Gerard B. M. Heuvelink and Daniel A. Griffith
“Many branches within geography deal with variables that vary not only in space but also in time. Therefore, conventional geostatistics needs to be extended with methods that estimate and quantify spatiotemporal variation and use it in spatiotemporal interpolation and stochastic simulation. This article briefly summarizes the main concepts of space–time geostatistics. Kriging in space and time can be done in much the same way as it is in a purely spatial setting. The main difficulties are in defining a realistic stochastic model that is assumed to have generated data and in characterizing and estimating the space–time correlation of that model. This article uses a model-based geostatistical approach to characterize space–time variability. The space–time variable of interest is treated as a sum of independent stationary spatial, temporal, and spatiotemporal components, which leads to a sum-metric space–time variogram model. Methods are illustrated with a case study of space–time interpolation of monthly averages of detected background radiation for a 5-year period in four German states.”
GIS and Science 
Geostatistics is an invalid variant of mathematical statistics. It assumes spatial dependence between measured values in ordered sets rather than verifies spatial dependence by applying Fisher’s F-test to the variance of the set and the first variance term of the ordered set. Geostatistics interpolates by kriging, derives voodoo variances, and ignores the concept of degrees of freedom. Google my name, read my blogs, and study who lost what, when, where, and why.