ABSTRACT
One of the most common goals of geostatistical analysis is prediction at unobserved locations and there are multiple strategies by which to accomplish spatial prediction; two popular choices are kriging and Bayesian predictive process models. There is an extensive amount of research on these two methods and it is not possible to report all of them here, but we encourage readers to see [5], Chapter 3 for kriging and Chapters 4 and 5 of [2] for an introduction on Bayesian methods. Kriging is derived as the best linear unbiased predictor by minimizing the mean squared prediction error (MSPE), see [5], Chapter 3.
