Gaussian process (GP) models encounter computational difficulties with large spatial data sets, because the models’ computational complexity grows cubically with the sample size n. Although a ...