Using the scallop data (description), first transform to normality as in the paper.
1) Compute sample variograms along the coordinate axes and the 45-degree lines (so four separate variograms). Fit a variogram model for the data (for example, as in the paper, or choosing your own parameterized function).
2) Use your covariance structure from above and pretend that it can be treated as known. Fit a kriging surface to the data. For this assignment, you can use the sample average as the truth and do simple kriging, or you can do ordinary kriging or Bayesian kriging. To make sure the anisotropy is handled correctly, it would be good for you to write your own code to do this.