As part of a practical exercise in the MRes course, we were asked to create a map of London with the average GCSE scores. Since we needed a value for every pixel, and the data are about wards, we used a method of interpolation, and specifically the Kriging method. Also, I went one step further and used Kriging with both R (which is can be integrated in GRASS as well), by using the gstat package and ArcMap.Kriging is an interpolation method and can be used for “describing the surface” of an area with the specific sample values. It’s function is based on assuming the distance or direction between sample points, which finally show a spatial correlation (Dennett. A, 2013). It is decided that Kriging is the best interpolation method to be used in this specific case, since it comes with smoother results. Also, after using all the possible different methods, in both softwares the variograms used for applying Kriging were created with the exponential models, since they were the best to be used, statistically proven.
The same options were used in both programs, because I would like to compare the results afterwards. The proper semivariogram was created, which estimates the values of the London “grid” and then I created the surface.
The darker areas show higher average GCSE scores.
As anyone can notice in the results, even the fact that they are the same, the output is a bit different. R provides a smoother optical output, but ArcMap categorises the results which can be more practical in analysis afterwards.