Spatial Scale Problems and Geostatistical Solutions

Atkinson and Tate make a good point. I only wish I could find it. Their extensive use of mathematics is daunting, but a necessary evil when understanding what goes on under the hood of ArcGIS. With no personal experience in the matter, a quick Google search yielded that Variograms are the same, if not similar, to kriging, and require significant input from the user. Correct me if I’m wrong.

GIScience has managed to produce a slew of tools that produce right answers. That is to say, there is only one possible answer. The more complex processes, like the interpolation methods outlined by Atkinson and Tate, reveal that there sometimes must be a best answer. At which point it is the responsibility of the user to justify their reasoning behind choosing 10 lags instead of 5. At which point, it becomes a case specific example.

What makes me curious is, is there a right answer? Is it possible to create a set of parameters, possibly for an arbitrary set of scales, that would optimize the up-scaling and kriging process in all fields of use?
Written in 2000, there has been more than a decade for someone to answer the question and implement it in GIScience. As of 2013, there is no right answer, but there is a significant amount of mathematics to back it up.

In an ideal world, if the research field dedicated data mining and geographic knowledge discovery is successful, there may eventually be no need for interpolation as it is replaced by the overwhelming wave of high resolution, universal, data sets.

AMac

Tags:

Comments are closed.