Alternative to Environmental Settings in Geostatistical wizards?

05-01-2020 08:40 AM
New Contributor III

Problem: My study area is a simple rectangular polygon in which the sampled points are spread inside. The extent of this polygon is obviously greater than the extent of the bounding box of input points. Now, the Geostatistical Wizard does not honor the (extent) environment settings. Also, Kriging is not given as an independent tool in the GA toolbox which can be supplied with the extent environment settings. As much as I know, only the wizard can do together (a) data transformations, (b) semivariogram modelling, and (c) neighbourhood parameterization, but it won't produce the result in the entire extent of the polygon!

Alternative adopted: Added four points at the four corners of the study area polygon. I updated their z-values with the mean of the sampled points. This causes the wizard to produce output in the exact extent of the study area polygon.  The interpolation in the new 'extra region' would have used data from the sampled points and these 4 extra points. 

Discussion: Is the above method a good alternative to the environment settings problem? I believe that a lot of users would come across this problem because a lot of times the output extent would be bigger than the sampling extent. Regarding the use of spatial analyst kriging tool, there's the only the option of simple and ordinary kriging and the tool is less flexible than the GA one. Is there a workflow of tools available to perform the above kriging job? Please enrich my knowledge. 

0 Kudos
2 Replies
Esri Regular Contributor

Hi Bankim,

Please see my response here: 

Using the Create Geostatistical Layer tool will allow you to change the Extent of any geostatistical layer.  That post was about Diffusion Interpolation With Barriers, but the same thing works with every type of Kriging.


0 Kudos
New Contributor III

Hello Mr. Krause, 

The tool works as expected! One thing very interesting about the new GA Layer was that the cross-validated (CV) statistics remain the same, since I was expecting that the added extent would have lowered the quality of the overall results. BUT, the statistics remained the same. In my opinion, this is because no measurement changed for the CV statistics calculation. So, even if I perform validation by sub-setting the points, no result report is going to change since the number of points, their measured values and their predicted values, all remain the same. Only new predictions in the added region are added in the new GA layer.

Please correct me if my understanding of this concept is not correct. 

0 Kudos