Ordinary Kriging Question

5409
10
04-10-2012 11:26 PM
RyanWhitehouse
New Contributor
Hello,

I'm a graduate student working on a project that involves well points, each associated with a measure of nitrate. I also have HUC10 watershed delineations, which I plan to use to breakdown the data later. I ran ordinary kriging through the Geostatistical Analyst using the nitrate well points and choose the Gaussian model. I ran into two problems here:

1) The kriging output did not automatically format itself as a raster dataset. Instead, I had to convert it, but by doing so it distorted the surface. Is this normal? I tried kriging via the toolbox, and it seemed to have converted it just fine. Does this mean I can't use the Geostatistical Analyst?

2) When I ran zonal statistics on the interpolated raster surface, using the HUC10 as the zone layer, the ArcToolBox would crash. This happens on multiple computers at different locations. Is there a way to get around this?

Basically, I'm trying to do a hotspot/coldspot analysis of the nitrates by HUC10 zones. But there are over 600 HUC10 zones.

Thanks.
0 Kudos
10 Replies
SteveLynch
Esri Regular Contributor
What version of ArcGIS are you using?

The Geostatistical Wizard does not output a raster, it creates a geostatistical layer which you can convert to raster, polylines or filled polygons.

If you did kriging via the toolbox it means that you used the Spatial Analyst (or 3D) version.

How large is the output raster?

Have you tried converting the HUC10 layer into raster and use that in ZonalStatistics?

- Steve
0 Kudos
RyanWhitehouse
New Contributor
The output raster covers the entirety of New York State. No, I have not tried converting the HUC10 into raster, but I will definitely try that.

Any idea why it distorts the kriging when I convert to a raster?
0 Kudos
EricKrause
Esri Regular Contributor
Any idea why it distorts the kriging when I convert to a raster?


This blog should help:
http://blogs.esri.com/esri/arcgis/2008/09/29/understanding-geostatistical-analyst-layers/

The main idea is that a geostatistical layer is really a mathematical model that takes an (x,y) location and returns a value (as opposed to a raster that has values stored in each cell).  A geostatistical layer doesn't "know" its own values until you ask it to calculate them.  The contours that you see in a geostatistical layer are made from a quick contouring algorithm: the model makes the fewest calculations that it can in order to draw roughly accurate contours.

When you export a geostatistical layer to a raster, all it is doing is calling the mathematical function at the center of every raster cell.  This obviously requires many more calculations than drawing quick contours, so the surfaces may appear a bit different (sometimes very different).
0 Kudos
FirdesYenilmez
New Contributor
Hi,


I am PhD student and I have the same problem. If you solve the problem, may you share the solution with me? Thank you very much. Firdes.



Hello,

I'm a graduate student working on a project that involves well points, each associated with a measure of nitrate. I also have HUC10 watershed delineations, which I plan to use to breakdown the data later. I ran ordinary kriging through the Geostatistical Analyst using the nitrate well points and choose the Gaussian model. I ran into two problems here:

1) The kriging output did not automatically format itself as a raster dataset. Instead, I had to convert it, but by doing so it distorted the surface. Is this normal? I tried kriging via the toolbox, and it seemed to have converted it just fine. Does this mean I can't use the Geostatistical Analyst?

2) When I ran zonal statistics on the interpolated raster surface, using the HUC10 as the zone layer, the ArcToolBox would crash. This happens on multiple computers at different locations. Is there a way to get around this?

Basically, I'm trying to do a hotspot/coldspot analysis of the nitrates by HUC10 zones. But there are over 600 HUC10 zones.

Thanks.
0 Kudos
JL1
by
New Contributor
I have a similar problem.

I created a bathymetry map of a lake using the geostatistical analyst (i.e. creating a GA layer). I re-classified the depths for my purpose and would now like to export the GA layer to a raster file (GA layer to Grid). When exporting I want the output cell size to be 1.
The interesting thing is that I did all this successfully in arcmap 9.3, yet in arcmap 10.0 it doesn't work.

Thanks for any kind of help.
0 Kudos
EricKrause
Esri Regular Contributor
What happens when you try to export the geostatistical layer to a raster?
0 Kudos
SteveLynch
Esri Regular Contributor
What is the default cell size that shows up in the dialog after you enter the ga layer into the control?

Please also state clearly what happens when you run the tool. I'm not sure what you mean by "I  have a similar problem."

Thanks
Steve
0 Kudos
JL1
by
New Contributor
So, I can run the GA (Kriging/CoKriging; default parameters worked best for my purpose) without any problems and do get a nice GA layer. This is what happens when I try to export the GA layer to a raster file:

1) Nothing happens (accepted default values in the GA Layer to Grid Window [output cell size: 1,573181523475E-05]): error message says that the output surface raster is <empty> (ERROR 999998: Unexpected Error), failed to execute.

2) When I choose to change the output cell size to 1 (which is in fact what I want), then a raster file is in fact produced. However, the high and low values are completely off (high: 3,40282e+038; highest value should be something around 4,2). When I try to classify (Properties, Symbology), I get the following error message: Failed to initialize classify renderer (possibly too many unique values).

Does this help you?
Thanks for your help.
0 Kudos
SteveLynch
Esri Regular Contributor
- I guess you are in a geographic coordinate system (i.e degrees decimal), not a good idea as 1 degree longtitude <> 1 degree latitude.
- default size cell produces an output raster of roughly 250 columns and 250 rows
- with you changing the cell size to 1 will mean that your output raster will have roughly 1 row and 1 column
- I notice that you are in a comma locale and using ArcGIS10, have you installed the latest Service Pack (SP5)?
- 3.4e38 is a NoData value, which basically means that your output raster is empty.

Regards
Steve
0 Kudos