topic Geostatistical wizard - Cokriging - model equations under variogram in ArcGIS GeoStatistical Analyst Questions

Geostatistical wizard - Cokriging - model equations under variogram

N_G_ — Fri, 02 Dec 2011 10:59:29 GMT

I have to use ordinary cokriging (variables are precipitations and elevation).
I build an automatic model (in model builder) to perform interpolation.

However, when I build the .xml file (reference file) with the geostatistical wizard why the last equation noted when we fit variograms and Cov. aren�??t the same when we open again the model in the method properties after computation?

I don�??t understand why parameters of the first variable (first variogram) are automatically applied to the model (nugget and partial sill), other parameters are the same.

It�??s not possible to have specific parameters for each variable and so each variogram and to keep them as we want (in the model equation under the variogram) ?

Maybe there is something I don't understand with the use of several fit-models .....

For examples:
Manual fitting and Reopening of the method properties after computation: see model equations

Re: Geostatistical wizard - Cokriging - model equations under variogram

EricKrause — Fri, 02 Dec 2011 13:47:58 GMT

The first graphic is the semivariogram for variable 2 (elevation), and the second graphic is for variable 1 (precipitation). You can see this at the top-right of each graphic: "Between: Var2 - Var2" and "Between: Var1 - Var1".

The first graphic is pulling Nugget[1][1] and Partial Sill[1][1] from all models. Graphic 2 is pulling Nugget[0][0] and Partial Sill[0][0]. The cross-covariance (not pictured) between elevation and precipitation is from Nugget[0][1] and Partial Sill[0][1].

I'm not seeing any problems with the graphics you posted. Can you try to clarify your question?

Re: Geostatistical wizard - Cokriging - model equations under variogram

Anonymous User — Mon, 05 Dec 2011 07:12:25 GMT

Original User: Nagosa

Thanks for your answer.

However, what I want to know is: why, when we adjust variograms:
1- Precipitation variogram [Var 1 -Var 1];
2 - cross-variograms [Var1][Var 2];
3 - Elevation variogram [Var 2 - Var 2];
we have one type of model equation (equation under the variogram, manual_image.jpg), as follow:

Model:70658*Nugget+0*Spherical(129460,56953,90,0)+56868*Spherical(25056,25056,0,0)
+1748700*Gaussian(129460,93934,100,4)

[1]*Nugget+ partial Sill [Elevation, model1 ]*Spherical(major range,minor range,anisotropy direction)
+partial Sill [Elevation, model2 ]*Spherical(major range,minor range,anisotropy direction)
+partial Sill [Elevation, model3 ]*Gaussian(major range,minor range,anisotropy direction)

And after cross-validation and the record of the .xml model we have (when we come back to method properties in the co-kriging data) this equation and not the first shown above, i.e. (after_computation.jpg):

Model: 25,326*Nugget+0*Spherical(129460,56953,90,0)+182.42*Spherical(25056,25056,0,0)+61,062*Gaussiant(129460, 93934,100,4)

[0]*Nugget+ partial Sill [precipitation, model1 ]*Spherical(same major range,minor range,anisotropy direction)
+partial Sill [precipitation, model2 ]*Spherical( same major range,minor range,anisotropy direction)
+partial Sill [precipitation model3 ]*Gaussian( same major range,minor range,anisotropy direction)

Does the geostastical wizard used the first or the second model equation to compute data? It's not clear. If we want to keep our adjustment equation (the first): how can we do and is it possible?
That's one of the elements which are not clear for me.

I hope that explanations are more clear?

Thank you for your help.

Re: Geostatistical wizard - Cokriging - model equations under variogram

EricKrause — Mon, 05 Dec 2011 14:09:06 GMT

Does the geostastical wizard used the first or the second model equation to compute data?

It uses both. When you do cokriging, you have to model the primary variable (Var 1), the secondary variable (Var 2), and the cross-covariance between the primary and secondary variable (Var1 - Var 2). The cokriging equations require you to model all three processes to make predictions.