Latest Contributions by EricKrause

The Lag Size is shared between the primary and secondary variables in cokriging.  You cannot set different values for each of them.  However, the lag size is only used in order to estimate the semivariogram parameters (range, nugget, partial sill, etc).  The actual interpolation does not directly depend on the lag size.  This means that you can experiment with lag size for the primary variable, then note down all of the parameters like range, nugget, sill for a lag size of your choosing.  Then switch to the secondary variable and find semivariogram parameters using a different lag size. Note down those parameters, then manually type them all simultaneously.  ... View more

Hi Alex, If I'm understanding correctly, you have points with a measured elevation, and you want to interpolate that elevation.  In this case, you should actually just use regular 2D Empirical Bayesian Kriging and provide the elevation as the "z-field" parameter. EBK3D is used when the location of the points are in 3D, and each 3D point has a measured value.  For example, you can interpolate oxygen measurements taken at various locations and depths in the ocean into a full 3D model that predicts oxygen throughout the ocean (this particular example will be shown in an upcoming LearnGIS lesson). Let me know if that isn't clear. -Eric ... View more

Also, if you really want to use EBK with those points in ArcMap or ArcGIS Pro, you can use the Coincident Points geoprocessing environment to force the tool to use all the points and not perform any averaging.  Set the environment to INCLUDE_ALL in order to do this.  However, you should keep an eye out for artifacts in the output because including all points with a relatively few number of sample locations will very frequently cause instability and strange behavior in kriging methods. ... View more

That explains it; I should have thought of this earlier.  IDW and Kernel Interpolation With Barriers both only require 3 unique locations to run.  Interpolate Points actually uses Empirical Bayesian Kriging (EBK) to perform the interpolation, and that method requires at least 10 unique locations.  You should get an analogous Not Enough Data error if you try to use the EBK tool with this data. More data is needed in EBK than in IDW and Kernel Interpolation because many more parameters must be estimated in EBK, and more data is required in order to do that. ... View more

It actually needs 10 unique locations with measured values.  All measured values at the same location get averaged into a single value. The same thing should happen in Desktop though, unless you happened to change the default value of the "Coincident Points" environmental setting.  Which tool are you running in Desktop that can execute successfully with these points? ... View more

Hi Elisios, I'm happy to take a look, but I will need a layer package rather than a layer file.  A package will also send the underlying data for the layer, which I need in order to view it.  Only do this is the data is not sensitive and you are ok with the public being able to download it.  Thanks, Eric ... View more

Hi Musfira, The number of points that come out of the Export Result button should be equal to the number of rows in the cross validation table in the Geostatistical Wizard.  I don't know any reason they would be different.  I can also see in your table that all Source ID values are -1, which is also strange. As Steve asked, what format are the points that you used in the wizard?  Shapefile, file geodatabase, something else?  Regardless, you should try clicking Finish then OK in the wizard to create a geostatistical layer in your map.  Then use the Cross Validation geoprocessing tool with the layer.  The optional feature class output from the tool produces the same feature class as the Export button in the Wizard.  Using the geoprocessing tool instead of the Export button may give what you are expecting. ... View more

We generally recommend using cross validation and validation to assess the accuracy of kriging models.  ... View more

A formula for EBK could probably be written, but it would be so complicated that no one would be able to understand it.  To understand EBK, you need to understand the procedure that generates the predictions.  There is more information in the documentation on Empirical Bayesian Kriging, but the general procedure is: Subset the input data. In each subset, estimate a semivariogram using maximum likelihood.  This semivariogram will have a relatively simple formula and will serve as the seed for simulations later. Use the estimated semivariogram to simulate new values at each of the input point locations. Estimate a new semivariogram from the simulated data values. Simulate a new set of data from the seed semivariogram. Repeat steps 4 and 5 many times (by default, 100 times).  This results in an empirical distribution of semivariograms in each subset. Use the empirical distribution in each subset as a Bayesian prior and calculate the posterior distribution. Finally, when making a prediction at a new location, identify the neighbors based on the search neighborhood and merge the posterior distributions from the subset associated with each neighbor.  The distributions will be weighted proportionally by the number of neighbors from that subset. The only difference for EBK Regression Prediction is that in step 2, you will estimate a coefficient for the explanatory variables that you provide in addition to the semivariogram parameters. ... View more

The export options allow you to export the blue crosses and red points that you see in the semivariogram pane.  That is one way to compare semivariograms, but there are others.  There are several methods, for example, to extract the semivariogram parameters (range, nugget, sill, etc) as well as cross-validation and validation statistics.  ... View more

Hi Eliseos, The only available options for exporting the semivariogram are available in ArcMap.  On the semivariogram page, there is a section to the right of the semivariogram map with options to export the variography.  These options only appear in the user interface of the Geostatistical Wizard, so the process unfortunately cannot be automated. If these options are something you would find useful in ArcGIS Pro, you should put your idea up on the ArcGIS Ideas site.  -Eric ... View more

All interpolation methods require a single measured value at every location. A few interpolation methods (EBK Regression Prediction and Gaussian Geostatistical Simulations) allow the use of a measurement error field in addition to a data field.  The measurement error field can be used to indicate uncertainty in the measured value.  However, the value of the measurement error field should correspond to one standard deviation of the measurement error rather than a confidence interval.  There may be a way to convert your confidence intervals into standard deviations if you know how the confidence intervals were constructed (in many cases, they are built from standard deviations).  ... View more

I apologize, I posted some outdated information.  I've edited my post above. ... View more

Hi William, I do not think this issue is related to your data; I was able to run Interpolate Points with your attached data without any errors. Instead, the issue is almost certainly related to licensing.  Interpolate Points requires an Advanced license for ArcGIS Enterprise in order to use the Geostatistical Analyst extension (which only comes with an Advanced license).  If you have a Standard license, Interpolate Points will fail on execution, and this is probably what is happening for you.  This will also happen for the Calculate Density tool because it requires the Spatial Analyst extension.  Can you confirm which version of ArcGIS Enterprise you are using and your license level? Starting in ArcGIS 10.6, Interpolate Points and Calculate Density will no longer appear in the tool list if you have a Standard license. We apologize for the poor communication of this problem, but I hope this clears up why the tool is failing.  ... View more