Latest Contributions by EricKrause

‎02-10-2014

To do this, you would have to come up with some kind of distribution for root-mean-square crossvalidation errors. Unfortunately, we're not aware of any valid way to do this (or if it's even possible). If you ever come across a methodology for this, please let us know.

‎02-03-2014

Do you have access to Geostatistical Analyst? It can only be done in that extension.

‎01-14-2014

It's hard to answer this question without looking at your data, but your best bet is to use Empirical Bayesian Kriging if you aren't sure which type of kriging to use. It's available in the Geostatistical Wizard and as a geoprocessing tool.

‎01-03-2014

Here is the response from the same developer, along with a zipped attachment of some Mathematica code, test data, and graphics: The change in sign is not a mistyping. This is to do bias correction. I wrote a simple test in Mathematica and the result for quantile map matches. Attached the test: Code.txt Source code to simulate data with Exponential covariance function (Exponent(-3*h)). Nugget=0 Sill=1 Range=1 Quantile=0.9 The prediction done in point (0,0) . Remark. This code is only for testing. OrdinaryKrigingQuantilePrediction.png �?? Wizard first page. SemivariogramParameters.png �?? Wizard page of semivariogram parameters. QuantilePrediction.png �?? All data are used as neighbors. Quantile prediction at point (0,0). TestData.* - Source data in shapefile and txt formats. Result in test code 2.15823 Result in Wizard 2.1582296949642736 If you have any more questions, it would be easier if you email me directly at [email protected] and I can put you in direct contact with the developer.

‎12-30-2013

The formula for quantiles contains a bias correction. I'm attaching the response from a developer as a pdf. It contains the more general quantile formula for universal kriging with a transformation.

‎12-26-2013

Hi Suzann, Does your input data have a join? If so, try using Copy Features to persist the join, then recreate the geostatistical layer and try again to export to raster. If that doesn't work (or you didn't have a join), try running GA Layer to Contour and let me know if it runs correctly. There are a few other things we can try, but hopefully this is enough to figure out the problem.

‎12-23-2013

The assumption is that there is an underlying smooth surface for the phenomenon. The surface is assumed to be a second-order stationary Gaussian field. The values in the polygons are assumed to be the average of the underlying Gaussian surface within that polygon. Areal interpolation is the process of estimating the underlying Gaussian field from the averages in an arbitrary set of polygons. Regarding how important the Gaussian assumption is, this is difficult to answer. Kriging theory assumes a Gaussian distribution in order for the predictions to be the "best linear unbiased predictions." If the data is not Gaussian, this property no longer holds. Kriging is known to be fairly robust to non-Gaussian distributions, but none of its attractive statistical properties hold if the data is not Gaussian. I'll actually change that documentation. Saying areal interpolation "requires" a Gaussian distribution is too strong of a statement. Instead, it should say that the kriging equations assume a Gaussian distribution.

‎12-09-2013

This looks like an Extent issue. On the tool, go to Environments, and specify a larger extent. By default, the raster will only get created in the rectangular extent of the input points, so any points near the border will not display the entire circle.

‎12-09-2013

Sorry, it is not implemented in Geostatistical Analyst or elsewhere in ArcGIS. If this is a useful functionality that you think should be in the software, you should recommend it on the ideas site.

‎12-06-2013

We'd prefer to leave these attachments up in case someone else has this problem.

‎12-04-2013

The default major/minor semiaxes are calculated as one quarter of the diagonal extent of the input points (because we also use minimum and maximum neighbors, this parameter is not very important). The default cell size is calculated as the smaller of the width and height of the input data extent, divided by 250. You can expose these as parameters in your script tool so that the defaults can be overridden. About the kernel parameter, you are correct that it is optimized to find the minimum root-mean-square crossvalidation error, treating all other parameters as fixed. When you change a different parameter, you will get a new optimal kernel parameter that reflects the change in the other parameter. Empirical Bayesian Kriging does variography automatically, which is why it is available as a geoprocessing tool. No interactive variography is required, and we suggest using EBK for automation of kriging. It is easier to automate, and it is more trustworthy when run blindly. If you still want to automate kriging (other than EBK) as it is implemented in the Geostatistical Wizard, you should start by reading this topic: http://forums.arcgis.com/threads/92508-Need-Help-Automation-of-Kriging-using-Model-builder-or-python-Reg. It's quite long, and a lot of it won't apply to you, but all the information about automating kriging is in there. The only difference is that if you're using 10.1, you won't be able to use the GeostatisticalDatasets arcpy class. You should consider upgrading to make this workflow easier. If you don't upgrade, you'll have to pass the new datasets as comma- and semicolon-delineated strings.

‎12-03-2013

Everything you said is correct. This has always been a known problem for simple and ordinary kriging; you have to estimate the semivariogram from the data, but then you have to assume that your estimation is completely correct. In practice, this means that standard errors will almost always be underestimated. This is one of several problems that we addressed with Empirical Bayesian Kriging in ArcGIS 10.1. Instead of estimating a single semivariogram and assuming it is correct, EBK simulates many semivariograms, so you end up with an entire spectrum that we weight by likelihood. By accounting for some uncertainty in the estimation of the semivariogram, this weighted spectrum does a much better job of estimating the covariance structure than relying on a single semivariogram. If you work with some data using ordinary, simple, and empirical Bayesian kriging, you'll notice that EBK usually gives larger standard errors than the others. This might seem like a disadvantage, but the larger standard errors will usually be more accurate. For example, if you use simple or ordinary kriging and make 90% confidence intervals, it will probably only capture 75% of the data, whereas 90% confidence intervals from EBK should capture closer to 90% of the data.

‎11-25-2013

We use a simple kurtosis formula in Geostatistical Analyst. The formula is the fourth moment divided by the square of the second moment. We do not correct for small samples or calculate excess kurtosis. It appears that the KURT function in Excel calculates excess kurtosis and corrects for small sample sizes. In general, the difference between these two calculations should be approximately 3. You can see from your data that if you add 3 to the Excel values, you will get close to the ArcGIS values (the remaining difference is due to small sample corrections in Excel).

‎11-20-2013

The append all fields parameter simply copies fields from the input polygons. For example, suppose you're predicting to a polygon feature class representing lakes, and one of the fields of the feature class is the name of the lake. If you choose to append all fields, the field with the name will be copied to the output. If unchecked, all fields will be dropped. Whether you append or not will not affect the predicted values in the polygons.

‎10-31-2013

We have special models for count and proportion data, but if your data is a continuous variable (like average temperature or median age), you may end up with negative predictions. You may be able to change the semivariogram parameters to get only positive predictions, but it's difficult to say exactly what you should change because everything is so data-dependent.

Online Status	Offline
Date Last Visited	‎02-25-2026 06:39 PM

My Ideas

Latest Contributions by EricKrause

Re: Comparison of Root Mean Square Error

Re: Kriging input with uncertainties

Re: Kriging to estimate vegetation cover - simple, ordinary or universal?

Re: How are kriging quantile maps calculated?

Re: How are kriging quantile maps calculated?

Re: Geostatistical layer to raster empty

Re: Areal Interpolation and data normaility.

Re: point density area expansion

Re: plurigausssan simulation method

Re: where is the view settings panel in the geostatistical wizard for semivariogram?

Re: Using certain Geostatistical Wizard from a Python script?

Re: Universal Kriging with External Drift

Re: What is wrong with kurtosis in Geostatistical Analyst?

Re: Areal interpolation reference/predicting to polygons 10.2

Re: Negative values after areal interpolation

Re: Why exporting EBK 3D to voxel layer change org...

Re: Access geostatistical layer created using ArcP...

Re: Why exporting EBK 3D to voxel layer change org...

Re: K-Bessel Semivariogram Equation

Re: Kriging Model Types

R-ArcGIS