Latest Contributions by EricKrause

I need some clarification for your question.  Are you talking about the Kernel Density tool in Spatial Analyst?  Or our ESDA tools?  Or Kernel Interpolation in Geostatistical Analyst? Can you link to the help topic that you're having trouble understanding? ... View more

We don't have anything like that in Geostatistical Analyst.  You might have more luck on the Geoprocessing forum, or maybe the ArcGIS Desktop - General forum. ... View more

You can also use Python's random number generator to get random subsets. Let's say you have 100 points in your feature class, and you want a subset of 70 random points.  Make a new Short Integer field in the feature class, then use Calculate Field.  Make sure you change from VB to Python, and use this code: arcgis.rand("Integer 1 100") Then sort the Attribute table by the random number, and select the first 70 records. ... View more

Universal kriging removes systematic trends in the data before kriging on the residuals.  Ordinary kriging assumes that all apparent trends are due to spatial autocorrelation (and thus aren't actually "trends"). http://help.arcgis.com/en/arcgisdesktop/10.0/help/index.html#//003100000048000000.htm http://help.arcgis.com/en/arcgisdesktop/10.0/help/index.html#//00310000003s000000.htm ... View more

Kriging is more for parameter estimation than making contours Can you explain what you mean here?  Which parameters are you talking about? If kriging is the optimal interpolator (under certain conditions), wouldn't it also be the optimal method for making contour maps? ... View more

When you manually set a lag size and number of lags, the software will never use a Major Range that is larger than the lag size multiplied by the number of lags (unless you manually specify a Major Range).  So, if the software is using (lag size x number of lags) for the Major Range, this indicates that the optimal Major Range is larger than the product of lag size and number of lags.  This means that your lag size is likely too small (or the number of lags is too low). If you use the "Optimize Model" button (available in ArcGIS 10), the software will calculate the Major Range (among other parameters) in order to minimize the Root-Mean-Square error (RMS).  Using this button is our recommendation for finding the best semivariogram parameters, including Major Range. ... View more

Constant trend removal either adds or subtracts a constant value to the whole surface (if it is done globally).  This value is the same everywhere in the surface.  A first-order trend removal will use the X,Y coordinates as predictors in a first-order linear model. Models: Constant: z = Intercept + Error First-order: z = Intercept + aX + bY + Error ...where z is the variable you are interpolating, and a,b are coefficients that are calculated using ordinary least-squares.  Kriging is then performed on the Error term.  When the trend removal is done globally (as in the Kriging tool in Spatial Analyst), the values of [Intercept, a, b] will be the same across the entire surface.  When you do a local trend removal (only in Geostatistical Analyst), the values of [Intercept, a, b] will be calculated locally using weighted least-squares, so the coefficients will change across the surface. ... View more

Model : 861.57*Nugget+8817.6*Tetraspherical(150000) 861.57 = Nugget 8817.6 = Partial Sill 150000 = Major Range In the Geostatistical Wizard's variography page, you can see that values correspond to the parameters on the right. There are various export options on the variography page in the Geostatistical Wizard.  You can output the red dots (binned values), the blue crosses (averaged values), and averaged values using geometrical intervals (you don't see these in the semivariogram graphic, but they are the values that are used in our weighted least-squares algorithm to fit semivariogram parameters).  The main reason to export is that you can graph the semivariogram in another program like Excel, where you have more control over the graphic (like changing colors, labels of axes, etc). You can also export the Semivariogram/Covariance map, either as a raster of weights or a raster of values (where the value is either the semivariance or the covariance, depending on whether "Variable" is set to Semivariogram or Covariance). ... View more

Even after changing the extent of the GPI layer, Map Tips still works for me.  Are you using ArcGIS 10? The default extent of any geostatistical layer is the rectangular extent of the input points.  However, you should think of a geostatistical layer as a function, rather than a final result.  It doesn't "know" its own values like a raster does.  To find the value at some (x,y) location, the layer has to calculate the result, whereas a raster has a saved value for each cell.  The geostatistical layer draws rough contour lines by making the fewest predictions it can to draw roughly accurate contours (this is why the surface often looks different after exporting to raster... this conversion simply entails making a prediction at the center of each raster cell).  Since the geostatistical layer is essentially a function that takes an (x,y) location and returns a predicted value, you can change the range of the (x,y) locations that it will accept.  ie, you can change the visible extent of the layer.  Changing the range of the x and y coordinates won't change the predicted value at any locations, so that is why the contour lines do not change when you change the extent.  Also, since the geostatistical layer does not have to store all of its values (instead, it stores a function of x,y), it essentially has infinite resolution (up to the precision of your computer); you can keep zooming in, and you will never see pixels like you will with rasters. ... View more

I'm sorry, but I can't reproduce your error.  Map Tips show the value of the GPI geostatistical layer when I hover over the map.  Please contact support services.  They should be able to help you resolve this issue. https://service.esri.com/index.cfm?fa=homepage.feedback.customer_serviceform ... View more

You're right, it doesn't print the GPI equation anywhere in the Wizard or the gp tool.  This is a good catch, and we may add that enhancement in the future.  Thanks for pointing it out. However, you can calculate the equation fairly easily.  GPI is just doing Ordinary Least Squares using the coordinates as predictor variables.  So, you can use the Ordinary Least Squares tool in the Spatial Statistics toolbox to calculate the coefficients.  If you're using a second-order polynomial, you'll need to calculate X, Y, X^2, Y^2, and X*Y and use them as predictor variables.  For even higher-order polynomials, you'll need to calculate all the high-order terms and interactions. The OLS tool only prints the first 6 decimal places of the coefficient, but the coefficient will often be smaller than 10^(-6) if you're in a small map unit (like meters), so it will just appear as 0.000000.  If the OLS tool doesn't print enough digits by default, there's another technique that will work: In the OLS tool, under "Additional Options" at the bottom of the tool, you can create a optional Coefficient Output Table as a dbf file.  After running the tool, add the dbf file to ArcMap, and open the table.  As shown in the attached graphic, right-click the "Coef" field and choose "Properties".  Click the "Numeric" button (highlighted in green in the graphic), and you can set the number of decimal places to as high as you need. Sorry for having to resort to a work-around, and thanks again for bringing this to our attention. ... View more

I'm not sure what you mean by "overlapping points."  Do you mean points that have the exact same x,y coordinate (coincident points)?  If that's what you mean, you should use the Collect Events gp tool in the Spatial Statistics toolbox. http://help.arcgis.com/en/arcgisdesktop/10.0/help/index.html#/Collect_Events/005p0000003s000000/ ... View more

Yes, the model you just described is Universal kriging with first-order trend removal, as implemented in ArcGIS (other implementations allow covariates other than just the x,y coordinates).  In the Kriging tool in Spatial Analyst, a global first- or second-order trend is fit to the x,y coordinates.  In Geostatistical Analyst, you can do local trend removal (ie, the coefficients of the x,y coordinates in the model will change across the surface). ... View more

Also keep in mind that the optimization is done after you do any transformations, trend removal, or declustering.  Optimizing the model won't correct any mistakes that you may have made during those previous steps. ... View more

The optimize button finds the parameters that result in the lowest root-mean-square error (RMS).  In kriging, for a given semivariogram type that you choose (Stable, Exponential, etc), it will optimize the range, nugget, and partial sill to find the lowest RMS.  If anisotropy is turned on, it will also find the optimal direction and major/minor semiaxes. ... View more