Select to view content in your preferred language

Why exporting EBK 3D to voxel layer change orginal vaules

199
2
Jump to solution
08-23-2024 08:06 AM
JV_
by
Emerging Contributor

I have 3D data representing measured soil parameters, and when I performed EBK 3D in ArcGIS Pro, the predicted values maintained the correct range, with the minimum and maximum values consistent with my original data.

However, when I export the results to a voxel layer and then import them as a Geostatistical Analyst (GA) layer to create sections and better visualize the model, I notice that the range of the data becomes narrower. Both the minimum and maximum values are lower than those in the original dataset.

I’m concerned that the interpolation process might be altering the true values of the observed data instead of preserving them exactly as they are. This is particularly important in my case, as it's crucial that the original data values are retained at the known locations.

Could someone explain how this process works and why these changes might be occurring?

Best,

0 Kudos
1 Solution

Accepted Solutions
EricKrause
Esri Regular Contributor

Hi @JV_,

EBK 3D is not an exact interpolation method, meaning that the range predicted values is usually more narrow than the range of the original values.  This is also true for all kriging methods, 2D and 3D.  The difference in the Symbology range is because geostatistical layers build their class breaks based on the input data values, and voxel layers (or rasters in 2D) build their symbology on the actual values of the voxel/raster.  Usually, the range of the output will be more narrow than the range of the input (called "smoothing").

For a longer explanation, please see this old blog post.  It is from ArcMap and talks about rasters, but the reasoning is the same as EBK 3D and the range of values of the voxel layers:

https://www.esri.com/arcgis-blog/products/product/analytics/understanding-geostatistical-analyst-lay...

If you have access to ArcGIS Pro 3.2 or later, you can use the IDW 3D tool which will honor the minimum and maximum of the original data.  The range of the voxel layer will still change a small amount because the calculations are made at the 3D center of every voxel, so unless you smallest/largest values align with the exact center of a voxel, the range will be slightly more narrow than the original points.

Please let me know if you have any other questions or need any clarifications.

-Eric Krause

View solution in original post

2 Replies
EricKrause
Esri Regular Contributor

Hi @JV_,

EBK 3D is not an exact interpolation method, meaning that the range predicted values is usually more narrow than the range of the original values.  This is also true for all kriging methods, 2D and 3D.  The difference in the Symbology range is because geostatistical layers build their class breaks based on the input data values, and voxel layers (or rasters in 2D) build their symbology on the actual values of the voxel/raster.  Usually, the range of the output will be more narrow than the range of the input (called "smoothing").

For a longer explanation, please see this old blog post.  It is from ArcMap and talks about rasters, but the reasoning is the same as EBK 3D and the range of values of the voxel layers:

https://www.esri.com/arcgis-blog/products/product/analytics/understanding-geostatistical-analyst-lay...

If you have access to ArcGIS Pro 3.2 or later, you can use the IDW 3D tool which will honor the minimum and maximum of the original data.  The range of the voxel layer will still change a small amount because the calculations are made at the 3D center of every voxel, so unless you smallest/largest values align with the exact center of a voxel, the range will be slightly more narrow than the original points.

Please let me know if you have any other questions or need any clarifications.

-Eric Krause

JV_
by
Emerging Contributor

Thank you! I didn't know about the IDW 3D, this was a great tip! And the article is also helpful! Thank you for answering

0 Kudos