I'm getting different results when using the ZonalStatisticsAsTable tool in 10.4.1 when compared to earlier versions (10.2 and 10.3).

ArcGIS 10.2:
enter image description here

ArcGIS 10.3:
enter image description here

But in ArcGIS 10.4.1, the same data produces this table as a result:
enter image description here

For each test, I used the same statistic type and checked the option to ignore NODATA in the calculation.

enter image description here

Has the earlier versions been incorrect, and 10.4.1 is correct? Or is 10.4.1 incorrect?

I have a set of reproducible test data here:

  • 2
  • Duplicate? Seems to me that it is the exact opposite of that question. It would appear that the bug IS present in 10.4.1...NOT fixed in 10.4 as asserted in the first line of the referenced question: gis.stackexchange.com/questions/157766/…. – Jay Cummins Sep 22 '16 at 12:06
  • Also...if this is a difference in 10.4 vs 10.4.1...my question does already differentiate between the two. Wasn't sure if the gis.se tagging should include the patch level. So I didn't create one. – Jay Cummins Sep 22 '16 at 12:09
  • I can't find the meta post at the moment, but I believe that we decided to not tag versions past the major release version (also evidenced by the lack of patch level tags!). – Evil Genius Sep 22 '16 at 13:12
  • 2
    That's what I thought--patch level would make a real mess of the tags. And the 'Ignore NoData in Calculations' didn't change the behavior. – Jay Cummins Sep 22 '16 at 15:15

I was experiencing the same issue as you. I wrote a python script using arcpy on version 10.2.2, which worked fine. But, when I upgraded to 10.4.1, I was getting MEAN values returned from arcpy.sa.ZonalStatisticsAsTable() that were way out of the value range of the input value raster.

For me, the fix was to stop using "OBJECTID" as the zone field. Looks like you are also calling "FID" as your zone field.

Try using a different field than OID, FID, or OBJECTID to delineate your zones.


I opened a support incident w/Esri and confirmed this to be an existing bug for 10.4:
BUG-000096325 - The 'Zonal Statistics as Table' tool from the Spatial Analyst toolbox shows different values in ArcMap 10.4 as compared to ArcMap 10.3.1.

Since I was calling zonal statistics from a python script, my current workaround is to use rasterstats instead which works well, unless you are executing in an ArcSOC.exe process (see How to resolve "'NoneType' object is not callable" TypeError when calling rasterstats.zonal_stats from ArcPy script?).

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.