Intersect fails but Pairwise Intersect succeeds

[GabrielMarcus1]

I have a precipitation layer from the NWS that I have run through a process to "flatten" the overlapping polygons through erase and merge. When I try to intersect the finished polygon layer with a point layer, it fails. 

Error messages below:

Reading Features...
WARNING 000117: Warning empty output generated.
ERROR 999999: Something unexpected caused the tool to fail. Please refer to the error help for potential solutions, or contact Esri Technical Support http://esriurl.com/support.
The table was not found.
The table was not found. [LocsJan19Wnd_Intersect]
Invalid Topology [Topoengine error (-40917539, 2046).]
Failed to execute (Intersect).
Failed at Tuesday, January 08, 2019 12:56:58 PM (Elapsed Time: 1.43 seconds)

I've tried

• checking and repairing geometry
• exporting feature and running intersect on the exported file
• exporting it to a shp file and running intersect
• closing and reopening the project
• rebooting
• running my script in an entirely new project

all with the same results. 

Pairwise intersect succeeds. For my use, intersect and pairwise intersect should produce the same results so it doesn't matter. Does anyone know why I would be getting these error messages when running intersect?

[Robert_LeClair]

Gabriel - in researching the WARNING 000117 error message, I found a few things that would cause the error.

• The data being intersected was not on a local disk rather an external disk drive.
• How many layers are being intersected?  10?  20?  50?  If many layers are being interested, consider using the Merge GP tool to merge them into one layer for the Intersect GP tool.
• Sometimes the data is bad and can be fixed using the Run Check Geometry GP tool.  Based upon the output, resolve the error(s) and then run the Intersect GP tool.
• What type of geodatabase are you using - single or multiuser?  There may be a related BUG.
• Were there any feature selected when you ran the Intersect GP tool?

I have not used Intersect vs. Pairwise Intersect before but there is a good description in the Help that compares the two GP tools.  In reviewing the Help, I don't think it's common for the Intersect and Pairwise Intersect to have the same number of features in the output.  Pairwise will have considerably less.

Hope this helps!

[GabrielMarcus1]

Thanks for your reply Robert. 

In order of your bullet points:

• All data are on local disk
• 2 layers are being intersected. A point and polygon layer.
• I mentioned using the checking and repairing geometry tool as my first step. I used both the ESRI and OGC methods. The check geometry tool does not output anything in the output table with either method. 
• Single user GDB. I'm the only licensed ArcGIS user in my org.
• No features were selected. 

I reviewed the help just now. I'm intersecting polygons and points. The number of output points are what I expected. It's the same number of points that I get when clipping the point layer with the polygon layer. 

[DanPatterson_Retired]

in your case, a very small simple sample would be useful.

compare spatial join, union, clip, intersect and pairwise intersect (and other overlay tools

Assume in all cases that the number of points returned is the same (plausible in many circumstances)

The choice one makes should be controlled by 

• the desired geometry output... in your case points
• the contents of the table that is produced... ie clip adds nothing, intersect adds more

If you switch geometry types to polygon and polygon, my above examples will become evident as to the difference that various selection and overlay methods produce in terms of the geometry outputs (intersect can produce way more than pairwise for example), but equally important (or more in some cases) is the difference in the tabular outputs.

When you get errors, it could be that geometry errors/inconsistencies occur during the process.  A point and polygon intersect would be unlikely to produce a geometry error, but higher geometries with other higher geometries could (ie self-intersecting polygons).

Checking and repairing geometry is always a good step.

If you get a 'failure', the next step is to subsample your inputs and run the process again.  If it succeeds, it could be the unsampled geometry areas cause the errors... switch the selection, repeat the process and run again,  if it fails, then it something to do with the geometries in the area... if it passes, then the source of the error is higher up the food chain.

You have done a good job of ruling out the obvious errors... except you didn't report the coordinate system being used and whether both files have the same coordinate system... check that.

You have found a solution which you can confirm works, you can check the 'selection' using the other methods, if they all pass except 'intersect', leave it for a rainy day