topic Re: Union results in new polygons in ArcGIS Pro Questions

Union results in new polygons

PhilipWhite1 — Mon, 27 Apr 2020 21:20:03 GMT

Hello,

I'm consulting a student who is running into a problem with a Union operation, and I've been struggling to figure out why the results look as they do.

My student has about 100 vector layers they're attempting to Union. The output file is a lot of strange new features that were not present in the any of the original input layers. I can't figure out why this is happening.

Troubleshooting this, I started running a Union on just 2 layers at a time. Discovered this was happenning here as well.

For example, here are the two input layers displayed on top of one another (participant2, participant3):

I run the Union on these two, and I would expect the shapes of all the features/new features to appear identical to the input layers, but they do not. See the new Union2_3 layer in red, and note the red ellipse at center that is now a different geometry than the input:

Why is this happening?

I ran the Check Geometry tool to see if there was something wrong with the input data, but it doesn't detect any geometry errors.

Why would new features get created? Or why would the geometry of the original not be preserved in the new output Union?

FYI: this operation worked as expected in QGIS.

Thanks,

Phil

Re: Union results in new polygons

DanPatterson_Retired — Mon, 27 Apr 2020 22:46:01 GMT

map and geometry of the featureclasses all in the same coordinate system?

If you suspect this is a bug, then reporting it to Tech Support would be useful since you seem to have a simple use case.

Re: Union results in new polygons

PhilipWhite1 — Mon, 27 Apr 2020 23:05:21 GMT

Hi Dan,

Yep. All are in the same coordinate system. Also, all input feature classes are stored in a gdb feature data set.

I'll also note that the student I'm working with was using ArcMap, but I'm getting the same results in Pro.

Thanks,

Phil

Re: Union results in new polygons

DanPatterson_Retired — Mon, 27 Apr 2020 23:18:27 GMT

Shifting isn't normal... I suspected a datum shift of the output, but... could be the ellipse.

See if it can be replicated with other simpler geometric shapes.

It could be unique to ellipses since "may" be being defined by the long and short axes and a central point rather than an n-gon of points (can't remember offhand)and any small 'issue' with resolution or something would possibly be compounded.

Worth a look. But I would "Tech Support" for a definitive answer.