2

I have a problem with creating Thiessen polygons for a set of points. I am creating them with Create Thiessen Polygons tool in ArcGIS 10.1). I am able to create them in a few steps (for parts of the data) but I would like to create it at once.
The error I get is:

ERROR 000537 error in converting points to Voronoi Polygons

It doesn't say much, when click at the error in arcgis help I found this message:

Description A TIN surface is created from input points and is used to generate the output Thiessen (Voronoi) polygons.

Solution Check the input point features and make sure that features exist in this particular feature class. As well, make sure that if the extent environment is set, the input point features do not fall outside their coordinate range.

But if it works "partially", where the problem may be?

  • You may have some errant points that fall out of the tool's extent capabilities, or there may be some other issue with the feature class. You might look at the feature class and see if there are any obvious issues, and/or attempt to repair geometry, or export the feature class to see if that fixes something. Lots of possibilities. – Baltok Jun 20 '14 at 13:06
  • Hi, thanks a lot for the answer. I will try it. I also thought there may be something wrong "between" some objects as it works when split in parts but does not want to work as a whole set :(. What kind of other obvious issues do you mean? – Izabela Jun 20 '14 at 14:07
  • Ensure that you don't have duplicate points. – user681 Jun 20 '14 at 18:33
  • Are you in a projected coordinate system or a geographic one? Per the help file, "This tool may produce unexpected results with data in a geographic coordinate system since the Delaunay triangulation method used by the tool works best with data in a projected coordinate system." – Chris W Jun 20 '14 at 19:04
  • 1
    @PolyGeo I can try, but it seems to me she really worked it out on her own (I'm still not clear on why the points didn't have a projection) and should combine all of her comments into an answer, so I'll delay until tomorrow before doing so to give her the opportunity. – Chris W Jun 25 '14 at 1:01
1

Several comments have good suggestions to check for duplicate points stacked atop each other and points with no or bad geometry. However the statement in the question that the tool worked partially cued me to take a look at the help file where I noticed the following:

This tool may produce unexpected results with data in a geographic coordinate system since the Delaunay triangulation method used by the tool works best with data in a projected coordinate system.

Based on that I questioned which coordinate system you were working in. In your investigations it appears you found that your points weren't projected at all which would probably lead to similar behavior. Apparently getting the points into a proper projection solved the issue with the tool.


I do want to point out a couple of things regarding projections in ArcGIS, just for clarification:

  • From your comments, the original points were in shapefile format. There should have been a .prj file among the files comprising the shapefile. If it was missing or damaged, that would lead to the issue and warning you got when you tried to add them to a map.
  • The Dataframe (with a default name of Layers) can have a projection set in its properties. This can differ from the projection of the individual datasets added to the dataframe. ArcGIS reprojects those individual layers/datasets on the fly to whatever projection the dataframe is using if it can. It can't if there isn't one defined in the first place - it knows what to project to because of what you set the dataframe to, but it doesn't know what to project from because the data doesn't say. It's like saying 'can you give me 532 in meters?' 532 what?
  • Projections also involve Datums, and ArcGIS handles them in separate steps. If the datum between two projections doesn't match, you must specify an appropriate transformation to use when converting between them or your points may end up in entirely the wrong place. You can't just pick the projection you want to use, you must specify both (though it's in the same dialog box).
  • Geodatabases themselves, be they file or personal, don't have projections/coordinate systems. It's the individual feature classes (or feature datasets) inside the geodatabase that do, and you can have several feature classes with different projections/coordinate systems in the same geodatabase.
  • You should be cautious about just 'defining' the CRS of an unknown dataset (which is essentially what you did with the export). If you don't know what it was really in, you have to check that it's coming in 'close enough' to where it really should be based on other known reference data. You've given it a projection now, but it may not be the right one (just the one you want to use). The difference may be fractions of a meter or hundreds of kilometers. For your purposes it may not matter at all where the points are except relative to each other, or your accuracy requirements may deem whatever projection you used to be 'close enough'.

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.