4

It's quite common to find these topology errors when dissolving data:

Census tracts dissolved using ST_Union on PostGIS Census tracts dissolved by neighborhood using ST_Union (PostGIS)

It's quite simple to understand the concept of a polygon donut hole in these cases: when the geometry really had a gap. But when your resulting holes are lines or points, my logical thinking get stuck.

What exactly makes this happen or not? If the polygons share the same boundary lines without gaps, how does it create a line hole? A point feature within this 'line hole' wouldn't intersect with the polygon feature?

I'm looking for a more detailed explanation, rather than the general and common "This happens when working with bad quality data"

  • 1
    If you zoom WAY in on them, I think you'll find that those aren't points and lines at all, they're very slender gaps in between your polygons. Just like your "donuts holes", just shaped differently. – Dan C Nov 28 '16 at 17:14
  • Indeed, most of the time it's that. But in this case I can zoom in as much as I want and it's still a simple line. If I look at the vertices, I can see there are only two vertices... Do you think there is no such thing as line/point holes? The only explanation would be really really small (maybe microscopic since my scale was even inversed at a point) areas? – Beatriz Viseu Nov 28 '16 at 17:57
  • Add some geometry that behaves like that as WKT so others can have a look as well. – user30184 Nov 28 '16 at 18:13
  • I don't know PostGIS very well, which is why I commented instead of answering, but in ArcGIS you wouldn't get a line or point feature in the results of a Dissolve operation on a polygon layer. Any topology errors like the ones in your screenshot would appear due to gaps between the polygons and would also be polygonal (although they may not appear that way on-screen if they're small enough). I'm just guessing that PostGIS is the same way. – Dan C Nov 28 '16 at 18:17
  • I think you're right @DanC! I thought it wasn't that because I zoomed in as much as I could and there was still no hole, but then I exported as KML to check its coordinates boundaries. It should have only two pair of coordinates for the inner boundary to be a line. There were more than that. Because they were microscopic gaps, even when I managed the vertices tool in QGIS, it appeared to be two pairs of coordinates. I thought maybe there was a theorical-crazy explanation for points and lines being able to be holes as well.. but I guess not! lol. Thanks! – Beatriz Viseu Nov 28 '16 at 18:39
3

Your logical thinking is fine: those are polygonal topological gaps in your original polygon layer, but the gaps are so small (or narrow) that they appear visually as lines or points. If you examine the geometry, either by going through the XY coordinates of the vertices of the gaps and/or by zooming in really close, you should see that they are all polygonal. As you say, if they were actually points or lines, they would be 0- or 1-dimensional features and wouldn't constitute an actual gap in the polygon layer.

|improve this answer|||||

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.