search

root676 / QNEAT3

QNEAT3 - Qgis Network Analysis Toolbox 3
GNU General Public License v3.0
63 stars 12 forks source link

Change default topology tolerance to 0.000001 #34

Open josephholler opened 4 years ago

josephholler commented 4 years ago

I'm attaching a very simple example geopackage with a layer, roads to be used as a network and another layer, parks to be used as start points for creating Iso-Areas. qneat3test.zip Specifically, I'm running Iso-Area as Pointcloud (From Layer) with distance optimization of just 500 meters to test. I originally digitized this data with snapping enabled and densified the line geometries such that some start points are exactly coincident with vertices in the road network-- or at least coincident within the margin of error of precision with which the coordinates are stored. The crazy thing is that without modelling one-way travel, the park with ParkName FC only has an iso-area going one direction on the road. However, if I change the topology tolerance to 0.000001 from 0.000000, it works just fine. That's just a fraction of a millimeter, so I'm confused as to why this makes a difference. I can only guess that something is going awry in adding a start point to the network if the start point is coincident with an existing vertex. The same thing happens with the QGIS algorithm Service Area (From Layer), so the root of the problem is probably in the QGIS code. I can't attribute this to an error with the network data, because the startpoint in question is in the middle of a line feature. In other words, it's clearly not a failure of one line to connect to another in the network due to digitizing errors. Either way, an extremely small tolerance of less than 1 millimeter apparently fixes the problem, so could this be the default?

root676 commented 4 years ago

I have never encountered this problem but I am glad you found a way around - we can definitely change the default value of the tolerance parameter as you proposed it.