Errors regarding Helsinki, Finland

[tuukka]

First of all, thank you for providing this great service!

I wanted to use your tiles for online maps about Helsinki: https://loom.cs.uni-freiburg.de/global#tram/geo/13/60.1760137780708/24.930896759033207

You can compare with this (unofficial, daily) map from the transport authority which looks correct (but has its own UI issues): https://kartat.hsl.fi/linjakartta/

I'm mostly interested in the geographical map of tram routes right now:

• Route 4 to Katajanokka is not showing. https://loom.cs.uni-freiburg.de/global#tram/geo/17/60.16738991376385/24.974702596664432
  • 4: https://www.openstreetmap.org/relation/52941
• Some routes such as route 5 to Katajanokka are only showing on zoom levels >14. Is this intentional? At least it's very confusing. https://loom.cs.uni-freiburg.de/global#tram/geo/15/60.16386713981777/24.969863891601566
• On zoom levels 13 and 14, a route is drawn as if it could take a sharp turn at the Reijolankatu triangle (which is not possible in the Y-crossings). https://loom.cs.uni-freiburg.de/global#tram/geo/14/60.1884332864269/24.922227859497074
• Many other errors that look like perhaps the data is just very old?

Are you rendering everything that's in the GTFS file (which one?), or do you try to filter down to currently running routes and leave out the very infrequent variants at the beginning and end of the day?

• The subway looks fine in the geographical map but the other maps have layout issues that make them look more or less bad - perhaps because the subway system is such a trivial one?
• Commuter rail is missing at least the routes A (to Leppävaara) and K (to Kerava), and more depending if you want to include sub-urban lines as well (now missing).
  • A: https://www.openstreetmap.org/relation/7440363
  • K: https://www.openstreetmap.org/relation/7435652
• The routes I and P, which are visible, are drawn in a too complicated way: they are simply two routes from Helsinki to Helsinki via Airport/Lentoasema (Helsinki-Pasila-Tikkurila-Lentoasema-Vantaankoski-Pasila-Helsinki and vice versa). Perhaps the loop causes issues in the rendering? https://loom.cs.uni-freiburg.de/global#rail-commuter/geo/17/60.2020210763314/24.932173490524296
  • I: https://www.openstreetmap.org/relation/358139
  • P: https://www.openstreetmap.org/relation/357191
• Long-distance rail is missing everything else but Helsinki-Turku and Helsinki-Tampere.

Let me know if I can help with the GTFS or OSM data, I'm quite familiar with them.

[tuukka]

Ok, I didn't know the service tag can be used for route relations: https://wiki.openstreetmap.org/wiki/Key:service#Train_Routes

As an alternative, you could also support the passenger tag: https://wiki.openstreetmap.org/wiki/Key:passenger

I've now added these tags to A and K.

I'm not seeing what's missing for route 4 - do you have validation error messages?

If you are interested in comparing with other semi-schematic maps, here's some really nice ones of Helsinki (not quite up to date though):

• https://elmoallen.name/raitio/
• http://www.inat.fr/metro/helsinki/
  • https://www.linkedin.com/pulse/helsinki-metro-map-integration-scales-jug-cerovi%C4%87

[hannahbast]

Thanks for the feedback, Tuukka.

Patrick (@patrickbr) is on vacation this week (it's a holiday week in Germany). He will reply when he is back (maybe earlier, but not sure).

A few things I can say:

The data for these maps comes exclusively from OpenStreetMap. You can see the respective SPARQL query when you click on "Statistics" in the lower right corner of https://loom.cs.uni-freiburg.de/global .

We also have a visualization based on (publicly available) GTFS feeds: https://travic.app . But that is a different application, it shows all the vehicles in real time. In that application, you get a particular line when you click on a vehicle. You can also search for particular stations.

[tuukka]

Thank you for the pointers Hannah! I solved what I was working on in another way, so have a nice holiday week and get back to this only whenever it's convenient for you again.

I can see the missing data in the Sparql results. My guess would be perhaps the tile generation ran out of disk space or is otherwise broken - why else would the lines appear and disappear like that when zooming?

[tuukka]

Update: I had another look at this (for a small project) and realized that 1) all HSL's tram routes are being combined into one because they use the same route color (you should ignore the colors in this case), and 2) no HSL's tram stops are present because we have mapped public_transport=platform and railway=tram_stop but not public_transport=stop_position for trams.

1. Here's another way of getting an id for grouping the variant routes without depending on the colors. This uses the route_master relations instead: https://wiki.openstreetmap.org/wiki/Relation:route_master

https://qlever.cs.uni-freiburg.de/osm-planet/5EYv2f

PREFIX osmrel: <https://www.openstreetmap.org/relation/>
PREFIX osm: <https://www.openstreetmap.org/>
PREFIX osmkey: <https://www.openstreetmap.org/wiki/Key:>
SELECT ?color ?id ?label WHERE {
# ?id osmkey:network "HSL" .
  ?id osmkey:type "route_master" .
  ?id osmrel:member ?route_master_member .
  ?route_master_member osm:id ?rel .
  ?rel osmkey:route "tram" .
  OPTIONAL {?rel osmkey:colour ?color} .
  ?rel osmkey:ref ?label .
}
ORDER BY ?id

2. Here's a modification of the original query to include nodes with public_transport=platform or railway=tram_stop too: https://qlever.cs.uni-freiburg.de/osm-planet/N9gPfo

[tuukka]

Update 2: I got it mostly working with the modified queries. To get the stops showing, I additionally snapped them to the route geometries (in QGIS; working input to topo). Route 5 still has some topology errors at its both ends that should form loops (in one direction, in the other direction).

[tuukka]

Hi! I'd like to hear if you can tell something about your plans. For instance, it looks like the global tiles have not updated since May 16 (based on the timestamp of the tiles).

[patrickbr]

Hi, sorry for the late reply and thank you for your work and investigation into this problem. We spend the last weeks preparing a publication of our approach and are now preparing regular updates of our tiles. I will experiment with your updated query and if everything works well, we will incorporate it into our process :)

Other things we are currently working on: (1) getting the line directions out of OSM using SPARQL queries. (2) filtering out irregular routes, or visualizing them as dashed lines.