I’ve noticed that some railway stations (particularly those of the Purple Line in Bangkok) are currently redundantly tagged both with a node and an area. I’ve discussed this a bit here, and it would probably be best for the community to agree on a single preferred method.
So the question is, for railway stations in Thailand (both the national railway and rapid transit lines), should we prefer to:
(A) Always place the railway=station and name=* tags on a node representing the centre of the station (tagging the buildings/areas only as building=train_station / landuse=railway), or
(B) Place the railway=station and name=* tags on an area representing the entire station if one is mapped, or on a node representing the centre of the station otherwise?
I think it will be difficult to make a hard rule about this because so many of us do a lot of our mapping using satellite imagery and therefore don’t know which particular building is the actual station. If I’m familiar with the station and have visited it, I can be more specific but in many cases, I am working from a geolocated photo snapped while driving past the station entrance and imagery that may or may not be clear enough to resolve fine details on the ground.
My preference would be to tag using your option B whenever possible simply because it’s easier for me given the way I gather my OSM data. I don’t do any mapping in Bangkok and I’ve never visited any the stations you mentioned so from my point of view, you’re free to handle them any way you like.
I think mapping as node is more useful, because an area of railway station is maybe hard to find, as the area along the railway is all owned by The State Railway of Thailand and can’t be told exactly where is the boundary of the station. Moreover, some large station may have a large boundary that if we tag the area, the center shown in the map is might far from their real center which people using the map want to know.
Generally speaking I understand the OSM preference, is to tag as a building (closed way) rather than a single node, if the entity occupies the entire building. Obvious exceptions are shopping malls, and businesses situated in “terraced buildings” where to draw each one, is not only difficult, but very time consuming.
From a personal standpoint, I do uprate most petrol stations nodes, to building=roof as they show up on my GPS, and therefore negate the need for me to add them as I fly past at some silly speed on the bike
I also might draw some of the larger/important buildings in a poorly mapped city, this way. This helps you get your bearings on a rendered map.
I realise that some Wats & Schools are mapped in a very haphazard way here, with some drawn as areas, (with or without buildings), and some as just nodes. I might convert the main temple from a node to a building, and give it a name, but I feel it is wrong to then copy that name=* & place_of_worship=* tag to every other building on the site. This just clutters the map and makes waypoint selection difficult. The same would typically apply to school buildings.
I certainly agree with Paul’s comment about the “one feature, one OSM element” principle applying, and its infuriating when I see people add an amenity=fuel tag, slam in the middle of a PTT gas station canopy I have drawn, simply because they either don’t realise the fuel tag is already incorporated in the building … or they are part of the Maps.Me brigade !
To answer Paul, in the same vein as above, I would probably draw the main station/terminal building, and add all the name=* and other informative tags to that alone. I tend to map from a “navigation standpoint”, which is to say, if I wanted to be directed to a certain point, where would OSM put me ?
So for a school/terminal/Wat etc, the main building is where you want to be taken to, and not some geometrically placed node in the middle of an area.
My opinion is wats and schools should be tagged as an area (with fence or wall if exists), not building, because, for wat, it’s not like church or mosque that the main building can be identified. Most wat in Thailand is a group of many buildings, in both law terms and people’s perception. Tagging the most important building of wat like ubosot (for some wat) is lead to a problem because for many wats the ubosot is not the place that people usually go. Moreover, some religious activity for wat is happening outside of the building. When navigating to the wat, the navigation should let people go through the gate at the fence, and that’s it.
For school, same as a university, people going to school should be navigated to the school fence gate and that’s it, because most people is not going to the administrative building of the school.
Yes, I agree that a Wat is really more like an area, and of course, if you are sure of the boundary, go ahead and draw it. Then put in every building, and of course, add a name for each building. Add the service roads, and then finish off with a entrance=main tag if appropriate.
That would be wonderful, if anyone has the time, and information, of course.
However, for most of us we all prioritise our additions, and at a time when I’m still changing rural unpaved tracks up-country, to the 2 lane asphalted roads they have now become… , Im afraid when I see an un-mapped Wat closeby, it just gets a quick building and place of worship tag. I can see the main building or *ubosot *if you prefer, but the boundary is often very vague.
So, I think the point I am making, lets not all get too pedantic about the perfect way to map a Wat, when some of the major roads and highways still need to be added.
A long time ago we had the railway station as part of the railway line. It was nice from a data evaluation and checking perspective.
As we are mapping now in a much higher detail level, with high-resolution orthophotos available, we can certainly advance from what we used at old landsat 2km resolution times.
If the photo allwos to uniquely identify the main building, you can tag it.
For larger stations we probably have a situation similar to a mall. The station building provides the room for multiple smaller shops. Think of Hua Lamphong with all the shops surrounding the main waiting area.
I am not too deep into the current tagging scheme for public transport. Maybe one could volunteer to check on the tagging list or cross-check different styles and the wiki.
I would be ok to have the building simply tagged as building=yes (or building=whatever) and then have a node inside to represent the station. Other nodes representing the other shops inside the station. As ticket counter, food, pharmacy, …
I kind of forgot to follow up on this. From what I see, I think we’re fairly agreed that wat and schools should have tags placed on the entire area, if possible, and otherwise with nodes as a temporary measure. This makes sense to me, as both are most often large fenced compounds.
Railway stations seem less clear. In some cases, the area of the station might be clearly defined, but not in others. Some consist of a single building, some have multiple. Since the station area includes not only the building, but also platforms, sidings, yards, etc., the building alone should probably not carry the tags for the station. It would seem that tagging the station as a node would be optimal as a general rule.
This mostly applies to national long-distance heavy rail stations, though. Should the same apply to elevated stations whose areas are quite clearly defined by the building footprint? I’m now tending to think yes, for the sake of simplicity.
Keeping the mapping scheme simple as long as fulfills the needs makes a lot of things easier.
A railway station as a node, some other nodes representing ticket counters, shops, etc are all easy to follow. A building footprint with building=yes is also easy.
This simple data structure can be digested by almost all data consumers. Even on old Garmin Handhelds you have something useful.
If the journey goes into more complex multi-polygon relations the numer od data consumers correctly parsing the data rapidly degrades. Also the likelyhood of someone (or some editor) breaking that data structure increses.
So as long as we can live with a simplification I feel happy to accept that a map is always a simplification of reality.