I started mapping what I saw in Germany (even little things like benches), but some days ago my attention was put to other cathegories:
- several buildings missing / old / misaligned (probably because data is old; was e.g. at https://www.openstreetmap.org/node/51248076#map=15/49.8319/8.6970))
- most roads, wood, landuse and addresses mapped, but no houses at all (was e.g. at https://www.openstreetmap.org/#map=14/53.7568/20.9390))
- areas where there is nothing mapped besides major roads. (e.g. https://www.openstreetmap.org/#map=9/7.0859/18.6905))
How to address these?
is it ok to map them only based on arial imagery (whereever possible with existing gps tracks, but probably they are only available in cat 1)
- how can I know the age of arial imagery
- why is it that there is areas / tiles nearly whithout any map data
It feels incorrect to pixel-perfectionate detailed maps in Germany when there is places where there is not at all a basic coverage to be mapped.
What do you think about this?
Are there guidlines or helpful resources?
Germany is one of the well mapped countries on OSM because there is the biggest community.
Many regions in the world aren’t mapped or have only roads, because nobody has mapped it - OSM is based on the work of volunteers.
Yes! Any help is welcome.
For imageries for the whole world it varies by region. For local imageries (country, city) there is usually information about it.
I saw you mapped also Poland, for Poland’s imageries dates are more or less:
Geoportal 2 - 2017 (in some regions 2019)
Bing - 2019
polska.e-mapa.net Buildings - this is the best and the newest one for drawing buildings. But there still may be also buildings already demolished.
For some imageries you can check it in the iD editor after pressing Ctrl+Shift+B.
Filling in largely empty areas based on aerial imagery is usually considered ok. Editing existing data from afar, though, can sometimes be problematic. It’s easy to make mistakes, and even if it is mostly an improvement, it can be demotiving to have outsiders pop in and “fix” your work. So personally, I’d proceed with caution on your first example if I had no personal connection to the place. Although the building outlines are clearly not as precise as they could be with today’s imagery, the mapping still seems quite good overall.
Usually because there are fewer mappers there. Which can have many causes – population density, standard of living (access to computers, available spare time), cultural differences (likelihood to volunteer, and for which activities), language barriers and many other factors. There’s surely also a bit of a network effect going on, and once you have a few people, they can recruit others. So areas where OSM got popular early on have a head-start.
Of course, you can only get so far without mappers on the ground. So the longer-term goal is to allow a local community to form and take care of the map. There has been some debate whether having all the “easy” work already done by remote mappers makes it more or less likely that a local community will eventually form, but we ultimately don’t know.
Local knowledge is a unique selling point of OSM, so I don’t think it’s wrong to map what you know. Mapping far-away buildings from aerial imagery is something “everyone” could do just as well as you, including the various corporate editing teams, whereas you’re probably one of only a few people available to add details to your home area.
If remote mapping is fun to you, like it is for many others, go ahead. But don’t feel obligated to do it – mapping your local area is when OSM is at its best.
Thanks for your evaluation maro21 and Tordanik,
thanks for the tipp with Ctrl+Shift+B in iD, I will try that out.
Is there a possibility for getting offset data in iD like josm seems to have?
For the first place in Germany: I had been there physically, taking note about some unmapped buildings, but when I at home checked there were a lot more missing and quite some misaligned (relative to each other and the roads).
Due to the notes I took I could decide which is newer and which older aerial imagery data.
Unfortunately I did not took a lot of gps data myself to check what would be a good approximation for the aerialimagery offset. So I used existing public gps tracks and compared buildings that looked well mapped with those that looked not well mapped. I added missing buildings and of the existing ones only touched those, where the offset was large, shape was different or obviously the mapped building was a different one then the one now existing.
Hope this was a good approach - actually could have been more precise by using better gps data for alignment, but that’s what I had at this point.
I also got immediate feedback from a local contributer that was happy to have some new housenumber quests in StreetComplete now.
I did the same for a village near my home afterwards.
The second place in Poland I found via https://resultmaps.neis-one.org/unmapped and mapped the existing buildings. I did not apply an offset as other mappings seem to have been done similarly.
Also I was not sure how old the maps are and how reliable polska.e-mapa.net data is.
I mainly mapped by Geoportal comparing it afterwards to polska.e-mapa.net (and also Bing). Natural features I mainly mapped from Bing.
I stopped when an area looked very different in Bing and Geoportal: https://www.openstreetmap.org/note/2355456
Also then I began to wonder if it is a good approach to map like this and so came here to ask.
I understand that some places are still not happed, because simply there was nobdy yet doing it (although I actually was not aware that levels of available data differ so much).
For our well-mapped region Germany: was everything really drawn manually or have there been some big imports initially to kick off the thing?
For the third sample location in Africa I was wondering about the rectangle shape of data that was mapped vs. not mapped.
This does not look like a natural “just this has not been mapped” border to me - and there is multiple of it in this area.
For what it’s worth, it sounds like you’re putting a lot of thought and effort into the quality of your work. Also great to hear that you got positive feedback from a local, that’s always nice!
Are you thinking of something like the Imagery Offset Database? I’m not aware of anything, unfortunately.
The map in Germany is largely the work of volunteer mappers. I can still remember when my town was mostly a blank map with a motorway running through. Germany enjoys an unusually large OSM community and regularly tops rankings like number of mappers, number of OSMF members and so on.
I’d say there were fewer imports in Germany than in many other places. In part that’s simply because of less generous licenses that didn’t allow imports (compared to, say, the US, who had public domain government maps). In more recent times, some governments and data providers have started opening up data, and there have been regional imports, but these are generally smaller initiatives.
Yeah, this is very characteristic of imports or organized mapping. Africa in particular is seeing a lot of humanitarian mapping efforts. This particular region seems to have seen UN-related editing.
Do you want to say just copying the shapes from polska.e-mapa.net is enough and only look for new / destroyed buildings? Of course this will take less time
What building shapes I created currently from aerial imagery does mostly not match 100% to the polska.e-mapa.net outline, but is mostly close.
In Germany Baden-Württemberg there is similar (actually more) data from Maps4BW, but also in the wiki it is warned about simply copying it.
polska.e-mapa.net is really helpful imagery but it doesn’t mean that is 100% correct.
We draw buildings from it. It comes from official data and is updated quite frequently. New buildings are added there but it depends on region. Removing demolished buildings from that imagery isn’t as fast as adding new buildings. So it’s always good to compare with aerial photos. Sometimes there are errors such as underground parking lot as a building but it’s rare.