I would like to announce a new MapRoulette challenge: Sweden - Fix intersecting buildings
(Sweden - Fix Intersecting Buildings)!
We’ve identified buildings on OSM that are intersecting with the outlines of other buildings.
Addressing intersecting buildings is important for improving OpenStreetMap’s accuracy and usability. These inconsistencies can make maps appear cluttered and confusing, hindering their effectiveness for both humans and machines. By correcting these issues, we ensure that the map accurately reflects the real-world layout.
Please don’t hesitate to reach out to me via this forum or other channels if you have any questions.
Ah!, okay that’s very useful.
Do you as Swedish community already have an idea if you want to import the buildings, and how to go about this?
Because it will for sure be of great value to the map to try and get all those buildings without it taking too much time because you have to trace everything.
It might be worth looking into how other countries have done their building imports, and how they make sure they are kept up-to-date.
We have had some smaller discussions so far but no method has been selected yet. The article linked is only a couple of days old, so the project has just started. I think some form of consensus will be agreed upon in time, but we just got the details of APIs and formats used. There are still some details like API specifications missing for some stuff so more input is needed there.
I will definitely check out the other projects as well. Thanks for the links.
Hopefully the other projects give some inspiration on how it could be done, I know there are many more building imports than just those two. But I think it’s good to think about not only putting the buildings on the map just once, but also keep the data up to date if it’s a dataset that gets updated every once in a while.
What data is it that will be available in February? I heard address locations and building outlines mentioned.
I just wonder where I can read up a little bit how this will change how we are mapping in Sweden. Will we after the import of address locations have all addresses as address nodes instead of being set on amenities or buildings, like how I heard they have done in Denmark? What if surveys reveal the addresses are more or less granular in reality than the imported data suggests, what edits are we allowed to do to addresses when it is being auto-imported?
What about buildings? Should we no longer map buildings Lantmäteriet are unaware of, or no longer delete non-existing buildings they still think exists? Will all existing buildings be deleted to fit the new data, or is it only buildings in Lantmäteriet’s data that isn’t overlapping any currently existing or recently existing building in OSM that will be added?
Where are the discussions being held?
Because I am currently adding lots of buildings from aerial photos, but maybe there is no point in doing that now, and just wasted effort? I haven’t heard anything about an upcoming import.
I don’t know the answer to your question, but I hope you have position calibrated your image background? If not, most of your added buildings will have an more och less error in their position. In Sweden we have the “Trafikverket Road Map” which can be used for calibration of the background image.
That will highly depend on exactly what data will become open, down to the attribute level.
In the data Lantmäteriet has there is a connection between addresses and buildings (as well as entrances and dwellings/“lägenhet”, sadly those have no geographic component). So If that connection is made available then we could add the addresses to buildings when applicable. If not, we can still do so in many cases (isolated buildings/addresses where it is clear that they belong together). But there will be a lot of address nodes added, yes (but any import should take existing addresses into consideration).
Don’t need a survey for that - sadly there are a lot of cases of very badly placed addresses (usually this is the case in newly built areas, where the municipality initially just places the addresses more or less at random in each parcel and then don’t move them once the buildings are there, but there are also many other cases).
This is mostly educated speculation, but: I don’t think that we’ll buy Lantmäteriets data wholesale (though I think there is an argument to fully replace some types of data in some areas entirely, like landuse in northern Sweden). So we’ll keep buildings missing from Lantmäteriets data (but might want to double check them) etc.
This forum (in various topics).
I can only speak for myself, but I have paused any large-scale mapping, mostly because of a lack of time but also because of the soon-to-be available new data.
@CasGroenigen The problem is that the authorities have classified almost all data at the highest possible end of GDPR, and a CC BY 4 license. The data will be free to use, but it’s not going to be truly open. Individual human approval of each usage request will become the norm, they say.
