Building Footprints vs Roof Areas: Best Practices 2024+

Very good “best practice”. This also helps to reduce the bad practice of mapping adjacent buildings with separate nodes, as I asked in this thread: Map Buildings next to each other with a space between them?

1 Like

I would add that in France, the land registry (cadastre) was imported in many places, so the actual walls are mapped. And having myself mapped lots of building (not roofs) before the import, using the trick you very well illustrated (and some other tricks) I know that building walls can often be mapped with a reasonable precision even from simple aerial imagery.

2 Likes

Thanks for such valuable feedback again!

Determining feature placement accuracy

So for the original discussion, both roof area and building footprint are acceptable approaches in OSM, but the more accurate the better!

It would be great if the feature could be tagged to distinguish between ‘roof mapped’ or ‘footprint mapped’, so the user knows the level of accuracy.

There is also a distinction between ‘footprint mapped satellite’ and ‘footprint mapped drone’ and ‘footprint mapped, ground truth/verified’.

What would be best tagging practice to distinguish between all of these levels of accuracy?

Deletion / modification of data

As for the topic of data deletion, feedback on the following scenario would be fantastic to clarify things :smile:

A mapathon produced data for an unmapped town in a developing country.

The buildings are mapped using satellite imagery that is two years out of date.

A few years later a new mapping campaign is started using drone imagery for the town, with the idea that the buildings are also ground truthed and tagged in the field.

The original OSM data was incorrectly placed (offset) geometries, simply with tags building=yes.

The old data is conflated with the new, attempting to match the existing geometries and merge the existing tags where possible (I am assuming it’s ok to replace building=yes with more useful building={field_verified_building_type}).

The end result has a diff, where some of the old geometries no longer exist or are simply in the wrong place. Can they be deleted? (considering the new data is up to date and ground verified)

Have read about mapping building overhanging roof edges, here it’s kind of standard, 1 meter sort of a going value, just enough to have the sun not enter the windows and heat up the walls at high noon, but have never bothered to add these… only the roof outlines as seen in sat imagery as building=*.

Wonder how they do this where import buildings is routine such as in The Netherlands where every new and revised building gets pulled in on request (a special thread for this on the NL community category) from ‘BAG’.