OSM landuse

I emailed Alexander Zipf last year and he replied to say the improvements would be available soon but no more details. I emailed again the other day to check.

However, this team has put a huge amount of effort into improving the accuracy of the landuse model, I don't want to nag them to add to OSM. I was wondering if there is already some collaboration? When this data is available (to my mind it is already very accurate and does not need to wait for checking) who would implement the changes?

There's a fair amount of additional material about how to use OSM Landuse elsewhere: for instance I looked at how OSM compared with the CEC Urban Atlas for various places. Presentation at SotM-11. Igor Brekic also looked at Vienna some time past. The basic approach as by the Heidelberg term is: some basic conversion of tags to landuse categories, sensible defaults (farmland works well in lowland Britain), a polygon merge & clipping process, based on an order priority of categories (a simple painter's algorithm worked for me). Additional processes for recognising urban areas have also been documented. Lastly some judicious use of other sources can fill some voids (e.g., Corine woodland in Ireland, other national datasets).

Interesting to know. I will look up the CEC atlas.

What I am mainly getting at is the osmlanduse website is called OSM, implying it is some how connected with the OSM geofabrik data.

In any case rather than trying to replicate the work of Heidelberg, would it be possible to use the dataset they have put together?

If they haven't released it then you can't use it other than as a WMS. Yes of course the site is based on OSM data, but they enhance it using their own, probably proprietary, algorithms. They may of course provide some kind of paid service too (after all if you are using it for commercial purposes, it's not unreasonable).

OSM data is likely to be much better than CLC Corine when it is more than imported Corine data (which is why coverage of landuse is good in many EU countries). OSM is more likely to be properly aligned at changes in landuse; is often more granular, and because it is generally not dependent on remote sensing avoids some classification errors.

The accuracy and resolution of osmlanduse.org is far and away better than what is currently available from geofabrik for the UK and NL.