[Imports] Slovenia landcover import RABA-KGZ review

[colored stone]

I think this isn’t really an error. This estimation based on analyses of
existing elements raba:id=1410 (currently classified as natural=heath) –
the question was – “how would OSM mapper classify this element according to
the satellite image”. More than 80% of this elements would be classified as
natural:scrub, some of them as landuse=forest (sometimes it is very
difficult to distinguish scrub from forest), very few as landuse=meadow
(1410 are former agricultural areas now overgrowth).


Main reason for these classification “errors” is that the source for land
cover is different (digital orthophoto 1:5000, covering last three years –
the situation might already change, detail identification process made by
experts – “normal” OSM mapper might not “see” the same).


So, if someone would have the same knowledge and specially the same image
source, the classification error would be much lower.


I think this is not the case to go for indivudual verificatilon - if it
would be the case - the whole import is under question.



Martin

On 23 November 2015 at 09:41, Martin Koppenhoefer <dieterdreist at gmail.com>
wrote:

>
>
> sent from a phone
>
> > Am 22.11.2015 um 23:25 schrieb colored stone <coloredstone.si at gmail.com
> >:
> >
> > We have concluded that natural=scrub corresponds to raba:id=1410  much
> closely. Not really exact – but the estimation is than more than 80% of
> existing elements of raba:id=1410 could be classified as natural=scrub. We
> have concluded that these elements should be imported with tag
> natural=scrub.
>
>
> if there is really an error amount close to 20% for this class you should
> verify all of them individually, it's definitely too high to be tolerated
> to just import them blindly.
>
>
> cheers
> Martin
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/imports/attachments/20151123/1ae80155/attachment-0001.html>