[Imports] Announcement: Corine 2006 Land Cover Import for Republic of Ireland

[Jaak Laineste]

> The fact is that we've had the ability to trace landsat for ages now, Yahoo in
> places and Bing now in a bunch more areas. It isn't giving us full land cover
> information is isn't likely to any time soon, if ever.
> Part of this is the fact that mappers map partly what they realise they can map. I
> know that there are cases where imports have removed the motivation of
> mappers to participate. I don't see that danger here.
> 
> Basically, having reviewed the quality of the Corine data for Ireland, I have come
> to the following conclusions:
> 
> * It's not perfect, but...
> 
> * ...It's a lot better than the _nothing_ we currently have in a lot of the country
> 
> * It's too much and too details for a small community to be motivated to
> manually trace
> 
> * It includes stuff that's quite hard for a lay mapper to identify by other means
> (peat bogs, different types of agriculture, categories of tree in forests)
> 
> * Even though the Corine lakes may be crude, they are often better aligned than
> the stuff traced off uncalibrated landsat.

I had very similar motivations when we undertook huge work of importing Corine in our country. We had pretty empty map before, so no much overlapping, I spent a lot of time to match it half-manually with coastline (and Estonia has really lot of that) etc. But the real issues came out some months afterwards, when we tried to have do detailed maps of villages, lakes and rivers. In real street level Corine areas do not make any sense - just random lines around. You cannot even just delete them easily, as they are technically quite complex and huge relations (often larger than OSM API download limit). 

 I suggest to do deeper testing (than I did) to see will happen after import: import data to OSM dev API machine, and then try to get it fully fixed for some area; merge with existing lakes, city building regions etc. Polygon and relation modification tools have become better in later JOSM versions, but bunch of polygons and nice topological model of OSM just do not fit together in any easy way.

Jaak