[Tagging] tagging single trees

[NopMap]

Alan Millar-2 wrote:
> 
> The solution seems pretty simple to me.  Add something like  
> "denotation=landmark", and then you always know when you have your  
> significant landmark tree.  If you also want to add denotation=urban  
> on other trees, that's good also.
> 
> If you find a tree without any denotation, then you know you found a  
> tree without denotation.  If you want specific status, one way or the  
> other, tag it with denotation.  Don't trust the absence of a key to  
> tell you something important.
> 

That is not a solution. For 4 years people have done valid tagging, using
the definition in the wiki for significant trees. If you change the meaning,
no denotation=landmark will magically appear there, so the information gets
lost. The mappers who originally contributed them have no idea that you
changed the meaning on them, so nothing will happen to fix the damage.

I have done a statistical analysis of the distribution of tree nodes in
Germany. The result indicates that 4585 trees are actually single trees.
(They don't have another tree within 50m). That makes about 15.8 %. Assuming
the same rate globally, you'd throw away the information for about 59000
nodes that actually describe a lone and significant tree.

Statistics also show that the real significant trees are much older. Average
change set id is 3.1M as opposed to 19.2M on the badly tagged generic trees.
Your chances are much better that the mappers are still around who did the
mass generic tree tagging to fix it. And as those trees are clustered in
bunches of up to 2500 at a time, they can be very quickly fixed.

So again, the conlusion is: Fix the new nodes, don't destroy significant
information on 59000 nodes!

bye
          Nop



-- 
View this message in context: http://gis.638310.n2.nabble.com/tagging-single-trees-tp5501462p5504061.html
Sent from the Tagging mailing list archive at Nabble.com.