[HOT] [Ebola] Import of 14, 020 United Nations Mission in Liberia place nodes

[Ralf Stephan]

​Hello,
I would like to share some thoughts and experiences on this import task.​

   - ​Landsat-8 imagery can only sometimes prove existence of a hamlet.
   This may be due to the small size of UNMIL sourced places in general, but
   there are also resolution issues. so there is quite some number of
   place=unknown from the UNMIL data, the percentage practically corresponding
   to 100 minus the Bing coverage in the square.​
   ​
   - Alternatively to setting
   ​
   place=unknown
   ​ beforehand, I went ​
   in a final pass through
   ​a list filtered with ​"
   fixme
   ​:​
   Location
   ​ -
   ​
   place=unknown -source=GNS" to​
   make sure they are
   ​set. That's a bit less involved IMHO than having to set hamlet manually.
   ​
   ​
   - Some hamlets with person names seem to be neighborhoods but often it
   is undecidable if a single farm is meant or a nearby village without name.
   - The Wiki workflow page is unclear on unmil:ids when merging two UNMIL
   nodes but probably that would be ="id1;id2". Any objections against adding
   it to the Wiki?

​If I do something wrong please don't hesitate to criticize.

Regards,​
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/hot/attachments/20140924/cbda44c0/attachment.html>