[Talk-us] GNIS Import Done

[Alan Millar]

>>> 2. I'm planning on taking the planet dump and grepping out all of the
>>> GNIS
>>> data. This will be used to generate diffs that we can send back to the
>>> GNIS
>>> board, who has the option of putting it back into their data set.

>     Also, please report
> anything you find. It would be great if the USGS had the wherewithall
> to create an automated method to pull any changes out of OSM and back
> into our databases.

Well, I'm not planning on making a list of all my OSM edits to send back
to GNIS, so I'm hoping the diff or other extract method works out.  I'm
finding errors and I'm just fixing them in OSM.

I've been looking at schools and churches in the Portland and Seattle
area.  It didn't surprise me to find duplicates between GNIS-loaded nodes
and previously-entered OSM nodes.  The JOSM validator duplicate name
detection and node merge make them easy to fix.  If two identical names
are within 0.0005 degrees of each other, they're probably a duplicate of
the same item.

However, it did surprise me to find quite a few duplicates within the GNIS
data.  For example, I found duplicate nodes for schools, with identical
names, within a few hundred feet of each other, but each with its own GNIS
ID number.  Often there would be one node located right on top of the
school building, and the other node on the street in front.  (Looks like
perhaps two datasets got loaded into GNIS, one perhaps geolocated by
address?  Who knows.)

In the case of duplicated GNIS nodes, I've been keeping the one located on
the building or school grounds, and deleting the one at the street.

I hope the diff or extract method will be able to capture deletions like
this and feed them back to GNIS.  Right now, they probably know there are
duplicates but don't know which one is correctly positioned.  Our crack
team of highly skilled OSM cartographers, as well as nerds like me, can
help clean it up.

- Alan