[Imports] Vermont Town boundaries from VCGI

Andrew Guertin andrew.guertin at uvm.edu
Mon Dec 2 21:38:04 UTC 2013

Newer version, with counties and villages: 

This does not yet have admin_level on member ways, but that's coming 
(unless the discussion on that progresses in a different direction).

I have county boundaries marked with admin_level=6, cities and towns 
with admin_level=8, and villages with admin_level=10. This goes against 
http://wiki.openstreetmap.org/wiki/Key:admin_level#admin_level, but 
since a village is (to some degree) subordinate to a town it doesn't 
make sense to me to have them at the same admin_level. I'm open to 
advice here.

I have counties marked with place=county, cities with place=city, towns 
with place=town, and villages with place=village. While these match up 
with the official terms, they do not match the definitions at 
http://wiki.openstreetmap.org/wiki/Key:place very well. Advice 
appreciated on this.

I've taken another look at the state and national boundaries. As I 
mentioned before, the boundary with Canada is almost identical to what's 
currently in OSM--a difference of only a few cm--which is probably 
because the source for both is probably the IBC data and those few cm 
are due to projection changes or digits of precision. I plan to keep the 
data currently in OSM.

At the NY/VT/Canada border, there's a difference of ~1.5 meters. There 
are no IBC monuments in the middle of the lake, yet there are 3 
unsourced points, plus the actual NY/VT/Canada corner. I plan to remove 
all 4 points, connect the IBC monuments with a straight line, then set 
the NY/VT/Canada corner as the point on this line closest to the VCGI data.

The border with New Hampshire has some significant differences between 
VCGI and OSM, the largest ones dealing with oxbows in the river. Most of 
the current boundary has source="Complex Systems Research Center, 
University of New Hampshire" I'm inclined to believe VCGI here, since it 
looks like the OSM version is derived from a "western bank of the river" 
rule applied to aerial imagery whereas the VCGI data looks like surveyed 
monuments, which better matches the law. Unless anyone has other 
information, I'd like to replace this border with the VCGI data.

For the Massachusetts boundary, there are differences on the order of 
10m or less for most of the length. It looks like the eastern half is 
unsourced (or the source was lost in relation cleanups), and the western 
half came from Tiger. This surprises me, as I'd assumed it came from 
MassGIS. I trust VCGI more than Tiger, so I'd like to replace this 
boundary with VCGI data.

For the New York border, on land the typical difference is ~10m, but 
around some curves it can get up to ~50m. There are also a few 
differences in oxbows. In the lake, the differences can reach into the 
hundreds of meters. Again, it appears the boundary came from Tiger, so 
I'd like to replace it with the VCGI data.

More information about the Imports mailing list