[Talk-us-massachusetts] Talk-us-massachusetts Digest, Vol 45, Issue 3

[Greg Troxel]

Yury Yatsynovich <yury.yatsynovich at gmail.com> writes:

> OK, if I can not touch MassGIS imported borders to align them with rivers,
> what about aligning rivers (on their particular segments) with borders?:)

So I will take that as an expresion of frustration rather than something
you want to actually do.

I did not mean to say "we can never deviate from MassGIS".  What I meant
was that we should discuss a plan and find consensus about how to deal
with these boundaries, which involves asking questions and responding
them and trying to really understand what's going on.  I was objecting
to starting what amounts to a large-scale edit that deviated from the
previous consensus without having the dicsussion.

As for alignment, this is tricky (not so much in this case, perhaps),
but I think in general we can't expect things to align so well.   Not
too long ago, I was editing conervation land, doing random fixups of
trails and other things - normal hand editing.  As part of that, I moved
vertexes that were joint with with parcels a meter or so to better line
up with L3 parcels.  Stow has recently had an update and the quality in
two other places was quite a bit better, so I figured that the
no-lnoger-matching was an improvement.

In this case, the parcel in Stow and the one in Acton really are on the
town line, because they were split that way when acquired to two
conservation trrusts.  The L3 parcels line in massgis tiles and the town
line as shown in josm are off a meter.  Despite that, I felt that the
town line was the best representation of the town boundary while at the
same time the L3 parcels was the best representation of the parcel
boundary (and hence landuse=conservation/boundary=protected_area).

Another mapper was concerned that people who zoomed in would complain
because OSM would appear to show part of the acton trust land being in
Stow.  I think it shows no such thing - maps need to read with an
understanding of accuracy.  But, I moved the vertices to very near the
town line (about 1m) to avoid this - since after all none of this is
good to better than 1m anyway, and probably several meters in L3
parcels.

This alignment notion is very powerful; people said in another context
"osm shows that this trail crosses into private property".  Again, there
is a representation of a trail which has accuracy issues and a
represnetation of a conservation land boundary which also has issues.
So at best the map show that they are close.


(I'll write separately about towns/rivers.)