[Talk-us] Post bot cleanup

Alan Mintz Alan_Mintz+OSM at Earthlink.Net
Fri Jul 20 01:19:50 BST 2012


At 2012-07-19 13:41, Phil! Gold wrote:
>* Martijn van Exel <m at rtijn.org> [2012-07-19 14:22 -0600]:
> > On Wed, Jul 18, 2012 at 11:42 PM, Toby Murray <toby.murray at gmail.com> 
> wrote:
> > > Any other common problems that people have seen?
> >
> > Another thing I find is a lot of leftover stray nodes without any tags.
> > I select them in JOSM with type:node tags:0 -child and delete them in
> > one fell swoop.
>
>Eeek!  Be very careful with this!  The unconnected, untagged nodes were
>left deliberately to facilitate recreation of roads, particularly the case
>where a decliner split a TIGER way but did not change its geometry.  I've
>been making use of these nodes in South Carolina to reconstruct the roads
>there.

That _is_ a nice feature. One thing I would do (in JOSM) when first loading 
an area to work on is run the validator. This will find those orphan nodes, 
as well as crossing ways that don't intersect, etc. - indicators of things 
that need to be fixed.

While it might seem easier sometimes to delete and re-draw objects, be 
careful with objects that have source/source_ref values, as this says that 
a user has edited/confirmed that object (with the exception of some of the 
automated import sources), which should be considered the highest form of 
authenticity. Be careful of intersections with turn restrictions and their 
connecting roads. These render with little restriction signs on/near the 
intersection in JOSM and Potlatch.

Unfortunately, TIGER11 still has naming inaccuracies. I recently did this 
neighborhood with some new construction in Yorba Linda, CA: 
http://www.openstreetmap.org/?lat=33.8987&lon=-117.8017&zoom=14&layers=M 
and put note tags on a number of roads that required research that 
ultimately contradicted TIGER. I think there were a couple more that were 
wrong that I did not annotate. That would be roughly 1-5% of the roads in 
the area, and a larger %age of the new roads. I've had a similar experience 
with new construction at Fort Irwin, CA and Victorville/Hesperia, CA. It 
seems to be worse for new construction areas, presumably because TIGER 
eventually gets some feedback to correct it. The issues are often obvious - 
look for mis-spellings of common words, lack of prefix or suffix (i.e. 
missing Via, Camino, St, Ave, Blvd), etc. Also look for names that are 
different from the pattern used in a given area, e.g., Piazza Maria, Piazza 
Angela, Piazza Gina, George, Piazza Donnatella. "George" is probably wrong 
here (missing prefix, non-Italian man's name).

I will say that TIGER11 road positioning and connectivity generally seems 
good now, even enough for GPS guidance.

Bing imagery is available down to zoom 19 (~26cm/pel) in virtually any area 
with a road, and even zoom 21 (~6cm/pel) in the most populous areas (though 
some of the patches close to downtown LA are "jagged", and look better at 
z19). I've confirmed that alignment accuracy is within about 5m of known 
benchmarks (personal surveys, FAA VORs, USGS benchmarks, CORS stations, 
etc.) at many points - enough to satisfy me that the imagery can be used.

--
Alan Mintz <Alan_Mintz+OSM at Earthlink.net>




More information about the Talk-us mailing list