[Talk-us] Good work with remapping!
Alan_Mintz+OSM at Earthlink.Net
Tue Jul 24 07:54:34 BST 2012
At 2012-07-22 23:53, Toby Murray wrote:
>I also ended up doing some reimporting from TIGER 2011 in the Irvine
>area because some neighborhoods that were just too far gone to bother
>salvaging. For example here is a "before" shot of one area:
>http://i.imgur.com/pEuIm.jpg (and actually, a couple of those roads
>are already fixed. I happened to have a P2 instance open after I had
>uploaded the new stuff from JOSM and P2 picked up some of the new
>roads before I grabbed the screen shot)
Reminder: Orange County, CA has a decent source of official land and survey
info (including roads), though it is somewhat slow and only works in IE:
>Here is the current view:
>I decided to follow my previous TIGER remapping strategy and it
>worked out fairly well. Some of the neighborhoods were gated
>communities so they only had 2 or 3 roads connecting out to the rest
>of the network. This made it very easy to remove the existing garbage
>and import one section at a time. It's a lot more complicated to do a
>reimport in an area with a regular grid pattern of roads because there
>are so many more external connections to make.
BTW, this neighborhood is unincorporated OC, with USPS addresses in Newport
Beach, 92657. Three issues with two of the tracts I checked:
1. In these two tracts (TR15079 @ MM765/11-19 and TR15394 @ MM778/12-19,
both in RSB173/34-38), all the streets (except Ocean Heights Drive) should
have a prefix of Via, which was apparently missing in TIGER.
2. "Christallo", despite being shown as "Via Christallo" on the basemap, is
actually "Via Cristallo" according to the Tract Map, as well as the
addresses shown for the parcels. It's also wrong on the overview sheets of
the survey and tract maps, but is correct on the respective detail pages
(this type of discrepancy is, unfortunately, more common than you would
think). Finally, it is confirmed by looking up the 9-digit ZIP for an
address on the street (e.g. 7 Via Cristallo, Newport Beach, CA) on the USPS
confirming that it is Via Cristallo. This was one of those that "just
didn't look right", given the Italian spelling of the other street names in
the area, and the "h" in the incorrect "Christallo", which was confirmed by
the records. After making the correction and checking alignment with the
Bing imagery in the area, I tagged it with:
* Removed the tiger:reviewed=no tag
* Added a note tag about the naming discrepancies.
Note that I normally use the first date in the imagery range in my source
values, indicating the "worst case". In this area, however, the range shows
a "zero" date - "01/01/1980-08/30/2010" - so I used the ending date (in
yyyymm format - 201008) instead. I similarly aligned and/or checked and
tagged the other roads in the tracts.
3. There were a number of orphan nodes tagged with "highway=turning_circle"
left in the area, presumably remnants of the previous ways? In JOSM, these
can be found with the following sequence, though there should be a better
way that just isn't apparent to me at the moment:
a. Find, replace selection, search string="highway=turning_circle type:node"
b. Find, add to selection, search string="parent selected type:way"
c. Find, remove from selection, search string="child (type:way selected)"
d. Find, find in selection, search string="highway=turning_circle type:node"
This left 6 orphan turning-circles, which I deleted.
It's a good idea to turn off background imagery for a moment after wiping
an area to catch these leftovers that might otherwise be hard to see. Maybe
even do a select-all to make sure they are highlighted for you.
Alan Mintz <Alan_Mintz+OSM at Earthlink.net>
More information about the Talk-us