[OSM-talk] TIGER Import Update

[Ben Gimpert]

Grr,

For those of you following the TIGER -> OSM import progress, I have an
update:

I've gutted the import code (again) to support prioritizing certain
counties in the county over others.  As much as I love Alaska, there are
a good propaganda reasons for starting with counties like San Francisco,
Cook (Chicago), and Washington DC.

Here's an ordered list of which counties will be done first:

        06075, San Francisco  <-- going as I write this
        06097, Sonoma (greater San Francisco, CA)
        06041, Marin (greater San Francisco, CA)
        06001, Alameda (greater San Francisco, CA)
        06085, Santa Clara (greater San Francisco, CA)
        06081, San Mateo (greater San Francisco, CA)
        17031, Cook (Chicago, IL)
        36061, New York (Manhattan)
        11001, District of Columbia (Washington DC)
        (...then the rest, ordered by state code like "AL" for Alaska)

Here's San Francisco, gradually getting more "grid-y":

http://www.openstreetmap.org/index.html?lat=37.79947548999389&lon=-122.4065783203125&zoom=11

Also the TIGER log now has better progress statistics for folks wanting
an answer to "How long will it take?" finer grained that my "A long
time...  Verily legion."  A quick bit of math on the last half hour's
progress shows about 1.32 streets per second, so about four hours for
the rest of San Francisco, and about 5 months for the entire import (!).

These numbers could be *hugely* wrong if San Francisco's streets are
particularly more "segment-y" than most places.  Which is probably the
case.

Of potentially more interest, folks may be interested to know that the
TIGER import is now adding a "highway" key on ways, with a value mapping
based on:

        http://wiki.openstreetmap.org/index.php/Key:highway

And specifically:

        CFCC code "A00" => "unclassified"
        "A1?" => "motorway"
        "A2?" => "primary"
        "A3?" => "secondary"
        "A4?" => "residential"
        "A71/2" => "footway"  <-- look here, Nick W.

(See "http://www.census.gov/geo/www/tiger/tiger2005fe/TGR05FE.pdf" for
more details.)

If someone more familiar with the TIGER format knows a better mapping
between our "highway" key and a corresponding TIGER CFCC code, please
let me know.  I just went with the quickest.

Oh and eventually I'll re-import Alaska and (some of) Alambama with the
"highway" key values.

And of course, from my fellow Americans I'll happily take bribes to bump
YFC (Your Favorite County) up in the priority list...  Except Texas.

                Cheers,
                Ben