[OSM-dev] Duplicate county - Huntingdon, PA

[Andy Allan]

On Tue, Jun 23, 2009 at 6:45 PM, Frederik Ramm<frederik at remote.org> wrote:
> Hi,
>
> Andy Allan wrote:
>>
>> Anyone fancy a challenge? It looks to me like an entire county has
>> been duplicated in Pennsylvania - Huntingdon, PA. Looks like Milenko
>> and DaveHansen both uploaded copies a day apart from one another.
>
> How could this have gone unnoticed for so long?

I've made some more investigations, and it looks like some fairly
extensive problems in Pennsylvania. The list that I've come up with
(using some postgis jiggery pokery and checking each on the data
layer) is that there are 16 counties duplicated. See
http://wiki.openstreetmap.org/wiki/TIGER_fixup#Duplicate_Counties for
the list.

Unfortunately none of them are exactly duplicated, i.e. I suspect some
areas have been fixed up a bit, but only partially. That makes it
impossible to 'trivially' remove one copy of the road network without
destroying topology. Something more advanced may be needed. I was
thinking along the lines of identifying (somehow) identical-twin
roads, then pick one to remove and stitch all the attached roads onto
the remaining roads. Then repeat until no twin roads remain.

Aaargh. I wish someone had noticed much earlier - before everything
was "partially" repaired! For those interested I reckon there's
approx. 86,600 ways across those counties that need de-twinning. If
anyone can find previous discussions on this problem please let me
know.

Cheers,
Andy