[Talk-us] [Imports] Importing Buildings and Addresses for Austin, Texas

Eric Ladner eric.ladner at gmail.com
Tue Nov 10 03:22:05 UTC 2015


I'll second Paul's bit rot on unique ID's comments.   They're all well and
good until somebody upgrades their GIS system, then all those unique ID's
get flushed down the drain.  And actually doing a conflation with hundreds
of thousands of buildings which may or may not have the ID's preserved over
years of editing is a daunting task in itself.

Also, I wouldn't be so concerned of the existing buildings.  Of the 4000 or
so existing buildings I looked at in Austin (and a good percentage of those
were the accidental import ones that I couldn't filter out), 90% of them
are poorly drawn contain only "building=yes" on them.  Most are hand drawn,
not orthogonal and only vaguely representative of the buildings on the
ground.

With the exception of the university buildings, it would be worth the time
to import the whole data set from CoA and find the conflicts at import time
and replace the existing geometry with the GIS system data.  I conflated
hundreds of buildings in the New Orleans import manually.  Time consuming?
Yes.  Worth it?  Yes.  The address data and decent building footprints
alone are worth it.

My 0.02.

On Mon, Nov 9, 2015 at 8:40 PM Andy Wilson <wilson.andrew.j at gmail.com>
wrote:

> On Mon, Nov 9, 2015 at 12:59 PM Martin Koppenhoefer <
> dieterdreist at gmail.com> wrote:
>
>>
>>
>> sent from a phone
>>
>> > Am 09.11.2015 um 18:07 schrieb Eric Ladner <eric.ladner at gmail.com>:
>> >
>> > Replacing existing hand drawn buildings with imports from the city's
>> GIS system would probably be preferable given the quality of the hand drawn
>> buildings.  (compare the outline in [1] to satellite imagery).
>>
>>
>> it might depend, unlike the city's data the things in osm may vary a lot
>> in level of detail and quality. Also there could be other information
>> attached to these buildings so they should be carefully reviewed before any
>> deletions are executed.
>>
>>
>> cheers
>> Martin
>
>
>
> Thanks for taking time to look things over. Adding to what Martin said,
> some of our reasons:
>
> 1. From the belief that individual contributions are more valuable as a
> whole than contributions from an automated import process. We are trying to
> grow a local OSM community in Austin, and I feel that throwing away the
> work of past contributors works against that.
>
> 2. The data we are importing was collected about 3 years ago at this
> point, so OSM data will be more relevant and up to date in some cases.
>
> 3. Efficiency. There are enough buildings in OSM already and as Martin
> pointed out, comparing them individually and properly merging tags, etc
> would be time consuming.
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk-us/attachments/20151110/68d56cec/attachment.html>


More information about the Talk-us mailing list