[HOT] Comment to project managers of Building projects

john whelan jwhelan0112 at gmail.com
Wed Dec 27 23:39:01 UTC 2017


It was more of a heads up comment about data quality.  Certain projects use
building counts to estimate population for one reason or another.  The one
I spotted was about did it make financial sense to install solar panels.
So if you count the number of buildings this gives you an indication of
demand.

However for some areas if you count the number of buildings in OSM you are
counting the real buildings twice.  So if your project depends on knowing
an accurate number of buildings then I suggest you run the duplicate
building script to ensure none are counted twice.

So its very much aimed not at maperthon organisers but at the people who
use or analyse the data.  It's a way to strip out some of the errors.
Project managers perhaps.  The problem with maperthon organisers is not all
of them know how to map so they aren't a good target audience.  The project
managers need good accurate data so they are the ones who would be
motivated to do these sort of checks.

I don't think we need worry too much about it.  We have a tool to spot them
and I've probably deleted a thousand or more already and others will be
cleaning up the map so hopefully they'll disappear in time.

If we extended the time lock on a tile to four hours that might be enough
to prevent a few or if we can get people to upload every thirty minutes
that might eliminate some.  The real problem here is the number of extra
buildings that could have been mapped with the same amount of effort.  Say
it that way and you might catch a few maperthon organisers attention.

The problem with validation is as soon as you come up with a basic set of
tests then someone comes along with a new twist.  Using JOSM to check for
untagged ways, crossing highways, highways terminating close to another
highway, unclosed buildings, duplicate ways. e-arching for highway=road is
fairly quick and simple plus scanning the tile to see that everything has
been done to a reasonable standard.

Adding in things like highway=yes, or ways tagged highway=unclassified,
building=yes can be done but adds to the time taken when validating.

Cheerio John

On 27 December 2017 at 17:23, Bjoern Hassler <bjohas+mw at gmail.com> wrote:

> Hi John,
>
> Thanks for this suggestion.
>
> Would it be an idea to enhance the mapathon instructions, to include
> suggestions such as running scripts like these (at the validation stage, or
> perhaps run by the task owner)?
>
> If anybody wants to get together to discuss those planning/ best
> practice-related issues, I'd be very happy to  join in!
>
> Bjoern
>
> On Wed, 27 Dec 2017 at 19:12, john whelan <jwhelan0112 at gmail.com> wrote:
>
>> A snapshot of 10% of Uganda gave me 870 duplicate buildings. This is
>> where two buildings overlap by more than 50%. This is merely to give you an
>> idea of the scope of the problem.
>>
>> If you are relying on the data then running Mike Thompson's detect
>> building script will find any duplicates in your project.
>>
>> Typically I'm seeing clusters of up to 200 duplicate buildings in a small
>> area, that one was on a project to count buildings for a solar panels.
>> These are not uploaded twice but the buildings have been mapped twice with
>> small differences.
>>
>> https://github.com/MikeTho16/JOSM-Scripts
>>
>> Select Duplicate Buildings
>> Selects duplicate, or near duplicate, area buildings in JOSM's active
>> datalayer.
>> A "near duplicate" is a building whose area overlaps another building's
>> area by more than 50%. Only the first building encountered of an
>> overlapping pair is selected. This is done so the issue does not have to
>> be
>> looked at twice. The selected buildings are added to the current
>> selection.
>> Currently only works with buildings that are ways (not multipolygons).
>>
>> To Run:
>> * Install JOSM's Scripting Plugin (only necessary once)
>> * Place file in a convenient location on your system (only necessary once)
>> * Click "Scripting" (on top menu bar)
>> * Click "Run"
>> * Click "..." button and select the script file.
>> * Click "Run"
>>
>> If you decide to delete some I'd recommend downloading the area fresh in
>> JOSM then running the script, inspect each building deleting if need be
>> then upload the results quickly to reduce the risk of a another mapper
>> deleting the other building of a pair.
>>
>> Cheerio John
>> _______________________________________________
>> HOT mailing list
>> HOT at openstreetmap.org
>> https://lists.openstreetmap.org/listinfo/hot
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/hot/attachments/20171227/742ce910/attachment.html>


More information about the HOT mailing list