[OSM-talk] HOT Community Webinar - Wed July 26th 3:00 PM – 4:00 PM UTC

john whelan jwhelan0112 at gmail.com
Tue Jul 25 16:24:28 UTC 2017


In an ideal world there would be as many skilled mappers per square
kilometre as there are in Germany.

Unfortunately the density of mappers available elsewhere is not the same.

In many parts of the world there are very few maps and these are not of a
high quality.  HOT and the missing maps projects are bringing resources to
map many of these areas.  Ideally the armchair mappers' work is validated
as they map.  This validation process is not part of the normal OSM
workflow.
http://wiki.openstreetmap.org/wiki/OSM_Tasking_Manager/Validating_data I
think everyone in HOT is sensitive that high data quality is something to
strive for.

I note that the original webinar was mailed to the HOT mailing list but
your reply was to the OSM talk mailing list.

HOT armchair mappers I think are in the business of roughly mapping the
area.  To add street names etc you need boots on the ground but its a lot
easier to adjust and refine if the basic ways are already there.  The HOT
community has come up with some guides for armchair mappers.
http://wiki.openstreetmap.org/wiki/Highway_Tag_Africa which include the
words "The road conditions in African countries do not always correspond to
their economic and social role. A road typology should be based on the road
importance and not on the surface or the visual appearance of a road."  In
other words it is accepted an armchair mapper will have difficult tagging a
highway perfectly.

There have been discussions about how far you can tag a building.  The NGOs
would very much like to know which are residential.  My feeling is you can
label a building=yes from imagery but after that you can go no further.
landuse=residential I think the convention is if you see a settlement with
side streets in a grid pattern its more than likely a settlement and can be
labelled landuse=residential.  If it looks like a village then tagging it
landuse=residential is reasonable. If a group of warehouses get mapped and
labelled landuse=residential at least the tags can be corrected with a
smartphone at a later date by boots on the ground.  We are seeing more and
more boots on the ground by the way.  It's fun to see cafes and other
buildings suddenly sprout tags.  Project managers, well some are better
than others.  Again HOT is setting up training to bring the standards up.
Many do not have a background in GIS, they just want to find out where the
population is so they know how many vaccines to order and where to deliver
them.  Without reliable census data populations can be estimated by
multiplying the number of buildings by the average number of people per
building.  It is an estimate.  If you don't have a building count then is
there a settlement there and roughly how big is it?

Training is a problem, I think in Nepal 70% of the mappers were mapping for
the first time.  The quality was not good.  What HOT has concentrated on
are fairly simple things of value to the NGOs etc which I think is the
right thing to do, so highways, buildings and settlements are priorities.
Fences, cattle pens and trees are of a lower priority.  Volunteers are
giving say three hours of their time to map are happier the sooner they can
start mapping.  Spending two and a half hours being trained for thirty
minutes mapping isn't practical.

Having said that the maps that are being produced are valued as being
better than what was available before.  They are having a positive impact
on life.  HOT is always looking for ways to improve the instructions and
training to new mappers.  For many their first language is not English,
French or German.  LearnOSM http://learnosm.org has had a lot of input from
the HOT community.

We are seeing more boots on the ground, in Nepal Kathmandu Living Labs
recognise the need for better mapping and they know GIS very well.  In
Zambia, Africa I'm seeing joint projects with the local government using
their GIS specialists with a mixture of armchair mappers and boots on the
ground.  The armchair mappers can cover a large area fairly quickly but the
local knowledge helps as well. The locals have a much better idea of the
priorities to be mapped. That one was fun the city thought there might be a
thousand buildings in one section of the city, the project mapped six
thousand which had implications for schooling and other services.

The roots of OSM has always been the local community.  Currently I think
you'll find that the local communities in Africa and other places are quite
happy to accept groups of buildings being tagged as landuse=residential
provided they have the correct the tags if need be.  If they are unhappy
with the present tagging process then I feel sure that HOT will take their
opinions into account.

If you can identify an issue then please identify it and people such as the
HOT training group will see if it can be clarified but to say only those
settlements in Africa that have been verified by boots on the ground which
is what you appear to be saying is an issue, is more difficult.  More than
95% will be correctly tagged.  OSM never has been prefect, I recall a very
negative reaction when locally I retagged highway=traffic_light to
highway=traffic_signals as per the wiki map features.

Cheerio John

On 25 July 2017 at 09:54, Christoph Hormann <osm at imagico.de> wrote:

> On Monday 24 July 2017, Gysler Castelino wrote:
> > [...] Typically, Missing Maps
> > (see e.g. http://tasks.hotosm.org/) often includes the instruction to
> > map (1) buildings, and then (2) group buildings into residential
> > areas. However, the 'residential area' definition is tricky, and
> > volunteer mappers don't apply it consistently, plus it may cross your
> > task boundary. So we'd like to find a way to consistently add
> > 'residential area outlines' after the buildings have been added.
>
> It should be clearly said that the reason for widespread incorrect use
> of the landuse=residential tag in HOT project areas are the incorrect
> instructions typically given in HOT tasks to mechanically cluster
> buildings and draw abstract wrapper polygons around them independent of
> both the primary purpose of the buildings and the actual use of the
> land.
>
> I am frequently amazed by the complete lack of distinct mentioning of
> the need of verifiability for information entered into the OSM database
> (both geometries and tags) in HOT instructions.  Mapping instructions
> like in [1] to me seem just schocking - not only in light of the
> projects in question but also because of what is presented there to OSM
> newbies as good practice.
>
> Fixing that is not a matter of technological measures and mapping tools,
> it is a matter of developing and communicating a mapping culture based
> on OSMs good practice [2].
>
> [1]
> https://docs.google.com/presentation/d/1PR7xn9ooqguswP0PuzMPD_
> hYHiLsr3THbtlhVBXL-Lk/edit#slide=id.ge869b0762_2_195
> [2] http://wiki.openstreetmap.org/wiki/Good_practice
>
> --
> Christoph Hormann
> http://www.imagico.de/
>
> _______________________________________________
> talk mailing list
> talk at openstreetmap.org
> https://lists.openstreetmap.org/listinfo/talk
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk/attachments/20170725/32111faa/attachment-0001.html>


More information about the talk mailing list