Hello Pierre, hello John!
Thanks a lot for your quick and detailed replys! I did't expect a
"simple" answer anyway ;-)
Your hint, John, that HOT-mapping is more standardised makes the
guidelines (http://wiki.openstreetmap.org/wiki/Highway_Tag_Africa) more
than only a rough orientation for us. And I agree with you, the most
important tasks to work on is the traffic network and that's what we
actually concentrating on. We're just trying to figure out how to do it
"as right as possible". I found conventions that seem to underline the
HOT-approach:
"In some situations it is however appropriate to base the tag value on
the physical attributes of a highway:
- The official classification for the road is not known or easy to
establish.
- The administrative classification for a road does not match the
physical attributes for some historical reason."
(https://wiki.openstreetmap.org/wiki/Highway_tag_usage)
So, is there a general favour for tagging based on physical attributes
instead of administrative levels? Would a wide frequently used and well
biult county-level road get a secondary-tag although it belongs to the
3rd administrative level?
In fact, our work has been initiated by a group of Cameroonians running
a small non-profit association for setting up connections and provide
support in terms of research and education. They are helping us
translating our questions and transfering them to local people they
know. So we asked about official criteria the government uses for
classifying their roads and for establishing a network hierarchy. We've
been told that the guys have already approached the administration and
are waiting for an answer. What ever this answer might be, we hope that
some of the criteria we'll get back can be applied in our work based on
satellite images. But we're still waiting ...
John, you've mentioned a "list of tasks". Where can we find this list?
Does this list indicate, what features or regions are "under
contruction"? We found a reasonable amount of "undefined" objects
(mostly roads) in our area which are not visible in OSM or the
iD-Editor. Is there someone working on these objects? What shall we do
at locations where we know that these exist? I'm afraid of finally
ending up either with an inhomogeneous, incomplete mapping result or
with objects being mapped twice after the "undefined" showed up again.
So far for now! Thanks again for your support! kamak
Am 2014-11-09 19:49, schrieb john:
I'm currently mapping #684 - Polio outbreak and Ebola preparedness,
Meiganga, Cameroon and I must confess I have questions about the
project.
I can understand French especially a European or African accent so the
video was useful to some extent.
To me HOT is different to normal OSM. In normal OSM people basically
do
their own thing. Sometimes they use "standard" tags but there is a
wide
range and style of tagging.
HOT prepares maps for a particular target audience so to my mind takes
a
more standarised approach.
What I haven't seen is a list of requirements.
There is a list of tasks but that isn't the same as a list of
requirements.
For Meiganga there are thousands of buildings to be mapped. I
understand
the idea that ideally each building outline shape should be carefully
mapped but realistically with the JOSM building tool I can approximate
the
building size in two or three seconds. To carefully trace the outline
takes me twenty seconds or so. If I look at some tasks I see someone
has
mapped three buildings then given up. They are beautifully mapped but
when
there are another 98+ buildings to map in the task and another 180 odd
tasks to do? Yes we are using volunteers so their time doesn't cost us
anything but mapping buildings is tedious and how fast do we want the
information to be made available and how accurate do we need it? What
exactly is the requirement?
I think for this you need to go back to the AID agencies and the
Cameroon
government cartographers and get them to make a list and set
priorities.
Can we tackle the tasks differently? The road network and water really
need to be done first. It's better to have as few a number of segments
in
a read as possible. That way when you tag the name you only need do it
once. Water, in a task its difficult to see if its a river, ditch or a
clump of trees sometimes. From further out you stand a better chance.
Also if we break the tasks down then the grunts, sorry less skilled
volunteers, can tick off the task as done when they've mapped all the
houses and paths. If they are daunted by the idea that they have to
map
all the "residential and non-residential" buildings and forests before
marking the task done they maybe reluctant to tick the box and we end
up
with lots of tasks mapped but not ticked as done.
Then we get to the quality of the map. It sounds dumb but different
satellites have different accuracy. DigitalGlobe’s WorldView-2 is one
of
the better ones. There was presentation by *Kevin* Bullock on the
subject
at a SOTM US recently but the video seems to have disappeared. If you
can
find it the relevant bit is 15 mins in for 90 seconds. If you look at
in
task 684 you'll notice that the Bing imagery and the Mapbox imagery
don't
quite line up. Some mapping has been carefully done from Bing and some
from the DigitalGlobe imagery.
Can we clean the data up? Interesting question, from a satellite
imagery
I'm unable to tell if a building is residential or not, however many
buildings mapped from satellite imagery are tagged building=house
rather
than building=yes. I would suggest that if building=house is tagged
from
satellite imagery this be changed to building=yes by bot but only if
its
the initial tag on the building.
Some small blobs might be a car or an outhouse. Perhaps it might be
worthwhile to scan for a minimum size building?
If we go back to the idea of requirements again it seems likely that to
get
a better map we need someone on the ground. worldbicyclerelief.org do
reasonable bikes for Africa, I'm from a technical background so I like
the
idea of some sort of computing device to enter data on. Smart phone
perhaps
or can we work with one of the local schools? I assume that Internet
access is not ideal but text messaging might work. I'd envisage
compressing / encoding the information so it fitted into the
constraints of
text messaging to get the updates back. It would need some programming
effort on the device and at the other end but there are a large number
of
programmers around OSM. This may well be already sorted out but as a
mapper I'd like to think that my efforts were used rather than left
waiting
for someone on the ground to do their bit. I might even dump some cash
into a charity that could sort this sort of stuff out and yes I know
its
not as instant as a bag of flour but it is important to have the
infrastructure in place.
Cheerio John
_______________________________________________
HOT mailing list
[email protected]
https://lists.openstreetmap.org/listinfo/hot
