On 18/08/18 18:44, Christoph Hormann wrote:
On Saturday 18 August 2018, Peter Elderson wrote:
Sure. But is there a standard method to indicate this uncertainty in
OSM, which can be processed by data consumers?
This is outside the scope of of OSM IMO and not practical for mappers to
determine in a meaningful way. You can specify the method used to
determine the width (using source:width=*) and different methods will
have different inherent measurement accuracies but the accuracy of the
values depends on lots of other factors beyond that.
Practically mappers have usually no way to verifiably determine the
accuracy of their measurements unless they have a reference value of
known high accuracy in which case the whole excercise is pointless.
Theoretically mappers could practice their measurement skills and
assess their personal measurement accuracy based on test measuring a
statistically significant number of cases under comparable
circumstances with a known accurate reference value but practically
this is not feasible.
You can usually find set distances measured out for taxis to test their
equipment on, those are fairly accurate.
The equipment is one source of uncertainty. Farlarger sources can be the
way in which measurement is performed and the ambient conditions.
I did come across one case where 100 1 meter steel rules were submitted
for testing. One was rejected as being 1 mm out in that 1 meter ... the
machine that made it had missed a single 1 mm mark along the length.
That said, equipment accuracy is fairly good for most items, you would
have to go really cheap to get some apparatus that would have a large
uncertainty.
While the width of a canal may be seen as important to visual
inspections when water is present,
in terms of the water you would want to know the depth of the water,
shape and smoothness of the walls and the rate of fall.
If your not going to state those things then there is probably not much
point in the uncertainty of the width.
-------------- In summary.
Uncertainty statements in OSM are going to be fairly useless.
I don't see data consumers wanting them.
I don't see mappers wanting to state them.
I don't see either of them being able to effectively determine them or
use them.
_______________________________________________
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging
