[Tagging] Marking waterway=brook as deprecated and problematic

[Stefan Tauner]

On Wed, 30 Dec 2020 21:18:23 +0100
Martin Koppenhoefer <dieterdreist at gmail.com> wrote:

> it looks as if you’re from Austria, and there it wouldn’t make sense to have a watercourse as shallow as 20cm and call it river, but in other parts of the world there are rivers that fall completely dry and are still regarded rivers.

Intermittent waterways was not our subject AFAICT. However I seriously
doubt that there are intermittent waterways that have less than 20cm
depth (when they do contain water) that are considered rivers (in OSM)
though.

> I also agree that many streams will be hard to jump over, but switching the definition to the amount of water per time seems not feasible for OpenStreetMap. 

Nobody suggested to use that as the (only) definition. I specifically
mentioned that this is hard to tell on the ground. My point is, that
current rule is not sufficient, subjective and should be refined and
more clearly specified. At the very minimum I would suggest defining
how far an able-bodied person can jump (i.e., specify the width where
one should definitely consider river)... but in general I think adding
some notion of depth as a second possible metric to help determining
the boundary would make a lot of sense without changing the meaning and
produce backwards-compatibility problems.

I think it would also make sense to give some hints on the length that
our metrics have to apply. If there is a short narrow, e.g., due to a
sandbank or rocks, this should not prompt a change of tagging
("temporarily"). I am not aware of any places where this has been done
and would be disputed but it does not hurt to give people some guidance.
Also related to this, we should write down where the change from stream
to river is supposed to be mapped, e.g., at places/nodes where the
junction of another waterway (or water from a spring) changes the nature
of the waterway instead of an arbitrary point downstream/in the middle
of nowhere just because the bed widens (just a suggestion, this would
obviously need more discussion/consensus).

IMHO the current rule has not been disputed because nobody really cares
about the correctness in this case (little harm done if it is not
completely off and since there is no clear out-of-OSM definition
anyway, what does correct even mean); and it is a very convenient rule
because it leaves a lot of margin for mappers but it leads to more
inconsistencies than necessary and leaves doubtful (new) users without
clear answers. I can actually remember that I felt quite disappointed
when I first read that definition years ago.

-- 
Kind regards/Mit freundlichen Grüßen, Stefan Tauner