[HOT] Ocean in Latin America

Philippe Verdy verdyp at gmail.com
Fri Mar 20 05:13:55 UTC 2020


May be, but then the riverbeds for the Cassiquiare should not be part
of the separate multipolygons for the Amazon or for the Orinoco.
So these multipolygons has to be fixed. It's very likely that Amazon
and Orinoco now share incorrectly some riverbeds. They must still be
completely disconnected, mapping the Cassiquiare has to be separated,
the 3 rivers should only have a common (unclosed) bordering line where
they join in each pair.
Note also that the tag waterway=riverbed has been deprecated
The tags "waterway=*" are now only for the linear oriented streams and
riverbeds should now be areas using "water=*".

But for the Cassiquiare, it's not clear in which direction it is
flowing (like for canals) as it would be a tributary for both the
Amazon and the Orinoco (which is OK): the main stream can be split in
two parts (quite arbitrarily somewhere in the middle between two river
junctions) so the tributary links can be OK. It's quite hard to
determine where is the cutting point (it may be over some lake area at
the junction of another smaller river or one of the source streams,
but there are also so many branches and islands in the Amazon area
(let's remember that the Amazon was initially a sea area in old times,
the terrain is very flat, and it's the growth of the maritime mangrove
that created the many islands and made the area partly emerge; then
the salinity of waters decreased with the mass of rain waters; the
resulking rivers between islands are flowing very slowly, only
regulated by the differencial levels of water and local weather
conditions of tropical rains; the water flows may change of direction
almost everywhere except on the easternmost parts with the mountains).

Fixing river directions should be made according to elevation profiles
in the eastern part (exactly in the area where the Cassiquiare
junction occurs: it's possibly not easy for the main stream of the
Cassiquiare, but should be possible for its smaller tributaries):
satellite imagery is not sufficient, you need also a map of elevation,
or some local measurements of elevations to compare and set the most
frequent flowing direction of ways for river streams. However this is
compeltely independant of riverbeds (water=* areas) that have NO
direction.

Le jeu. 19 mars 2020 à 18:38, Manfred A. Reiter <ma.reiter at gmail.com> a écrit :
>
> Hi did not checked it,
>
> but a connection between Orinoco and Amazonas exists! It's the Cassiquiare. (Not sure about the correct wording)
>
> ## Manfred Reiter - mobile -
> ## please excuse typos and brevity
> ## http://weeklyOSM.eu
>
> Philippe Verdy <verdyp at gmail.com> schrieb am Do., 19. März 2020, 12:28:
>>
>> There's certainly a problem in the way the multipolygon for the
>> Orinoco river was created, without importing the water tags to it.
>> And there's a suspiscious connection between Orinono and the Amazon
>> via a tributary, creating an "island" in it. This multipolygon
>> requires fixes (Osmose and other QA tools can help detect where this
>> happens).
>> I suspect that the multipolygon for the Orinoco River was modifioed
>> recently to add too many riverbeds in it and not really belonging to
>> it. This multipolygon is certainly broken.
>> I joined the OSM French talk list. There may be QA tools to do that
>> (it's not easy to fix as this covers a very large area, only JOSM
>> experts can locate it, with the help of QA tools to locate the broken
>> areas and superfluous tags. This will require loading lot of data, and
>> JOSM running in a 64-bit Java VM with enough memory, plus a solid PC.
>>
>> Le jeu. 19 mars 2020 à 18:20, Jorieke Vyncke
>> <jorieke.vyncke at gmail.com> a écrit :
>> >
>> >   Interesting!
>> > So is this an issue that can be fixed by the Humanitarian layer OSM France team? Or is it just a matter of updating OSM and waiting for the humaniarian layer to render it correctly?
>> > Thanks, Jorieke
>> >
>> > Op do 19 mrt. 2020 om 16:47 schreef Philippe Verdy <verdyp at gmail.com>:
>> >>
>> >> Most probably this is the "water bassin" of the Amazone river, which
>> >> was tagged incorrectly with some "water=*" that causes problems in
>> >> this rendering.
>> >> Water bassins for rivers (which do not include only riverbeds and
>> >> lakes/ponds, but also all surrounding lands whose drained waters on
>> >> soil are converging to rivers) should not use this tag.
>> >> This does not cause a problem however in the OSM Carto rendering. If
>> >> that tag was approved, then the rendering for humanitarian map should
>> >> be fixed (it is maintained by OSM France).
>> >> But if I look at the boundary, I only sea ways for small riverbeds.
>> >> So it is likely that some multipolygon for riverbeds areas of some
>> >> river has been broken and the renderers attempt to "close" it due to
>> >> holes, or that someone joined all these riverbeds into a single
>> >> multipolygon.
>> >> Given the size of the relation where it is used, this cannot be fixed in iD.
>> >> Note also that given the current delays in the OSM data servers for
>> >> data replication, this may be temporary and caused by lack of
>> >> synchronization of the slave database used by the French renderer for
>> >> HOT.
>> >>
>> >> Le jeu. 19 mars 2020 à 17:11, Jorieke Vyncke
>> >> <jorieke.vyncke at gmail.com> a écrit :
>> >> >
>> >> > Hello,
>> >> > Is there someone who knows why several countries in Latin America look like ocean on the humanitarian layer on OpenStreetMap? Check here: https://www.openstreetmap.org/#map=5/3.119/-61.436&layers=H
>> >> > Can someone fix that?
>> >> > Thanks!
>> >> > Jorieke
>> >> >
>> >> >
>> >> > _______________________________________________
>> >> > HOT mailing list
>> >> > HOT at openstreetmap.org
>> >> > https://lists.openstreetmap.org/listinfo/hot
>> >
>> > _______________________________________________
>> > HOT mailing list
>> > HOT at openstreetmap.org
>> > https://lists.openstreetmap.org/listinfo/hot
>>
>> _______________________________________________
>> HOT mailing list
>> HOT at openstreetmap.org
>> https://lists.openstreetmap.org/listinfo/hot



More information about the HOT mailing list