[Talk-us] Tagging Ridges as ways
steveaOSM at softworkers.com
Fri Jul 26 20:52:05 UTC 2013
I have tagged numerous ridges (dozens, at least), always as a way.
To me, it doesn't make sense for a ridge to be a single node, that
semantic might be better conveyed with natural=peak. I include a
medium-sized mountain range, the Santa Cruz Mountains
(www.osm.org/browse/way/174808173), almost 120 kilometers long, but
which I seem to have captured with only 22 nodes. True, I used
natural=mountain_range (rather than natural=ridge) on this. Neither
natural=ridge nor natural=mountain_range renders in mapnik/Standard
as anything besides text in the name=* tag. And that seems about
The proposed feature natural=massif (a group of mountains) has been
abandoned as inactive.
Just like any way (a road, a bike path...), use "as many points as it
takes to accurately convey the line of the way, but don't use too
many." Of course, these values are subjective, but there is a
certain middle ground to the calculus of curve-fitting. OSM (via its
human contributors) usually does a good job of finding this middle on
most ways, but there are exceptions in both directions (i.e. too many
points, as well as too few).
There are mapping tools (e.g. part of QGIS Desktop) which can help
you de-node/simplify over-noded ways. However, often the best way to
do these is manually with an extra imagery layer that lets you edit
the way while seeing a relatively sharp contrast of where the
ridgeline is. My example of a node every five or six km might be a
bit sparse, but it works. Find your happy medium: often, starting
with the highest peaks as a skeleton and connecting them with a way
is a good way to "find the line," then you just keep adding nodes as
you see fit until it is "accurate enough" for your taste. There are
such things as (national) map accuracy standards, but OSM doesn't
claim to adhere to them.
>seems to suggest that the appropriate way to tag a ridge is as a
>way. Any thoughts on this? I'm looking at 100+ ridges, wondering how
>to simplify (de-node) yet keep the way roughly following to contours
>of the feature it's supposed to be labeling.
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Talk-us