[OSM-talk] [tagging] Feature Proposal - RFC - incline up down

[Martin Koppenhoefer]

2009/8/22 John Smith <delta_foxtrot at yahoo.com>:
> --- On Sat, 22/8/09, Morten Kjeldgaard <mok at bioxray.au.dk> wrote:
>
>> Not really. There's a lot of elevation data available in the GPS traces, and since roads where incline=* is relevant are drawn along GPS traces, it's a matter of exploiting that data value. I'm aware that the GPS elevation data isn't terribly accurate on an absolute scale, but when determining inclines we will be making elevation differences which will decrease the error significantly.
>
> I doubt most GPS traces would be useful for this kind of thing, most vertical GPS data is +/- 10m under the best possible conditions, usually I seem to be +/- 20m most of the time, and it's not a stable 20m diff it jumps about just like GPS positions do.
>
> You'd need a GPS with an altimeter, or a stand alone altimeter, to do this kind of elevation differences to get accuracy better than up/down.

+1. E.g. the widespread 60 CSx from a well known manufacturer offers
this, but (I guess) almost noone does a calibration before every log
(at least I almost never do). That's why absolute elevation is not
reliable, but the relative data (i.e. difference from one junction to
another) is still "high"precision compared to GPS-elevation. I wonder
if some smart programmer could use this in some way for OSM. Provided
you have some reliable height-points, e.g. from public survey,
mountain-tops, other official signs, you could use this relative data
to calculate also the intermediate nodes.

cheers,
Martin