[OSM-talk] GPS Accuracy under Forest Canopy

[Paul Houle]

Stephen Hope wrote:
> I've done some rain-forest hiking, and I've noticed similar results.
> If you really want to see some wandering tracks, try hiking along the
> base of some cliffs, in dense forest.
>
> I have noticed that the errors do seems to be less the faster I'm
> moving.  If I stand in one place for a while, the path can wander over
> quite an area if there is dense cover.  If I walk fairly quickly, then
> it still has errors, but not as large.  I think it must be finding
> more open patches and correcting itself more often.
>
>   
    My Garmin eTrex HCx makes reasonable tracks under forest cover,  
although the tracks are certainly worse under forest than under a clear 
sky.  It's not the cheapest GPS unit you can get,  but it's reasonably 
priced and it's a great navigator to enjoy both OSM and commercial maps 
on foot or sitting in the passenger seat of a car.  The ability to see 
my own track has gotten me "unlost" more than once;  it seems that once 
I've gotten into GPS mapping I've been more aggressive about going into 
unfamilliar and confusing terrain,  so I've been getting lost more!

    I think of track accuracy from a practical viewpoint.  Having a 
trail off by 20 meters isn't so important so long as I get the topology 
right.

  I walked a segment of trail that followed a creek and always stayed by 
one side:  when I looked at the tracks overlaid with Garmin's Topo 
2008,  I saw the track crossing the creek.  I was often within 10 meters 
of the creek,  so this isn't 'crazy'  If I'm loading this into OSM and 
if the creek is there,  I certainly feel pressured to manually push the 
trail across the creek so that the trail doesn't show false creek 
crossings:  that's an error that people when they're using the map and 
could even cause confusion.

    As for speed,  it's an issue that GPS errors have a "brown noise" 
characteristic:  they look worse on longer timescales.  If you're 
standing at one place and your GPS seems to be swirling around in lazy 
nested circles,  it looks real bad.  It's hard to average the 
coordinates to get a betting point position.  If you take a track or go 
walking for 4 miles or drive 40 miles in your car,  that craziness is 
still there,  but it's made invisible by the scale of the map.