[OSM-newbies] GPS accuracy

[Robert Hart]

On 18/09/2007, Jeffrey Martin <dogshed at gmail.com> wrote:
> I went to Seoul Saturday on the bus and I have issues with accuracy.
>
> The satellite screen sometimes showed 11 meters or more. Is this kind
> of data useful?

Yes, all data is useful - it just depends what you want to do with it.
At an accuracy of 20m, you'd be able to put a road on a map with
reasonable accuracy. I wouldn't want to go editing some existing roads
without an accuracy of 10m though.

If on the other hand you are finding your accuracy is slipping to 30m
or more, then you'd probably want to make two or more trips down the
same routes at different times, and then load both traces into JOSM
and take an "average".


> I noticed after going through a tunnel that the accuracy would jump to
> 30 m when we
> exited, but on the mapping screen it didn't look like it was off that much.

30m isn't that far. I always like to try an visualise the error onto
my surroundings. Also remember that the accuracy reading is only an
estimate, there's no way the GPS can know what the true error is.
Sometimes the error is actually much lower, but other times (for
example if a GPS signal is reflecting off a building, or you are
moving around erratically) the error is really much higher. You will
sometimes have a better idea of where you are than it does!

> Walking in the city I sometimes got severe canyon effect.
>
> What strategies should I use in dealing with this kind of variable data?

When walking, there's lots you can do to improve the accuracy of the trace.

1) hold the GPS so that it can see the sky properly.
2) for mild cases of canyon effect it can help to walk closer to the
middle of the road in order to maximise the satellite visibility.
3) stop for a while in any more open areas, such as junctions, to get
a good position. some GPSs can accurately follow relative movement
with less satellites than you'd need to get a absolute fix.
4) slow down at corners.

Rob

-- 
Robert Hart