[talk-au] GPS accuracy

[John Henderson]

On 20/09/12 22:27, Peter Hoban wrote:

> Discovering the accuracy of a unit is easy.  Find a convenient spot near
> your house and with your GPS record its position.  Come back next day
> (or at least a few hours later) and do it again.  Repeat daily until you
> are sick of it and you will then have a good idea of how accurate any
> particular observation is likely to be.  No technical expertise required.
>
> The question of absolute accuracy is complex.  Survey marks mostly were
> placed before the current modelling of the earth was developed.  While
> these may now have GDA coordinates (typically about 100 mm different
> from WGS in Australia) there are complexities that arise (eg from
> continental drift and the instability of the earth's axis of rotation)
> which are significant variables.  There are many assumptions in the
> modelling.
>
> WAAS also works in Europe and Japan.  There is no likelihood of it being
> implemented in Australia as our population density is too low.  Switch
> it off.  If it is left switched on there is some risk that spurious
> signals from other systems may degrade the accuracy of your device.

I've been using a Garmin GPSMAP 60CSx for a number of years.  Normally 
for logging GPS tracks by car for use in OSM, I use it in conjunction 
with an external antenna (mounted above the driver's seat so it's closer 
to the centre of the road).

With a good view of the sky, this GPS unit usually claims its accuracy 
to be ± 3m, with one important exception.  And that's when cornering.

If I superimpose track logs from several days in JOSM, I do see 
incredible consistency in the tracks.  The exception is in the corners, 
where there's distinct variation.

The solution is to drive around corners more slowly (where safety 
considerations permit).  Then the GPS seems more inclined to accept 
cornering as the reason for the deviation from going straight ahead 
(rather than its interpreting the change in direction as resulting from 
a noisy or degraded satellite signal instead).

I've struck another situation where the GPS reports significant 
uncertainty about its position.  That's when bushwalking with thick tree 
cover and especially with cliffs or hills to one or both sides.  It's 
clear to me that the signal is weak, with echoes only making the 
situation worse.  Here the GPS might report an accuracy in the order of 
±15m, and superimposed logs confirm this variation.

John