[time-nuts] GPS position averaging
kb8tq at n1k.org
Mon Dec 15 19:39:47 EST 2014
> On Dec 15, 2014, at 7:19 PM, Bill Hawkins <bill at iaxs.net> wrote:
> This may be one of those cases where the human brain's ability to see
> patterns would help to see the distribution and eliminate outliers. Plot
> a manageable set of X and Y points, and another set of X and Z points.
> To make the three dimensions plot as one in a statistical distribution
> (bell) curve, use the magnitude of the vector.
> What do the surveyors instruments do that can get them down to a few
> centimeters in half an hour?
The ones around here use a survey grade GPS and log the data to a file. They do a bunch of post processing to get the solution to a desired level of accuracy. Last time I had it done, they had more fun looking at the GPSDO’s than running their fancy GPS.
> You might also want to study what's been said about the causes of
> variance. The atmosphere is not uniform.
Ionospheric issues are one of the things that gets corrected for to make a more accurate result.
> Consider the relative effect of position on time, where one foot or 30
> cm is equivalent to 10E-9 second.
…. and some of the errors between solutions are > 10 feet.
> Also consider the 'averaging' period of the PLL filter in the
> disciplined oscillator.
The problem shows up when the GPS constellation is in a configuration that accentuates the error. That’s often a 12 / 24 / 48 hour sort of thing.
> I'm no expert, especially not in math. These hints are thoughts that
> come to mind.
> Bill Hawkins
> -----Original Message-----
> From: Ole Petter Ronningen
> Sent: Monday, December 15, 2014 3:04 AM
> I would like to get a better idea of what my actualt antenna location
> is, so that I can manually set the position in the GPSDOs. For that
> purpose I set up a Ublox M8N with logging in U-center, and I've
> collected about 250K readings over several days. For each reading I also
> log number of SV's used, as well as HDOP, VDOP. PDOP, thinking I would
> filter the list and end up with a subset of really good fixes.The idea
> is then to average the "best readings" and use that as my position.
> My question is twofold; 1) is this for some reason a bad idea? And 2)
> How do I average the numbers? I can not put my finger on it but it
> "feels wrong" to average lat, long and height independently.
> Any hints are greatly appreciated.
> time-nuts mailing list -- time-nuts at febo.com
> To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
> and follow the instructions there.
More information about the time-nuts