[time-nuts] GPS position averaging

Dave Daniel kc0wjn at gmail.com
Mon Dec 15 05:36:41 EST 2014

```I am new to this list and to this topic, but it seems to me that if one
wants to come up with an average of a set of spatial measurements, one
would use distance as the parameter to be averaged. The distance would
presumably be that from a fixed spatial reference point (0,0,0). One
would then take the square root of the sum of the squares of latitude,
longitude and height. My problem with this is that I can't see what one
would use for the reference point, but maybe that is not important (or
maybe it is!).

DaveD

On 12/15/2014 2:48 AM, Hal Murray wrote:
> olepr01 at gmail.com said:
>> 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.
> It seems like an interesting idea to me.
>
> It would be interesting to run a test: setup a pair of identical GPSDOs
> running off the same antenna but using different locations and see if you can
> see any difference in the PPS or 10 MHz output.
>
> You can do a filtering pass to dump anything with fewer than 4 satellites.
> The survey code has a mode where it says "bad geometry".  I don't know the
> fine print on how that works.
>
> I can get my elevation off a topo map.  I'm not sure how to translate that to
> GPS coordinates.  Something like that might be another opportunity for
> filtering.
>
> The refclock part of ntpd filters a batch of samples by discarding outliers.
> Time is only one dimension, so the recipe is pretty simple: sort, compute the
> average, discard either the top or bottom, whichever is farther from the
> average.  I'm not sure how to do something similar in 2 or 3 dimensions.
>
> It might be interesting to make histograms of HDOP and friends.  (before and
> after any filtering you can come up with?)  That may cover the bad-geometry
> case.
>
>
>
>
>

```