"Gary E. Miller" <[email protected]> writes: > The 1 hour plot was nothing but RTKFLT. > > The 12 hour plot was: > > 146 "status": 2, DGPS > 552 "status": 3, RTK FIX > 139602 "status": 4, RTK Float > > So the two deviantions are probably, but not required to be, status > related.
That's interesting. Based on non-SPARTN reference data, I would expect the FIX points to be very tightly clustered at the cm level, and the FLOAT to have some wander at least 10s of cm, maybe more. I suggest 3 plots, one per subset, and if possible, to force the origin/true-postiion of FLOAT and DGPS to be the mean for FIX. Or, color the dots by mode or something, but I suspect the computed error statistics will be wildly different.
