[teqc] upcoming improvements to teqc's qc antenna position calculation
lou at unavco.org
Fri Mar 9 11:41:11 MST 2018
The first thing I'm going to ask you to do once the next official teqc version
is released is to get rid of your 2018Jan11 executable. Remove it. Delete it.
Expunge it from existence.
As some of you might know, I've been working to improve teqc's qc antenna position
calculations during the last year. The main corrections added in early 2017 were:
- the relativistic Sagnac correction to each SV position calculation,
- accounting for the rotation of the Earth during the travel-time of signals
from each SV to the antenna, and
- accounting for the time differences between the different time systems
used by different SV constellations (when delta UTC model information is
The last major improvement in the fall of 2017 was replacing the polynomial
position fitting code for GLONASS with the methodology prescribed in the
GLONASS ICD, i.e. a numerical integration of the equations of motion using
SV XYZ positions, velocities, and accelerations given in the GLONASS ephemerides.
This same approach was also adapted to also finally include SBAS data in the
position calculation when SBAS L1 and/or L5 pseudoranges were part of the observables
and when SBAS ephemerides were available. (Using the ICD methodology also
allowed the Sagnac correction to be added to GLONASS and SBAS SV positions.)
However, I appear to have been a little too focused on GLONASS and SBAS, so
a couple of bugs crept in which effected other the constellations. The best way
to describe the effect of the bugs was that qc position calculation in the
2018Jan11 version is quite "sub-optimal". The effect on the other qc parameters
is generally small, unless the sub-optimal position found was off quite a bit.
With those bugs now corrected, all thrusters are finally firing and in the right
direction for good position solutions. The following shows the results of teqc
qc positions for 1600+ global sites that we archive daily here at UNAVCO, differencing
the qc position of each site from the position that we have in our database (some
of which might be in error) -- after subtracting 15.7 meters from each teqc qc
position. (Keep in mind that the following includes solutions of all latitudes,
Antarctica to Greenland, and all elevations, from sea level to many kilometers above.)
QC Histogram N = 1646
distance count c/100m^3
-------- ----- --------
< 2.5 m 1053 1608.9
< 5.0 m 481 105.0
< 7.5 m 85 6.8
< 10.0 m 24 1.0
< 12.5 m 2 0.1
< 15.0 m 0 0.0
< 17.5 m 0 0.0
< 20.0 m 1 0.0
The first row shows 1053 vertically-adjusted qc positions within 2.5 meters of the
database position, the second row shows 481 vertically-adjusted qc positions between
2.5 and and 5 meters of the database position, and so on. The last column shows the
"density" of 3d differences per 100 cubic meters in each region, so within the 2.5-meter
radius, there are 1609 solution counts per 100 cubic meters; there is an order of
magnitude falloff in solution density for each 2.5 meter shell out to about 10 meters.
After the systematic vertical offset if removed, 1-sigma on the position appears
to be in the 2.5-3 meter range.
Most of these 1600+ sites are GPS-only, but a fair number have SBAS, GLONASS, and/or
Galileo data involved in the position determination.
What's the explanation for +15.7 meter offset in teqc's qc position? The best
explanation I've been able to come up with is that teqc assumes a vacuum between
each SV and the antenna, whereas in reality there are path sections of significant
delays in the ionosphere and the troposphere. But there's no consistent information
that's always available to teqc (via RINEX files or even raw data, in general)
that can be used to model the ionosphere or the troposphere. Probably because SVs
are fairly well azimuthally distributed over 24 hours, neglecting these path delays
almost average out azimuthally and end up only pulling teqc's position vertically up,
hence closer to the SVs (because of the vacuum assumption). The "best" offset
varies a bit from day to day, but so far the offset has been from 15.7 to 15.4 meters.
I've also noticed that there is a slight north-by-northwest statistical bias to these
1600+ qc position daily solutions, but this bias is well under the 2.5 meter range.
Another way of showing teqc position improvement is with a rapidly moving site: BATG.
BATG is in Antarctica, on the eastern Ross Ice Shelf, south of McMurdo, and is moving
a pretty constant velocity of +260 m/yr East and +620 m/yr North, or about 670 m/yr,
or about 1.8 meter per day. The daily solutions from the fixed teqc clearly show
this motion, day by day:
qc lat qc long qc elev
---------- ---------- -------
-79.005197 170.736879 51.8003 - qc of 27 Feb 2018 data
-79.005183 170.736897 51.4887
-79.005167 170.736943 50.6936
-79.005151 170.736967 50.4797
-79.005135 170.736990 51.1099
-79.005118 170.737044 51.7878
-79.005105 170.737052 51.1652
-79.005090 170.737109 50.3990
-79.005074 170.737128 51.1165
-79.005058 170.737165 51.2178
-79.005045 170.737199 50.7253
-79.005028 170.737226 51.0839
-79.005013 170.737258 51.5805
-79.004998 170.737305 51.0949
-79.004984 170.737321 51.3625
-79.004970 170.737356 51.2215
-79.004954 170.737385 51.1504
-79.004935 170.737413 51.4234
-79.004924 170.737438 51.2954
-79.004905 170.737488 50.5010 - qc of 8 Mar 2018 data
(The elevations above are teqc's raw qc elevations, without any vertical adjustment.
The actual WGS84 elevation of BATG is currently 31.8 meters.)
Anyway, I'll be curious to see how others find teqc's qc position improvements in
the next release.
Louis H. Estey, Ph.D. office: [+001] 303-381-7456
UNAVCO, 6350 Nautilus Drive FAX: [+001] 303-381-7451
Boulder, CO 80301-5554 e-mail: lou unavco.org
"If the universe is the answer, what is the question?"
-- Leon Lederman
More information about the teqc