1. WHY DOES THE IGS CARE ABOUT EOPs?
Summary of core products of the International GNSS
Service (IGS)
Ultra-Rapid (real-time), Rapid, & Final series
outputs: orbits, polar motion/LOD, clocks, & station positions
Ultra-Rapid products very widely used for many
demanding real-time applications
e.g., very rapid tropo water vapor soundings for meteo models
& natural hazards monitoring
Ultra-Rapid product quality depends on EOP
prediction accuracy
latest observed orbits projected into future with EOP predictions
EOP prediction errors limit accuracy of IGS real-time orbits
Jim Ray
IGS Analysis Center Coordinator
NOAA/National Geodetic Survey
NGA Future EOP Prediction Workshop, Springfield, VA, 17 November 2011

2. IGS Core Product Series (2011)ID
Latency
Issue times
(UTC)
Data spans
Remarks
Ultra-Rapid
(predicted half)
IGU
real-time @
03:00, 09:00, 15:00, 21:00
+24
00:00, 06:00,
12:00, 18:00
● for real-time apps
● GPS & GLONASS
● issued with prior IGA
(observed half)
IGA
3 - 9 hr
03:00, 09:00,
15:00, 21:00
-24
● for near real-time apps
● issued with following
Rapid
IGR hr
17:00 daily
±12
12:00
● for near-definitive,
rapid apps
● GPS only
Final
IGS d
weekly each
Thursday
12:00 for
7 d
● for definitive apps

3. IGS Ultra-Rapid Update Cycle
day 1
day 2
day 3 ●
00h
06h
12h
18h
= 24 hr of observations
= observed EOPs
= 24 hr of predictions
= predicted EOPs
IGU updates every 6 hr are always 3 hr after the beginning of each prediction interval

4. IGS Core Product Accuracies (2011)
Series ID
Product Types
Accuracies
Output Intervals
Ultra-Rapid
(predicted half)
IGU
● GPS orbits
~ 5 cm (1D)
15 min
● GLONASS orbits
~10 cm (1D)
● GPS SV clocks
~3 ns RMS / ~1.5 ns Sdev
● EOPs: PM + LOD
~250 µas / ~50 µs
6 hr
(observed half)
IGA
~ 3 cm (1D)
~5 cm (1D)
~150 ps RMS / ~50 ps Sdev
<50 µas / ~10 µs
Rapid
IGR
~2.5 cm (1D)
● GPS SV & station clocks
~75 ps RMS / ~25 ps Sdev
5 min
<40 µas / ~10 µs
daily
Final
IGS
<2.5 cm (1D)
<5 cm (1D)
~75 ps RMS / ~20 ps SDev
30 s (SVs) + 5 min
<30 µas / ~10 µs
● Terrestrial frames
~2 mm N&E / ~5 mm U
weekly
IGS aims for ~1 cm orbit & ~1 mm terrestrial accuracies
to satisfy most demanding mm-level user application requirements

5. Errors in obs EOPs ~ cancel out in forward/reverse transforms
Rotational Transform:
Observed EOPs(t)
Observed orbit:
Crust-fixed frame
Observed orbit:
Inertial frame 1)
+ Projected orbit:
Inertial frame
Observed orbit:
Inertial frame 2)
Rotational Transform:
Observed + Predicted EOPs(t)
Observed +
Projected orbit:
Crust-fixed frame 3)
Errors in obs EOPs ~ cancel out in forward/reverse transforms
but EOP prediction errors fully embedded in crust-fixed orbit predictions
typical prediction errors: ~0.4 mas/d for PM; 0.1 ms/d = 1.5 mas/d for UT1
0.1 ms = 1.5 mas = 4.6 Earth = GPS

6. Ultra-Rapid AC Orbit Comparisons (over 48 hr)
Performance among Analysis Centers has become bimodal
SIO & USNO have been excluded for >2 year
AC quality is more uniform over first 6 hr of orbit predictions
biggest differences occur for 6 – 24 hr orbit predictions

7. Some IGU AC Orbits Have Large Rotations
0.5 mas = 64 mm
GPS hgt
SIO & USNO have large Z rotational errors; also Y
CODE sometimes also has moderately large Z rotations
these AC rotations probably from poor orbit modeling, not EOP predictions

8. Ultra-Rapid Orbit Diffs (mm) wrt IGR (2009)DX DY DZ RX RY RZ
SCL
RMS
WRMS
MEDI
TOTAL
ERR
IGU 6-hr predictions:
mean
3.5
-0.6
0.3
0.8
3.1
-0.7
28.9
21.3
15.6
41.7
std dev
4.7
4.9
3.4
13.8
16.3
27.2
2.6
19.7
8.0
IGU 24-hr predictions:
1.1
-0.1
-0.5
-0.9
-1.3
64.7
47.3
30.2
80.2
1.8
2.0
3.8
21.9
31.2
52.0
1.9
33.3
6.0
IGA observations:
1.2
0.1
-0.2
0.9
-1.2
9.0
7.2
1.3
12.7
1.5
1.6
Orbit errors double when prediction interval increases by x4
IGA total err only ~40% worse than IGRs (but 175% worse for RZ)

9. Ultra-Rapid Orbit Diffs (mm) wrt IGR (2009)DX DY DZ RX RY RZ
SCL
RMS
WRMS
MEDI
TOTAL
ERR
IGU 6-hr predictions:
mean
3.5
-0.6
0.3
0.8
3.1
-0.7
28.9
21.3
15.6
41.7
std dev
4.7
4.9
3.4
13.8
16.3
27.2
2.6
19.7
8.0
IGU 24-hr predictions:
1.1
-0.1
-0.5
-0.9
-1.3
64.7
47.3
30.2
80.2
1.8
2.0
3.8
21.9
31.2
52.0
1.9
33.3
6.0
IGA observations:
1.2
0.1
-0.2
0.9
-1.2
9.0
7.2
1.3
12.7
1.5
1.6
Z rotation errors are largest RT error –
from UT1 prediction errors
Largest RT orbit prediction error comes from UT1 predictions
IGA accuracy also limited by RZ rotations

10. Ultra-Rapid Orbit Diffs (mm) wrt IGR (2009)DX DY DZ RX RY RZ
SCL
RMS
WRMS
MEDI
TOTAL
ERR
IGU 6-hr predictions:
mean
3.5
-0.6
0.3
0.8
3.1
-0.7
28.9
21.3
15.6
41.7
std dev
4.7
4.9
3.4
13.8
16.3
27.2
2.6
19.7
8.0
IGU 24-hr predictions:
1.1
-0.1
-0.5
-0.9
-1.3
64.7
47.3
30.2
80.2
1.8
2.0
3.8
21.9
31.2
52.0
1.9
33.3
6.0
IGA observations:
1.2
0.1
-0.2
0.9
-1.2
9.0
7.2
1.3
12.7
1.5
1.6
due to modelling of orbit dynamics
large X, Y rotation errors – from PM prediction errors
Next largest RT limits from orbit modelling (solar radiation pressure effects) & PM prediction errors

11.  Multi-technique EOP combinations mostly sub-optimal ! 
EOP Error Sources
Station-related measurements:
thermal noise
instrumentation
propagation delays
multipath, etc
σStation ≈ 1/√NStation
Geophysical & parameter models:
esp near S1,
K1, K2 tidal
periods
AAM/OAM
errors
Source-related errors:
orbit dynamics
(GPS, SLR, DORIS)
quasar structures
(VLBI)
σSource ≈ 1/√NSource
σEOP = + +
Possible improvements:
new subdaily EOP
tide model ?
better handling of
parameter
constraints ?
modern theory of
Earth rotation ?
more robust SLR,
VLBI networks ?
more stable site
installations ?
near asymptotic limit
for GPS already
new GNSS
constellations
better GNSS orbit
models ?
quasar structure
models (VLBI) ?
 Multi-technique EOP combinations mostly sub-optimal ! 

12. Conclusions
Generally, IGA/IGU near- & real-time orbits & EOPs are of very high quality
could use more & better input Analysis Center solutions
Rotations are leading real-time orbit error
due to UT1 & PM prediction errors used for IGU orbits
models for orbit dynamics also add some rotational errors for some ACs
EOP services could better use IGU products
provide updates at least 4 times daily
seek better input AAM + OAM predictions
improve combination algorithms
present IERS predictions generally not adequate for IGS requirements
IGS ACs generate better 1-day PM predictions internally from their own latest measurements; we cannot do that for UT1 though
Better model for subdaily tidal EOP variations also needed
errors of IERS model alias into GPS orbit parameters