IERS Working Group On HF-EOP - NASA

IERS Working Group on HF-EOP John Gipson, Chair NVI,Inc/NASA GSFC 2018 Fall AGU Washington, DC

Issues & Recent Results Current IERS model of HF-EOP is 20 years old and based on tidal models. Madzak (2016) derived a model based on satellite altimetry data and showed it was better for VLBI Desai and Sibois (2016) derived a model based TPX0.8 and showed that it worked better in GNSS Springer (2017) used an empirical model provided by Gipson derived from VLBI data. This reduced GNSS residuals. All this suggests the IERS model is no longer adequate and needs to be updated.

Formation & Goals of Working Group An ad hoc Working Group on ‘Diurnal and semi-diurnal Earth orientation variation’ was established as an outgrowth of the 2017 JULY GGOS/IERS UAW in Paris. At the 2017 IERS Directing Board meeting the working group was put under the auspices of the IERS. This working group includes representatives of all space geodetic techniques. The WG will evaluate a suite of models and make recommendations to the IERS. John Gipson NVI, Inc./NASA GSFC

Working Group Webpage https://ivscc.gsfc.nasa.gov/hfeop wg John Gipson NVI, Inc./NASA GSFC

Models to be tested Tidal Models IERS Conventions Desai and Sibois Madzak et al Ray Lyard Based on TPX 0.4 Current Based on TPX 0.8. Tidal model Based on TPX 0.9 FES2014 Better on GNSS Better on VLBI Newer tide model Newer tide model Empirical Models Gipson Derived from VLBI data Artz et al Derived from VLBI and GNSS Fit using VLBI, better on GNSS Untested on other techniques 1. We welcome other promising candidates. 2. All of the models are fairly similar. 3. We need your help to test! John Gipson NVI, Inc./NASA GSFC

Comparison of Two Models UT1 Predictions 40 micro-seconds 20 0 8 13 18 23 -20 -40 -60 August 2008 GOT4 JMG(ALL) John Gipson NVI, Inc./NASA GSFC 28

Comparison of Two Models Difference in UT1 50 40 30 micro-seconds 20 10 0 8 13 18 23 -10 -20 -30 -40 -50 August 2008 GOT4-JMG(ALL) John Gipson NVI, Inc./NASA GSFC 28

Comparison of Two Models Difference in UT1 10 8 6 micro-seconds 4 2 0 8 13 18 23 -2 -4 -6 -8 -10 August 2008 GOT4-JMG(ALL) John Gipson NVI, Inc./NASA GSFC 28

Some GPS Results IERS model Courtesy Tim Springer John Gipson NVI, Inc./NASA GSFC

Some GPS Results Gipson 2010 VLBI model Courtesy Tim Springer John Gipson NVI, Inc./NASA GSFC

Some VLBI Tests Cont17 Baseline Scatter 20 18 16 14 12 IERS 10 No model Linear (IERS) 8 Linear (No model) 6 4 2 0 0 2000 4000 6000 8000 10000 12000 14000 Applying IERS model reduces baseline scatter. All following tests use CONT17 data set. John Gipson NVI, Inc./NASA GSFC

Difference in Baseline Scatter 50% CONT17 Reduction in Scatter: NoneIERS 45% 40% 4 IERS is better 3 2 35% 30% 25% 1 20% 0 0 2000 4000 6000 8000 10000 15% 12000 -1 10% -2 ‘None’ is better -3 -4 5% 0% -4 -3 -2 -1 Each point is the difference in baseline scatter between ‘None’-IERS. If a point is above the axis, then IERS is better. Below, then ‘None’ Is better. Reduction in scatter for 65/86 baselines. Average reduction is 0.94 mm IERS Model is better than no model. John Gipson NVI, Inc./NASA GSFC 0 1 2 3 4

IERS vs Gipson (VLBI) Difference in Length Scatter (IERS2017a) 0.5 50% VLBI better 45% 40% 0.4 35% 0.3 30% 0.2 25% 0.1 20% 0 -0.1 0 2000 4000 6000 8000 10000 15% 12000 10% -0.2 -0.3 -0.4 -0.5 IERS better Each point is the difference in baseline scatter between IERS – VLBI model. Reduction in scatter for 49/86 baselines using Gipson. Average reduction is 0.05 mm John Gipson NVI, Inc./NASA GSFC 5% 0% -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 Gipson is better 0.3 0.4

IERS vs FES2012 model 50% Delta Length Scatter (IERS-fes2012) FES2012 better 0.6 0.4 30% 25% 0 0 2000 4000 6000 8000 10000 12000 14000 20% -0.2 15% -0.4 -0.8 40% 35% 0.2 -0.6 45% 10% IERS better Reduction in scatter for 36/86 baselines with FES. Average reduction is -0.01 mm John Gipson NVI, Inc./NASA GSFC 5% 0% -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 IERS is better But really not much difference

IERS vs Desai&Sibois 50% Difference in Length Scatter (IERSDesai) Desai& Sibois better 0.5 0.4 45% 40% 35% 0.3 30% 0.2 25% 0.1 20% 0 -0.1 0 2000 4000 6000 8000 10000 12000 15% 14000 10% -0.2 -0.3 IERS better -0.4 -0.5 Reduction in scatter for 48/86 baselines with Desai & Sibois. Average reduction is 0.02 mm John Gipson NVI, Inc./NASA GSFC 5% 0% -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 Desai&Sibois is better

Desai&Sibois vs Gipson 50% Difference in Length Scatter (DesaiGipson) Gipson better 0.5 45% 40% 0.4 35% 0.3 30% 0.2 25% 0.1 20% 0 -0.1 -0.2 -0.3 -0.4 -0.5 0 2000 4000 6000 8000 10000 12000 15% 14000 Desai &Sibois better Reduction in scatter for 46/86 baselines with Gipson. Average reduction is 0.03 mm John Gipson NVI, Inc./NASA GSFC 10% 5% 0% -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 Gipson is better 0.3 0.4

IERS-Gipson and Desai-Gipson 0.5 50% Difference in Length Scatter (IERS-Gipson) 45% 0.4 40% 0.3 IERS 0.2 0.1 30% 25% 0 -0.1 0 35% 2000 4000 6000 8000 10000 20% 12000 15% -0.2 10% -0.3 5% -0.4 0% -0.5 -0.4 -0.3 -0.2 -0.1 0 50% Difference in Length Scatter (Desai-Gipson) 0.5 0.4 0.3 0.2 0.1 0 -0.1 0 -0.2 -0.3 -0.4 -0.5 0.1 0.2 0.3 0.4 45% 40% 35% 30% 2000 4000 6000 8000 10000 12000 14000 Desai & Sibois 25% 20% 15% 10% 5% 0% -0.4 -0.3 -0.2 -0.1 0 For VLBI analysis, Desai&Sibois better than current IERS. John Gipson NVI, Inc./NASA GSFC 0.1 0.2 0.3 0.4

Anastasiia Girdiuk et al. (VLBI) IERS Desai & Sibois Madzak et al. Take model as a priori, and then estimate residual tidal terms. This was done using VieVs. Best agreement is with Gipson not too surprising, since this was derived from VLBI data. Gipson John Gipson NVI, Inc./NASA GSFC

Tobias Nilsson WRMS (VLBI) WRMS of residuals from 1072 R1, R4 and CONT sessions from 2007-2016. Gipson is best of empirical techniques. Desai & Sibois best of tidal models. Nilsson used VieVs@GFZ John Gipson NVI, Inc./NASA GSFC

CONT17 Hourly EOP 100 50 us 0 -50 -100 -150 -200 11/28/2017 11/30/2017 12/2/2017 12/4/2017 Values RMS Signal RMS Residuals Formal errors 30 µs 10 µs 6 µs 12/6/2017 IERS 12/8/2017 12/10/2017 Residual Can use the residual as a measure of the goodness of the model. John Gipson NVI, Inc./NASA GSFC 12/12/2017

Residual Hourly EOP over all CONTS Model IERS Desai&Sibois EOT11a FES2012 HAMTIDE Madzak Gipson ABN VLBI ABN COMB X-pole μas 107.1 105.4 114.5 110.7 111.3 112.3 105.5 107.1 106.7 Best and worst results. Y-pole μas 114.0 110.4 119.9 113.4 120.4 114.7 111.1 115.2 110.5 Total PM μas 110.4 107.8 117.1 112.0 115.7 113.5 108.2 111.0 108.5 UT1 μs 7.66 7.57 8.60 7.63 7.88 8.03 7.25 7.41 7.33 Total EOP μas 111.90 109.80 121.20 112.80 116.60 115.90 108.40 111.10 109.00 Work done by Tobias Nilsson John Gipson NVI, Inc./NASA GSFC

Residual Hourly EOP over all CONTS 130 125 120 115 110 105 Notice: Starts at 100 100 XP YP UT1 Avg As a result of this and several other tests, Nilsson’s general conclusion is that the Desai&Sibois and Gipson models were best. John Gipson NVI, Inc./NASA GSFC

Conclusions/Next Steps Testing of different models by VLBI groups is well underway and we have some preliminary results. Desai is better than current IERS. Not as good as Gipson. Waiting for feedback from other VLBI and non-VLBI groups. We need to agree on a model prior to ITRF2020—and IGS plans on beginning reprocessing in June 2019. Won’t you please help! John Gipson NVI, Inc./NASA GSFC

Difference in Baseline Variance CONT17 Reduction in Variance: None-IERS 80 60 IERS model is better 40 20 0 0 2000 4000 6000 8000 10000 12000 -20 -40 No model is better -60 -80 Each point is the difference in baseline variance between None-IERS. Reduction in variance for 65/86 baselines. Average reduction is 15 mm 2 IERS Model is better than no model. John Gipson NVI, Inc./NASA GSFC

IERS Each point is the difference in baseline scatter between 'None'-IERS. If a point is above the axis, then IERS is better. Below, then 'None' Is better. Reduction in scatter for 65/86 baselines. Average reduction is 0.94 mm IERS is better 'None' is better IERS Model is better than no model. 0% 5% 10% 15% 20% 25% 30% 35% 40% 45% .