290 likes | 314 Views
This study presents the preliminary results of the IGS 2nd reprocessing (repro2), aiming to reduce errors in GNSS data products and enhance ITRF2013 contributions. The analysis includes updates since repro1, improvements in data integrations, and new models applied. Detailed information can be found at the provided link.
E N D
The IGS contribution to ITRF2013–Preliminary results from the IGS repro2SINEX combinationsPaul Rebischung, Bruno Garayt, Xavier Collilieux, Zuheir AltamimiAGU Fall Meeting 2014, Abstract G11C-07, San Francisco, 15 December 2014
The IGS 2nd reprocessing (repro2) • Re-analysis of GNSS data collected by the IGS network since 1994 using the latest models and methodology • Reduce systematic errors in IGS products • Provide IGS contribution to ITRF2013 • Main updates since repro1: • Daily data integrations (instead of weekly) • GLONASS data processed by some ACs • IGb08/igs08.atx framework • IERS2010 Conventions • New yaw attitude models for eclipsing satellites • A priori modeling of Earth radiation pressure and antenna thrust • See details at: http://acc.igs.org/reprocess2.html
Analysis Center submissions • Submitted products include: • Satellite orbits and clocks • Terrestrial frames and EOPs (daily SINEX files) Contributions from 7 operationalACs + 2 TIGA contributions
1st combinations (1/2) WRMS of station position residuals(i.e., of daily « AC – combined » differences) • Outlying days for several ACs: ongoing re-submissions
1st combinations (2/2) Smoothed WRMS • ULR: • Large systematic errors in East • Issue in sub-network combos • Ongoing re-submissions • GRG: • Large systematic errors in Up • Pronounced semi-annual variations in North and Up • Under investigation → 2nd combinations with GRG, ULR & GTZ included for comparison only
2nd combinations (1/2) Smoothed WRMS Smoothed, unbiased WRMS • MIT seems to dominate in Up (and North). • Because « classical » WRMS are biased in favor of ACs with non-common stations. → Use Sillard (1999)’s unbiased WRMS # stations in AC solutions
2nd combinations (2/2) Smoothed, unbiased WRMS • GFZ’sNorth WRMS vs. meanionosphere TEC: • Error in 2ndorderionospheric corrections • New productssubmitted, but stillaffected by several issues • → 3rdcombinationswith GFZ included for comparisononly ― GFZ’sNorth WRMS ― a + (b x mean_TEC)
3rd combinations Smoothed, unbiased WRMS • Inter-AC agreementafter 2004: • Horizontal: ≈ 1 – 1.5 mm • Vertical: ≈ 3 – 4 mm • Comparable to the weekly repro1 results • Substantial degradation before 2000 (and 1997) • Much less marked in repro1: 3D RMSof repro1combinationresiduals
Spectral analysis Stackedperiodograms of station position residuals(computedusing stations with > 3000 days of data) • Background: flicker + white noise • Spectral peaks at: • Annual period; GPS draconitic harmonics • Fortnightly periods (14.8, 14.2, 13.7 & 13.2 d) • 9.1 d (MIT); 8.2 & 7.8 d (COD, ESA, MIT);7.0 d (MIT); 3.65 & 2.2 d (GRG) East North Up
Scale AC / ig2 scale offsets • Inter-AC agreement: 0.3 – 0.5 mm • Scale rate differences < 0.1 mm/yr • Combined scale rate wrt IGb08:-0.03 mm/yr • Contribution to ITRF2013 scale rate? ig2 / IGb08 scale offsets
Origin: Y component AC / ig2 Y origin offsets • Inter-AC agreement: ≈ 3 – 5 mm • nearly 10 times larger than for scale • Y component of combined origin: • Good agreement in phase with SLR • Annual amplitude slightly over-estimated ― ig2 / IGb08 Y origin offsets ― SLR / ITRF2008 Y origin offsets (cf. G11C-08)
Origin: X component AC / ig2 X origin offsets • Inter-AC agreement: ≈ 3 – 5 mm • nearly 10 times larger than for scale • X component of combined origin: • Annual amplitude under-estimated • Broad spectral peak around 3.12 cpy ― ig2 / IGb08 X origin offsets ― SLR / ITRF2008 X origin offsets (cf. G11C-08)
Origin: Z component AC / ig2 Z origin offsets • Inter-AC agreement: ≈ 6 – 9 mm • nearly 20 times larger than for scale • Z component of combined origin: • Spoiled by GPS draconitic harmonics • Annual signal out-of-phase with SLR ― ig2 / IGb08 Z origin offsets ― SLR / ITRF2008 Z origin offsets (cf. G11C-08)
Earth Orientation Parameters AC / ig2 X-pole differences AC / ig2 Y-pole differences AC / ig2 LOD differences • Inter-AC agreement: ≈30 μas; 150-200 μas/d; 15-30 μs/d AC / ig2 X-pole rate differences AC / ig2 Y-pole rate differences WRMS of EOP residual time series
Summary • Station positions: • Post-2004 inter-AC agreement comparable to weekly repro1 results • Substantially worse in early years • Scale: • Excellent inter-AC agreement • Contribution to ITRF2013 scale rate? • Origin: • No substantial improvement • EOPs: • Inter-AC agreement slightly improved compared to repro1
Next steps • Expected by end of January 2015: • Re-submissions (COD, GFZ) • Extensions to 2014 (COD, ESA, MIT) • Final combined solutions due by end of February 2015 • If time allows, study stationresidual time series: • Form long-termcumulative solution • Revised discontinuity list • Modeling of post-seismicdeformations ULAB Northresiduals
Pole coordinates AC / ig2 Y-pole differences AC / ig2 X-pole differences
Pole rates AC / ig2 Y-pole rate differences AC / ig2 X-pole rate differences
Length of day Normalizedperiodograms AC / ig2 LOD differences
Origin: X component Normalizedperiodograms AC / ig2 X origin offsets
Origin: Y component Normalizedperiodograms AC / ig2 Y origin offsets
Origin: Z component Normalizedperiodograms AC / ig2 Z origin offsets
Annual – East codemresa gfzgrgjpl mit gtzulr
Annual – North codemresa gfzgrgjpl mit gtzulr
Annual – Up codemresa gfzgrgjpl mit gtzulr
Relative formal errors: pole coordinates σYPO/ median(σsta) [mas/mm] σXPO / median(σsta) [mas/mm]
Relative formal errors: pole rates & LOD σYPOR/ median(σsta) [mas/d/mm] σXPOR / median(σsta) [mas/d/mm] σLOD / median(σsta) [ms/d/mm]
Relative formal errors: geocenter σYGC/ median(σsta) [mm/mm] σXGC / median(σsta) [mm/mm] σZGC / median(σsta) [mm/mm]