IGS Contribution to the ITRF2014 Paul Rebischung, IGN France The IGS contribution to ITRF2014 is based on the products from the second IGS reprocessing campaign (repro2). A complete list of the models and conventions used in the repro2 data analyses can be found at http://acc.igs.org/reprocess2.html, where individual modeling aspects that have been changed since the repro1 campaign, are new, or have not been widely adopted previously have been highlighted using red font. The main updates since the repro1 campaign are: * a switch from weekly to daily terrestrial frame integrations made to facilitate the study of station displacements, * the analysis, by some ACs, of GLONASS data in addition to GPS data, * the implementation of the IGb08/igs08.atx reference frame and calibration framework, * the implementation, mostly, of the IERS 2010 Conventions, regarding in particular the conventional mean pole model, the geopotential model, tidal station displacements, tidal variations in the Earth's rotation, tropospheric and ionospheric propagation delays, * the implementation, partly, of new attitude models for eclipsing satellites, * the modeling of Earth radiation pressure and, mostly, of antenna thrust acting on GNSS satellites. A total of 9 ACs contributed to the repro2 campaign and provided daily terrestrial frame (SINEX) solutions among other products. The AC repro2 SINEX time series were extended with consistent operational products up to GPS week 1831, when available, and daily combinations of the AC repro2+operational SINEX solutions were performed over the period 1994-01-02 to 2015-02-14 (GPS weeks 730 to 1831), based on the following AC contributions: * COD: - 0730_0 to 1772_6 : cf2 solutions included with weight - 1773_0 to 1831_6 : cof solutions included with weight * EMR: - 0769_0 to 1785_6 : em2 solutions included with weight - 1786_0 to 1831_6 : emr solutions included with weight * ESA: - 0782_0 to 1788_6 : es2 solutions included with weight - 1789_0 to 1831_6 : esa solutions included with weight * GFZ: - 0730_0 to 1720_6 : gf2 solutions included for comparison only - 1721_0 to 1827_6 : gf2 solutions included with weight - 1828_0 to 1831_6 : gfz solutions included with weight * GRG: - 0938_4 to 1825_3 : gr2 solutions included for comparison only - 1825_4 to 1831_6 : grg solutions included for comparison only * GTZ: - 0730_0 to 1720_6 : gt2 solutions included with weight * JPL: - 0730_0 to 1815_6 : jp2 solutions included with weight - 1816_0 to 1831_6 : jpl solutions included with weight * MIT: - 0730_0 to 1803_6 : mi2 solutions included with weight - 1804_0 to 1831_6 : mit solutions included with weight * ULR: - 0782_0 to 1773_1 : ul2 solutions included with weight - 1773_2 to 1825_2 : ul2 solutions included for comparison only The resulting daily combined SINEX solutions, named "ig2", constitute the IGS contribution to ITRF2014. The combination methodology comprises the following steps: * Pre-processing of AC solutions: - Recovery, when necessary, of the unconstrained normal equation - Fixation of UT1-UTC and of the satellite phase center offsets - Application of ocean pole tide corrections, when necessary - Calibration of LOD estimates with respect to IERS Bulletin A - Explicitation of apparent geocenter coordinates - Pre-elimination of specific parameters (stations with abnormally large coordinate differences with the other ACs or with metadata errors; ULR's geocenter coordinates) - Inversion of the normal equation under NNR and NNT constraints - A priori scaling of the covariance matrix so that the median 3D station position formal error is set to 4 mm * Iterative combinations: - The combined parameters include station positions, ERPs (pole coordinates, pole rates and LOD) and apparent geocenter coordinates. - Translation and rotation parameters are estimated between each AC solution and the combined solution. - After each iteration, unbiased WRMS of the station position residuals are computed for each AC and East/North/Up component. Stations with residuals larger than 5 times the unbiased WRMS in either East, North or Up are rejected from the corresponding AC solutions. It is however ensured that one station cannot be rejected from more than one AC solution during the same iteration. - Iterations are performed until no outlier is detected anymore. * Final combination: - A posteriori variance factors are computed for each AC based on a variance component estimation method known as "degree of freedom estimation", and used to rescale the AC covariance matrices. - A final combination is performed. - The origin and orientation of the final combined solution are defined by NNT and NNR constraints with respect to a subset of selected IGb08 core stations. - The scale of the combined solution is a "natural" mean of the AC frame scales, hence conventionally defined by the igs08.atx satellite PCO values. From WRMS values of the residuals of the daily repro2 combinations, the overall inter-AC level of agreement is assessed to be 1.5 mm for the horizontal components and 4 mm for the vertical component of station positions, 25-40 µas for pole coordinates, 140-200 µas/d for pole rates, 8-20 µs/d for calibrated LOD estimates, 4 mm for the X and Y components of geocenter motion, 8 mm for its Z component and 0.5 mm for the terrestrial scale. On the long term, the origins (resp. scales) of the AC terrestrial frames show relative offsets and rates within ± 3 mm and ± 0.3 mm/yr (resp. ± 0.5 mm and ± 0.05 mm/yr). The combination residuals also present AC-specific features, some of which are explained by known analysis specifics, while others remain under investigation. The overall set of daily repro2 combined solutions comprises a total of 1845 different stations. However, only 58% (1073) of the 1845 stations indeed have position time series with more than 1000 points, and 17% (321) have series with more than 4000 points. Only 31% (578) of the 1845 stations are or have been part of the IGS network. Most of the remaining stations are part of either regional (EPN, SIRGAS...), national (CORS, ARGN...) or geophysically-oriented (PBO, SONEL...) permanent GNSS networks. The metadata blocks of the daily repro2 combined SINEX files thus rely on site logs gathered from different sources, or, when no site log could be found, on the metadata given in the AC SINEX files. For more detail on the repro2 combination methodology and results, see: Rebischung P, Altamimi Z, Ray J, Garayt B (2016) The IGS contribution to ITRF2014. Journal of Geodesy (under review)