Sensitivity of Antarctic GPS velocity estimates to terrestrial reference frame transformation strategies
Updated: 2026-11-18
In this study, we systematically assess the sensitivity of Antarctic GPS coordinate time series and site velocities to the choice of Helmert transformation strategy when transforming from NASA Jet Propulsion Laboratory (JPL) daily fiducial-free solutions to the ITRF2020 reference frame. We evaluate the effect of: (i) including a scale parameter; (ii) applying robust estimators in place of conventional least squares; (iii) down-weighting vertical coordinates; and (iv) applying geometry-based spatial weighting to the tie station network. The accompanying data contain (i) computed daily IGc20 daily frame coordinates and their associated variance-covariance information; (ii) daily JPL non-fiducial station coordinates and their associated variance-covariance information, both in gdcov file format; (iii) Python scripts used to estimate the Helmert parameters; and (iv) the estimated Helmert transformation parameters in X-file format. We adopted IGc20, the IGS realisation based on the ITRF2020-u2024 solution, as the reference frame for the Helmert alignment. We estimated daily Helmert parameters by transforming JPL’s daily fiducial free station coordinates to daily IGc20 coordinates. Daily IGc20 coordinates were estimated from the IGc20 SINEX (Solution-Independent Exchange) products (last accessed 17-04-2026 via https://files.igs.org/pub/station/coord/IGc20/) using the piecewise linear coordinates (accounting for discontinuities/offsets at equipment changes and other breaks), seasonal harmonics and post-seismic deformation terms of the ITRF2020 trajectory model along with the accompanying variance-covariance information. We obtained the daily non-fiducial GNSS network solutions from JPL in the geodetic covariance (gdcov) file format via https://sideshow.jpl.nasa.gov/pub/JPL_GNSS_Products/Final/ (last accessed 30-03-2026). The supplied daily solutions contain station coordinates and their covariance information in a fiducial-free, but internally precise, network representation. Before transformation, we projected the coordinates and their associated covariance matrices from the downloaded non-fiducial solutions onto a minimally constrained datum using standard GIPSYX utilities. This procedure removes rank deficiency while preserving the internal network geometry. The resulting projected coordinate solutions and covariance matrices were retained in gdcov format and combined with the corresponding daily predicted IGc20 reference coordinate products.