A New Vertical Land Movements Data Set from a Reprocessing of GNSS at Tide Gauge Stations

The main objective of the International GNSS Service (IGS) Tide Gauge Benchmark Monitoring (TIGA) Working Group is to provide accurate coordinates and changes in them in the form of long-term trends for globally distributed Global Navigation Satellite System (GNSS) stations at or close to tide gauges (TGs). Mean sea level (MSL) records derived from TG observations measure sea level relative to benchmarks on the land and structures supporting the TGs. Therefore, any changes in land levels affect the MSL records and the computed estimates of sea level change, ie. the MSL trends. In order to compute regionally or globally averaged MSL required for climate studies, these MSL trends have to be corrected for the vertical land movements (VLMs) derived from the GNSS observations.
In this study, we have estimated a new set of VLMs at or close to TGs from the recent reprocessing campaign “repro2” undertaken by British Isles continuous GNSS Facility and University of Luxembourg TIGA Analysis Center (BLT). The position time series of more than 700 stations distributed around the world have been reprocessed for the period 1994 to 2015 using the latest bias models and processing strategies following the conventions of the International Earth Rotation and Reference Frame Service (IERS). It is well known that position time series are affected by discontinuities, which stem from different sources such as earthquakes, hardware changes and other artificial offsets that do not reflect real geophysical events. Since uncorrected discontinuities adversely affect the trend estimates, we have, after applying all known offset epochs, inspected the time series of all stations manually and added any further offset epochs required during the analysis. We have included a total of 2500+ discontinuities of which two-thirds are from hardware changes, 4% from earthquakes and 9% from unknown sources. We fit a deterministic model (sum of linear trend and seasonal terms) to the position time series using the Hector software package. As expected the annual terms show the highest power with amplitudes of a few millimeters. The stochastic model for estimating trend and associated uncertainties follows a power-law noise process as has previously been described as optimal for GNSS-derived position time series. The new set of VLM estimates from our repro2 solution is evaluated through comparison with a published GNSS solution, the recent ICE-6G model of glacial isostatic adjustment and by application to the latest release of MSL trends from the Permanent Service For Mean Sea Level.