Differences in annual signals between IGS- and NGL-derived position time series: testing different strategies of alignment to the reference frame

Annual signals derived from GNSS position time series have become a very important source of information on a variety of geophysical phenomena occurring in the closer or farther vicinity of the antennas. In this research, we investigated the vertical position time series obtained from the International GNSS Service (IGS) third reprocessing (repro3) and PPP times series from the Nevada Geodetic Laboratory (NGL). We selected 1019 globally distributed stations with good quality data and time spans longer than 5 years. First, we pre-processed them for outliers and identified the epochs of offsets using manual inspection. Then, we inspected the differences between the annual signals contained in both sets of time series. We noticed from the map of annual signal differences that there is a worldwide annual common mode in the series of “IGS-NGL” station position differences, with a median amplitude of the order of 2 mm, and a maximum around August. We hypothesized that those differences could be explained by different strategies of alignment to the reference frame, and in particular by the alignment of the NGL series to the scale of the ITRF. To investigate this, we produced additional series by applying different types of constraints to the daily repro3 normal equations. Namely, we compared four sets of time series: “IGS” which are the official repro3 solutions, aligned by no-net-rotation nor translation (NNR+NNT) constraints to the IGSR3 reference frame via the well-distributed IGSR3 core network; “IGa” which are repro3 solutions aligned by no-net-rotation, translation nor scale (NNR+NNT+NNS) constraints to IGSR3 via the same core network; “IGb” which are repro3 solutions aligned by NNR+NNT+NNS constraints to IGSR3 via the same daily sets of reference stations as used by Jet Propulsion Laboratory (JPL) to align their orbit and clock products; and “IGc” which are repro3 solutions aligned by NNR+NNT+NNS constraints to the IGb14 reference frame via the same set of reference stations as used by JPL. A comparison of these four sets of time series with the NGL PPP time series reveals that c.a. half of the systematic differences in vertical annual signals between IGS and NGL comes from the alignment of the NGL solutions in scale to the reference frame, and another half comes from the use of different station networks for the alignment to the reference frame.