Impact of network constraining on the terrestrial reference frame realization based on SLR observations to LAGEOS
The Satellite Laser Ranging (SLR) network struggles with some major limitations including an inhomogeneous global station distribution and uneven performance of SLR sites. The International Laser Ranging Service (ILRS) prepares the time-variable list of the most well-performing stations denoted as `core sites' and recommends using them for the terrestrial reference frame (TRF) datum realization in SLR processing. Here, we check how different approaches of the TRF datum realization using minimum constraint conditions (MCs) and the selection of datum-defining stations affect the estimated SLR station coordinates, the terrestrial scale, Earth rotation parameters (ERPs), and geocenter coordinates (GCC). The analyses are based on the processing of the SLR observations to LAGEOS-1/-2 collected between 2010 and 2018. We show that it is essential to reject outlying stations from the reference frame realization to maintain a high quality of SLR-based products. We test station selection criteria based on the Helmert transformation of the network w.r.t. the a priori SLRF2014 coordinates to reject misbehaving stations from the list of datum-defining stations. The 25 mm threshold is optimal to eliminate the epoch-wise temporal deviations and to provide a proper number of datum-defining stations. According to the station selection algorithm, we found that some of the stations that are not included in the list of ILRS core sites could be taken into account as potential core stations in the TRF datum realization. When using a robust station selection for the datum definition, we can improve the station coordinate repeatability by 8%, 4%, and 6%, for the North, East and Up components, respectively. The global distribution of datum-defining stations is also crucial for the estimation of ERPs and GCC. When excluding just two core stations from the SLR network, the amplitude of the annual signal in the GCC estimates is changed by up to 2.2 mm, and the noise of the estimated pole coordinates is substantially increased.