[G01-3] Reference frame methodology and implementation

[G01-3-05] Kalman filter terrestrial reference frame solutions based on time-variable process noise

^○Benedikt Soja¹, Richard Gross¹, Claudio Abbondanza¹, Toshio Chin¹, Michael Heflin¹, Xiaoping Wu¹, Kyriakos Balidakis², Tobias Nilsson³, Susanne Glaser², Maria Karbon², Robert Heinkelmann³, Harald Schuh^2,3 (1.Jet Propulsion Laboratory, California Institute of Technology, Pasadena, United States of America, 2.Technische Universität Berlin, Berlin, Germany, 3.GFZ German Research Centre for Geosciences, Potsdam, Germany)

In recent years, Kalman filtering has emerged as a suitable technique to determine terrestrial reference frames (TRFs), a prime example being the JTRF2014 solution by the IERS ITRS Combination Center at JPL. The time series approach allows variations of stations coordinates that are neither reduced by observational corrections nor considered in the functional model to be taken into account. These variations are primarily due to non-tidal geophysical loading effects that are not included in the IERS conventions (i.e., non-tidal atmospheric pressure, ocean, and continental water storage loading). It is standard practice that the process noise models applied in Kalman filter TRF solutions are derived from time series of the aforementioned geophysical loading effects and account for station dependent differences. So far, it has been assumed that the parameters of these process noise models are constant over time. However, due to the presence of seasonal variations and episodic events such as ENSO, this assumption does not truly reflect reality. In this study, we derive a station coordinate process noise model allowing for such temporal variations. Time series of daily non-tidal loading deformations between 1985 and the end of 2015 provided by GFZ Potsdam are used to calculate monthly process noise parameters for every single station. As a test case, this process noise model is applied in the computation of a TRF based on very long baseline interferometry (VLBI) data from over 4000 VLBI sessions during the same time frame. The resulting VLBI TRF is compared to a solution based on a constant process noise model, but otherwise on identical input data and Kalman filter setup. In particular, it is investigated which VLBI stations profit the most from a spatiotemporal process noise model and how frame defining parameters, such as the scale, are affected.