Towards More Frequent Terrestrial Reference Frame Updates

The Jet Propulsion Laboratory realizes global terrestrial reference frames (TRFs) using a Kalman Filter and Smoother approach where VLBI-, GNSS-, DORIS-, and SLR-derived station positions and EOPs are assimilated along with site ties and data measurement covariance matrices. Kalman Filtering provides an ideal framework wherein time-series based TRFs can be sequentially updated as new space-geodetic (SG) observations become available. Given the current availability of operational SG observations, TRF updates on a weekly or monthly basis are feasible using JPL's time series approach. ITRF realizations are currently determined every 3 to 5 years, thus allowing new observations to be incorporated and benefitting from the improvements to the space-geodetic data reduction and ITRF realization schemes. Yet such a long interval between ITRF releases makes the frame vulnerable to temporal degradation and limits its ability to accurately extrapolate station positions and Earth Orientation Parameters into the future. Kalman Filtering enables TRFs to be updated more frequently. Because the state vector and its full covariance matrix can be saved at the end of the solution, as additional measurements become available the filter can be re-started from the saved state and run forward in time to assimilate the additional measurements. The updates, produced with no need to re-generate the full solution, will therefore improve the predictions through the constraints offered by the newer observations. This is potentially appealing for all those scientific applications in which frame predictability is crucial, such as sea level change, vertical land motion determination, geodynamics and precise orbit determination. By using the input data set adopted for the JTRF2014 realization, we will investigate schemes to more frequently update the TRF based upon our Kalman Filter and Smoother approach. The sequentially updated state vector parameters, mainly station positions, velocities and frame defining parameters (origin, scale and orientation) will be compared to a baseline combined solution. Comparisons between updated state vector parameters and predictions derived from the baseline solution will also be shown.