VLBI Analysis in Real-time

Geodetic Very Long Baseline Interferometry (VLBI) is one of the primary space geodetic techniques. As it is providing the full set of Earth Orientation Parameters (EOP), it is unique for observing long-term Universal Time (UT1) and precession/nutation. There is an increased need for accurate real-time estimates of EOP, in particular for satellite-based navigation and positioning systems. Consequently, the International VLBI Service for Geodesy and Astrometry (IVS) is making strong attempts to reduce the time span between the collection of VLBI observations and the availability of the final results. This acceleration in data analysis is one of the main goals of the new generation VLBI system called VLBI2010, which is currently under development. Good progress has already been achieved towards providing the observations in near real-time by using fast fibre connections from the operating stations to software correlators, that are capable of immediately correlating the incoming raw VLBI data. New demands and challenges for the analysis software will emerge with the availability of the correlator output in near real-time. The main objective of project VLBI-ART is to meet these demands by considerably accelerating the VLBI analysis procedure on the base of an elaborate Kalman filter software solution, which represents a perfectly well suited tool for analysing VLBI data in quasi real-time. The Kalman filter will be embedded in the newly developed Vienna VLBI Software (VieVS) and it will be designed to work completely automated without any need for human interaction. Numerous investigations with real and simulated data are to be performed in order to determine and optimise the performance of the software. The tests involving real observations will not only be based on data from special experiments, where the correlation procedure is accomplished in near real-time, but also on post-processing of VLBI data in a quasi real-time mode. The results obtained will be compared to the output from conventionally used VLBI analysis packages, as well as to equivalent parameter series from other techniques such as Global Navigation Satellite Systems (GNSS). On the other hand, we will be able to simulate artificial observations for the complete future VBI2010 network and assess the real-time accuracy that can be achieved. Furthermore, it is intended to probe the promising possibility of including data from several other sensors in the Kalman filter. For example, we will investigate the possible potential improvements of the results by entering tropospheric delays from water vapour radiometers, Earth rotation from ringlaser gyroscopes, tropospheric delays and EOP estimated using data from GNSS, or atmospheric angular momentum calculated from numerical weather prediction models. The automation of the data analysis will include routine removal of ambiguities as well as detection of clock breaks and data outliers. Project VLBI-ART represents a significant augmentation of the state-of-the-art VLBI analysis software VieVS by enabling it to analyse VLBI data in near real-time and provide the parameters of interest with the best possible accuracy. Hence, when VLBI2010 observations will be started in 2015, a software will exist which is able to analyse them.

Knowledge of the rotation and the orientation of the Earth, commonly described by the Earth Orientation Parameters (EOP), is important for all kinds of positioning and navigation on the Earth and in space. The only technique able to completely determine the orientation of the Earth in space is Very Long Baseline Interferometry (VLBI). With this technique, several radio telescopes distributed all over the world observe distant quasars. Since the quasars are extremely far away, they can be considered not to move. Thus, by analyzing the VLBI observations, it is possible to precisely determine the EOP, as well as the positions of the telescopes and a number of other parameters.Ideally, the EOP estimated from VLBI should be available in real-time. Currently, however, this is not the case. The bottleneck is that the data recorded by the stations need to be brought together at one place (called a correlator). Due to the large data volumes, this is currently achieved though recording the data an disks at the stations and the shipped to the correlator, what can take more than one week. However, as more and more telescopes get connected to high-speed networks, the data could be electronically transferred in real-time. In order to then have the results, e.g. the EOP, available in real-time, also the data need to be analyzed in real-time. Current VLBI data analysis softwares do not have real-time capability and commonly require human interaction. Thus, the goal of the project VLBI-ART was to develop a software capable to analyze VLBI observations fully autonomously in real-time.This goal has been achieved by implementing a so-called Kalman filter module in a VLBI soft- ware, called VieVS@GFZ. Since VLBI observations are not yet operationally available in real- time, we have validated the performance of the software though extensive simulations. The results of these simulations show that the software is performing as expected. Furthermore, a Kalman filter has the advantage that many random error sources, like the propagation delay in the atmosphere, can be precisely modeled, what is also beneficial when post-processing the observations. Our results show that the using a Kalman filter results in more accurate estimation of EOP, station coordinates, and other parameter.