Improving GPS results using continuous processing strategies

The establishment of the International GPS Service for Geodynamics (IGS) in 1994 and the rapid development of its high quality products have opened promising, new capabilities for the global GPS positioning community. Two important products of the IGS are the precise satellite ephemerides and the precisely-determined locations of many permanent tracking sites distributed over the entire globe. In addition, we can obtain high quality RINEX data from the global tracking network on a daily basis. The combination of a multi-purpose regional GPS network and the nearby IGS fiducial stations, along with the use of IGS precise ephemerides, can provide results in a well-defined global reference frame with maximum accuracy and minimum computational burden. We are fast moving towards the situation of having multi purpose GPS networks containing medium length baselines (up to 1000 km), which require the highest accuracy level of positioning. As a general rule of thumb, the common-mode errors due to ionospheric, tropospheric and orbital uncertainties increase with baseline distance. This therefore introduces a range of problems in the processing of medium length baselines, such as ambiguity resolution, vertical precision associated with the treatment of the tropospheric delay, etc. The results from GPS processing can be used for many different applications, such as in the fields of glaciology, volcanology, oceanography, meteorology, etc., and so we increasingly require a high degree of flexibility and innovation in our GPS processing techniques. The aim of this thesis is to improve our processing techniques over medium length baselines making use of permanent stations, such as the IGS fiducial stations and the IGS precise ephemerides. A processing technique is required that exploits the continuity of the GPS data and that is able to satisfy various high precision GPS applications (at the 1 part in 107 level or better). We suggest a new approach for GPS processing, the so-called continuous processing approach, rather than the classical processing that pertains to batch least squares adjustment. For a continuous processing methodology, we have to resolve a number of related matters, such as cycle slip detection and repair; outlier detection; the use of sequential least squares adjustment, incorporating between-epoch correlation and the necessary modification of the unknown vector and cofactor matrix to maintain the continuity of data processing under the influence of a satellite combination change; ambiguity resolution, and estimation strategies of tropospheric delay. We demonstrate the efficiency of the new technique using several different high precision GPS applications in geodynamics. Three studies were undertaken using data from a tide gauge network within Australia, the Northridge earthquake event in North America, and a study of the motion of the Amery Ice Shelf in Antarctica. The results from using the continuous processing approach developed in this thesis are in good agreement with previous studies but yield additional insight into some of the physical processes, such as the motion of the Amery Ice Shelf. The utilisation of improved stochastic modelling of measurements and the continuous nature of the processing increases the effectiveness and reliability of outlier detection and ambiguity resolution. The posteriori variance factor is also improved by up to 30%. With greater flexibility in the monitoring of complicated movements of stations, the new technique will prove to be a very promising tool for high precision applications involving regional GPS networks.