Improving the precision and accuracy of geodetic GPS: Applications to multipath and seismology

The Global Positioning System (GPS) enables precise and accurate determination of position anywhere on anywhere on the Earth, a boon to the field of geodesy. Although great advances in geodetic GPS positioning precision and accuracy have been made over the last decade, improvements can still be made. This dissertation addresses GPS positioning error from two different directions---understanding and taking advantage of the repeating nature of some errors, or understanding and taking advantage of the relationship between errors in different contemporaneous GPS observables.; In the area of high-rate GPS positioning, repeating errors have a substantial impact on the solution. In this dissertation, I study high-rate GPS error reduction using data from the 2002 Denali Fault earthquake. I apply the techniques of modified sidereal filtering and spatial filtering to positions from 25 GPS stations throughout northwestern North America, and I develop improvements to these methods such as data equalization and careful selection of sidereal filtering sites. Substantial reduction in noise magnitude is achieved through proper application of sidereal and spatial filters, and the resulting 'GPS seismograms' show excellent agreement to records from seismometers.; Multipath, where GPS signals arrive by more than one path and thereby create a range error, can be understood through the GPS observables. Multipath effects on GPS carrier phase, pseudorange, and signal-to-noise ratio (SNR) measurements are different but linked by the same underlying principles. In this dissertation, I explain multipath effects on the GPS observables and define multipath in terms of conditions specific to geodetic GPS installations and receivers. I develop two approaches to multipath errors, both using SNR measurements---a graphical method for multipath assessment, and a computational method for multipath modeling and carrier phase error reduction. The graphical method shows great promise for understanding spatial and temporal variability in multipath errors, but provides no avenue for removing these errors. The theory behind SNR modeling is robust, but complicated to implement with geodetic GPS measurements of SNR. I discuss the difficulties inherent in SNR modeling and demonstrate how this technique is of limited utility for geodetic GPS even in the most simple of multipath environments.