Sensing atmospheric water vapour using the global positioning system

Includes bibliographical references .Atmospheric water vapour measurements are of importance to meteorologists, radio astronomers and geodesists. Precipitable water vapour (PWV) is a greenhouse gas to be reckoned with in numerical weather models and climate change studies, it is a nuisance in centimetre-wavelength radio astronomy and introduces range errors in space geodetic techniques. The propagation time of electromagnetic waves is the principal observable in the Global Positioning System (GPS). Accurate estimates of the delays experienced by the radio signals travelling from the satellites to ground-based receivers are made during the post-processing of GPS observations. In combination with meteorological observations made at the receiver, the estimated delays can be used to determine the amount of integrated precipitable water vapour along the signal path. In this thesis an overview of the basic GPS principles and components is provided, as well as a derivation, from first physical principles, of the mechanisms contributing to the delay experienced by a radio signal traversing the ionosphere and troposphere. Implementing this theoretical background, PWV and tropospheric delays are estimated and compared to measurements made by other techniques, namely radiosondes, water vapour radiometry and very long baseline interferometry (VLBI). A high degree of correlation is observed in all instances of inter-technique comparison. The usefulness of GPS-derived slant delays is demonstrated by their ability to reduce VLBI inter-station baseline repeatabilities when they are included in the VLBI analysis. However, this contributed to a higher mean formal baseline error. Furthermore, it shown that GPS-derived slant delay accuracies, when compared to radiometry, can be improved through the stacking of GPS processing residuals to make corrections for the effects of multi path and antenna phase centre variations. A modified residual stacking (MRS) method is proposed, in which data weighting is based on a measured autocorrelation function; however, in most instances the more complex MRS failed to significantly improve on the corrections made by normal residual stacking. GPS-derived PWV time-series from thirty South African stations for a four-year period are presented. A four-parameter model was fitted to the time-series to correct for seasonal effects and detect linear trends. It is shown that an autoregressive moving average (ARMA) model is required to estimate realistic trend uncertainties, rather than the white-noise model implicit in standard least-squares analyses. Furthermore, significant trends in PWV were observed in South Africa with the central parts showing a decrease in PWV during the study period, while an increase is observed over the southwest and northeast. These trends coincide with a temperature increase observed over the whole of South Africa for the study period. A hypothesis is presented to explain the different trends, based on the different sources of PWV in different climate areas. Lastly, vertical earth tide displacements (VETD) measured by gravimetry are compared to the modelled VETD applied during GPS processing. It is shown that rnismodelled VETD can contribute significant errors to GPS-derived PWV. A number of methods to mitigate this error are proposed and compared to each other, including a novel technique to accurately measure VETD by GPS