Sea surface computations from local satellite tracking and satellite altimetry
The user of satellite radar altimetry in acquiring geodetic data over oceanic areas is investigated using precise ephemeris information for satellite coordinates and GEOS-3 altimetry data collected within the context of a Canadian GEOS-3 altimetry experiment conducted in the Hudson Bay area. GEOS-3 Doppler tracking data from four stations in the vicinity of Hudson Bay are analysed for the purpose of establishing the extent of local improvement of the orbits required to reduce the altimetry data. The a priori decisions involved in the computation of mean sea surface were to use precise ephemeris information; to use the available Doppler tracking data to reduce the ephemeris errors by means of the semi short-arc technique in the translocation mode; to use these Doppler fitted orbits to reduce and adjust the altimetry data; and to compare the results to various versions of the geoid to establish the degree of usefulness of the satellite altimetry technique for the determination of the sea surface topography. The estimation of sea surface from satellite altimetry data is discussed. The problem areas and the various sources of errors inherent in the satellite altimetry observables used to define the sea surface are identified and their respective modelling with the context of this study is examined in detail. The basic principles of the semi short-arc method, conventionally used in Canada for geodetic positioning, are related here to the intended local improvement of the orbits required to utilize the altimetry data. From numerical results based on the combination of the available GEOS-3 Doppler data with reference orbits described in this case by a set of DMA precise ephemerides, precise ephemeris errors exceeding occasionally 10 m int he radial direction were found. The formulation of the least squares model of intersecting altimetry arcs are used here to remove long wavelength errors primarily due to unmodelled gravity field effects in the orbit determination is examined in detail. Residual orbit biases are found to be well represented by an absolute bias and a tilt parameter for each arc. From the study of a regional network of GEOS-3 orbital arcs in Hudson Bay the internal consistency of the estimated sea surface is found to be less than one metre, whereas the rms difference of the sea surface heights at the intersection points is found to be 1.1 m. Comparisons of the altimetry-derived sea surface with a GEOS-3 sea surface independently determine by DMA indicated a relative consistency of 0;98 m (rms), attributed primarily to the presence of unmodelled time varying effects in the sea surface and anticipated differences induced by the difference reduction procedures used to obtain the adjusted altimetry data. Comparisons with two combined geoids indicated an agreement of the order of 1.2 m (rms), attributed mainly to errors in the computed geoids and some level of spurius structure possibly introduced into the sea surface because of the Doppler orbit adjustment. Based on the analysis and the results of this study, several contributions relevant to the problem of acquiring geodetic data in oceanic areas form satellite altimetry have resulted from this research. Of these, the following are considered, int he authors opinion to be the most significant: 1. The practical demonstration of the feasibility of improvement of the orbits required to utilize the altimetry data from local satellite tracking; the implications from the present results are that if local Doppler tracking can improve orbital information as good as the precise ephemeris, it should be even more important in the case of worse available reference orbits. 2. A complete discussion into the diverse kinds of information which needs to be considered in any attempt of utilizing satellite altimetry data. While the results presented herein are based on assumptions and decision pertaining to the intended application of the Hudson Bay Experiment, the information contained in this thesis should serve to highlight the strengths and the weaknesses of the estimation process used in this study hence usefully direct further investigations towards future applications. 3. The development of a complete computer package of the programs (available at the Department of Surveying Engineering at UNB) designed to combine information similar to the one used in this study.