The dynamics of the Amery Ice Shelf from a combination of terrestrial and space geodetic data

Understanding and monitoring the changes occurring in the Earth's climate are of increasing interest to scientists, legislators and policy makers, environmentalists and to the wider-population. The Antarctic continent is particularly suitable for studies into climate change, as it is the major geographical region where the climate signature influences the production and storage of water and ice. The outlet glaciers and ice shelves are the most dynamic regions of Antarctica and as such can be expected to respond to variations in the climate signal sooner than the other regions. Unlike the glaciers, which may have response lags of hundreds or thousands of years, the ocean-ice interface of the ice shelves are sensitive to changes in sea level or ocean temperature, resulting in a much faster response times. If an accurate determination of change is to be made, it is of critical importance that baseline (and future) measurements of change and rate of change are of the highest accuracy and precision. The Lambert Glacier-Amery Ice Shelf system is a major feature of East Antarctica. This study uses a combination of historical-terrestrial and modern-space geodetic measurements to increase our understanding of the dynamics of the Amery Ice Shelf (AIS). Parameters of interest are horizontal and vertical velocity and strain. Vertical ice shelf motion induced by the tides is also of significant interest. The available data spans approximately 30 years, providing an opportunity to test for change in the state of the ice shelf. Continuous Global Positioning System (GPS) measurements that are longer than one day were processed using a segmentation technique for use in tidal studies. Tidal signals are evident as far south as 73¬¨‚àûS, confirming that the southern-most part of the grounding zone is more than 200km to the south of the previously adopted position. Upon analysis, significant north-south amplification is revealed in each of the tidal constituents caused by a reduction in the depth of the water column close to the grounding zone. First-estimates of the height of the water column are made for the southern-most sites based on an analysis of the GPS-derived tidal constituents. Higher order and non-linear tides are evident in the residuals and thus need to be modelled for high accuracy applications. Comparisons were made with three different numerical tidal models with a view to addressing fundamental corrections for future satellite missions, such as ICESat and CryoSat. These missions will require high resolution, accurate tidal models to remove the tidal signals from ice shelf measurements. Significant amounts of in situ tidal measurements (>2 months and up to 1 year) are required for input into the tidal models if these missions are to reach their promised accuracy. To provide sufficient ice shelf tidal measurements, an inexpensive autonomous GPS system was designed to run for periods greater than 8 weeks during the Antarctic summer, yet can be easily modified to make longer term measurements of up to one year. The terrestrial ice shelf surveys of 1968-69 and 1969-70 provide the backbone to the current glaciological understanding of the AIS. This investigation shows that the widely circulated coordinate and velocity results from this survey had significant error. The data are reanalysed using modern least-squares methods and new results provided in a modern, stationary (tide-free), highly precise and accessible International Terrestrial Reference Frame (ITRF96). Coordinate uncertainties (95% confidence level) of the new ITRF solution are generally less than Sm, while the velocity uncertainties are better than 3.7myr‚ÄövÖ¬™¬¨œÄ in magnitude and 0.6¬¨‚àû in flow direction. More recent GPS data, collected from 1988 onwards, were analysed using modern least squares processing techniques in which the results of each field season were held as quasi-observations and then combined in the ITRF97. Velocity uncertainties (95%) are generally less than 3.9myr‚ÄövÖ¬™¬¨œÄ in magnitude and 0.8¬¨‚àû in flow direction. The final velocities (both terrestrial and GPS) were compared to an independent set of velocities derived from sequential Radarsat images, revealing systematic biases in the remote sensing velocity magnitude of ¬¨¬±20-40myr‚ÄövÖ¬™¬¨œÄ and 1-4¬¨‚àû in direction. The combined terrestrial and GPS velocities represent the most comprehensive, consistent and accurate set of velocities from in situ data presently available for the AIS. A congruency test is performed to test the hypothesis that the ice shelf velocity profile is unchanged during the period 1968-1999. This approach provides an overview of both change and rate of change occurring on the Amery Ice Shelf for the first time and hence provides an important contribution with new information on the ice shelf dynamics. The test is shown to fail at the 95% confidence interval for velocity directions and velocity magnitude. While the sample size is small (9 station pairs), the sign of the differences are essentially uniform, with the GPS velocities being less. Consequently, we suggest that the ice shelf may have slowed by 1-5myr‚ÄövÖ¬™¬¨œÄ during the thirty-year period 1968-70 to 1998-99 where velocity comparisons are available. Such a slowing may be a result of changes in one of the other balance forces acting on the Amery Ice Shelf (temperature of the ocean, snow accumulation, ice discharge rate). Strain rate estimates can be deduced through the combination of the terrestrial data and the quasi-observations from the GPS analysis using four dimensional integrated geodesy. Strain rates are determined using both the individual terrestrial and GPS velocities along with an analysis combining the two data sets. This combined strain solution represents the most precise and spatially dense set of strain grids available for the AIS from in situ data. Further strain information is gathered from computing forward and transverse strain rates in the direction of the terrestrial traverse. These results agree with those strain rates computed from Radarsat images. Using a combination of the derived velocities (horizontal and vertical) and strain rates, combined with snow accumulation, surface slope, ice thickness and water and ice density information, melt rates were determined at four GPS locations. The results generally agree with estimates from two numerical ocean circulation models and measurements using velocities and strains derived from Radarsat images, although the precision of the available accumulation measurements limits the precision of the derived melt rates. Accurate ground data plays a vital part in the calibration and validation of remote sensing measurements, especially for radar altimeters (ERS-1/2), laser altimeters (ICESat), synthetic aperture radar (SAR) measurements and the new generation of satellite gravity missions (GOCE/GRACE). What is needed for this to occur is GPS data collected and processed to the highest precision in a consistent reference frame, while longer GPS time series are required to measure and hence correct biases introduced by ice shelf tidal motion. This thesis provides methods and recommendations to achieve these goals. Such calibration/validation activities will undoubtedly play an essential role in the integration of future data and measurements for ice sheet mass balance studies.