Ice height change in East Antarctica derived from satellite laser altimetry

Abstract

In the last two decades, satellite altimetry has given the scientific community an unprecedented amount of data, which has substantially increased our understanding of the rate of change of ice surface height (dH/dt) over glaciated regions. This can be attributed to better spatial and temporal coverage of polar regions and the increased accuracy of laser and radar satellite altimeters. This accuracy is dependent on minimising errors and reducing the uncertainties of estimates of dH/dt, which are derived from ice height measurements. There are a number of different factors that contribute to the overall uncertainty budget. In this thesis, an alternative method to crossover and along-track analysis is proposed and is applied to (ICESat) height measurements. A new method of estimating surface slope at crossovers is presented and used in conjunction with the newly proposed along-track method. Particular emphasis is placed on the formal propagation of interpolation uncertainty and surface topography bias, which is often given little attention in the literature. The proposed methods are tested using a number of simulated datasets for Enderby Land and surrounds in Antarctica. The simulated datasets are derived from ICESat data, with different levels of spatially correlated noise applied to each, dependent on regionally specific ice velocities. Both the error (the difference between simulated and estimated dH/dt) and the uncertainty (a function of the interpolation distance and surface slope) are derived. It was found that the formally propagated uncertainty made a good approximation of the error and both the crossover and along-track methods were found to have the lowest uncertainty and error when using Green's function spline interpolation. The errors and uncertainties due to interpolation were an order of magnitude smaller than those obtained from the slope correction method. The overall uncertainty was found to be approximately 50% of the ICESat single-shot uncertainty budget, showing the relative importance of including these often-overlooked contributors in the final uncertainty budget of ice height rate estimates from altimetry data. The proposed methods were then applied to actual ICESat data over part of East Antarctica, including Enderby, Kemp, MacRobertson and part of Dronning Maud Lands. The dH/dt results for the study site generally showed an increase in the rate of change of ice surface height. Although most of the study site was gaining height, there were some regions with negative dH/dt estimates, such as directly behind the grounding line of the Amery Ice Shelf. These negative rates tend to have little impact on the overall estimates of dH/dt, as they are localised to very small regions. The positive rate of height change in the interior was found to be statistically significant, especially near Dronning Maud Land. The uncertainties calculated for this study do not include the ICESat single-shot uncertainty budget, as the focus of the study was the uncertainty contributions of interpolation and surface slope bias. The combination of these uncertainties would decrease the significance of the inland signal, however the large number of positive dH/dt estimates found in the interior does suggest that the ice height surface is increasing for this region, implying a positive mass balance change may be occurring in the interior of East Antarctica.