Water-Content of Glacier-Ice: Limitations on Estimates from Velocity Analysis of Surface Ground-Penetrating Radar Surveys

Abstract

The rheology of ice is strongly controlled by its liquid water content. Since water content and its distribution also exert a strong control on radar propagation velocity and attenuation, this provides a potential remote technique for assessing ice-water content. A suite of surface ground-penetrating radar (GPR) surveys have been undertaken on two glaciers, Tsanfleuron Glacier in the European Alps and Bakaninbreen, Svalbard in order to determine their water content variation with depth. Common-offset surface radar profiling shows both glaciers have a two-layered structure, with a shallow layer characterised by low returned radar power and a deeper layer characterised by strong scattering. The thickness of these layers varies rapidly across the glaciers. In order to provide a robust interpretation of the properties of the layers within these glaciers we present a quantitative semblance analysis of two common midpoint surveys, making estimates of layer thickness and water content. This analysis includes a Monte Carlo estimate of the likely resolution of these estimates, as well as of the effects of the non-minimum-phase nature of the GPR wavelet. In Tsanfleuron Glacier, the shallow layer consists of relatively dry ice with an estimated water content of 1.18 [Formula: see text], whereas the deeper layer is interpreted as containing small water bodies, 3.90 [Formula: see text] by volume. At Bakaninbreen, the shallow layer contains no water and the deeper layer 1.29 [Formula: see text] water. At both glaciers the deeper layer will undoubtedly be rheologically softer, with implications for ice dynamics; furthermore, the layer will provide a store for a substantial water volume with important implications for the glacier’s water system. The uncertainty in the calculated water content is, however, significant in terms of ice dynamics, meaning that alternative methods of assessing water content, possibly using borehole radar, are required to provide the input for predictive models of glacier flow.