Abstract
The composition of ice from four case studies based on the facies, solute, stable isotope, and debris content reveals compositional differences reflective of different modes of ice formation. In Southern McMurdo Sound, there is a distinctive geochemical signature that differentiates between meteoric-origin and marine-origin ice. Analysis of the basal ice of three glaciers from the McMurdo Dry Valleys shows that liquid water does contribute to its formation. The basal ice sequences are structurally and compositionally different and are reflective of different modes of formation or entrainment active at the glacier margins. In the cases of the Rhone and Wright Lower glaciers marginal sediments and liquid water are key to understanding the accretion of debris-rich ice and both have basal facies consistent with refreezing in subzero conditions. The liquid water is formed by ephemeral melt during the summer. In the Rhone Glacier, melt water refreezes on the apron and is entrained into the advancing glacier. By contrast, by the Wright Lower Glacier adjacent streams or ponds saturate unconsolidated sediments which are entrained during ice advance. In the Taylor Glacier, the basal ice is comprised of a thick sequence of intercalated layers of clean clear ice and fine-grained debris layers. These laminated facies have a solute composition consistent with evaporites formed from a relict seawater intrusion. The combination of entrained debris, high solutes and laminations is consistent with interaction at the glacier bed and regelation. Interpreting empirically derived co-isotopic slopes is problematic, as highlighted in the case study of the Taylor Glacier where laminated facies have all the hallmarks of refrozen ice, yet plot on a co-isotopic slope that is typically interpreted as meteoric. Similarly, ice from the McMurdo Ice Shelf shows a clear difference in absolute isotope values which is interpreted as being refrozen from seawater, yet its co-isotopic plot is statistically indistinguishable from the meteoric water line. The ice compositional approach has highlighted several shortcomings. Firstly, solutes deposited in inland areas have limited solute pathways and do not distinguish between different types of ice but are useful in distinguishing between marine and continental salts. Secondly, co-isotopic analysis to reconstruct freezing history is dependent on statistically-derived interpretations which do not explain slopes that lie between physically-based models of meteoric and freezing slopes. In empirical studies, slopes between 5 and 8 are common, and are probably cosmopolitan samples. Finally, ice composition is inconsistent between similar ice types in the McMurdo Dry Valleys, as similar facies have different ice compositions, and origins. This underlines the problem with the premise that structurally similar ice facies are formed by the same process.