Numerical validation of the conjecture of a subglacial lake at Amundsenisen, Svalbard

The likelihood of a subglacial lake beneath Amundsenisen Plateau at Southern Spitzbergen, Svalbard, pointed out by the flat signal within the Ground Penetrating Radar (GPR) remote survey of the area, is justified, here, via numerical simulation.This investigation has been developed under the assumption that the icefield thickness does not change on average, as it is confirmed by recently published physical measurements taken over the past 40 years. As a consequence, we have considered admissible to assume the temperature and density in-depth profiles, snow and firn layers included, to be stationary. The upper icefield surface and the rocky bed surface are known in detail.The mathematical numerical model is based on an unsteady Stokes formulation of the ice flow and on a Large Eddy Simulation formulation of the lake water flow. Following the numerical sensitivity results that we presented on a recent issue of this journal, we have, here, upgraded the model by improving the description of critical aspects of icefield thermo-mechanics, such as the local water release within temperate ice as a strain heating effect and ice sliding on the bedrock. The first issue impacts on ice texture, i.e. its constitutive equation, while the second one drives icefield surging. Actually, we have obtained 13% enhancement of the numerical value of the ice top surface velocity versus measured one, and physically consistent numerical ice sliding velocity values at the rocky bottom.Adopting a new physically sound initial subglacial lake water temperature and velocity fields, we present the numerical simulation of the whole system, icefield and conjectured subglacial lake, within a time slot of 20,000. d (physical time), when its evolution trend was clearly captured. By then, although the maximum value of water temperature keeps rather low, metastability appears to be overcome on more than half of the conjectured basin, with a progressive trend in time in support to the subglacial lake existence. We stress that the numerical subglacial lake surface converges to the GPR flat signal spot with tolerance equal to the GPR measuring error.Finally, we observe that the numerical simulation results meet quantitatively and qualitatively the fundamental aspects of the conjecture, so that further on-site investigations on the subglacial lake (e.g. drilling operations) appear fully justified.