Melt Undercutting and Calving from Tidewater Glaciers: Observations and Model Results

Dynamic models incorporating crevasse-depth calving laws have enjoyed considerable success in simulating observed behavior of tidewater glaciers. Such models are based on the assumption that longitudinal strain rates exert a first-order control on calving, and that penetration of surface and basal crevasses provides the ultimate constraint on glacier extent. However, ‘second-order’ processes such as melt undercutting may significantly amplify calving rates, initiating seasonal and longer-term glacier retreats. We present high temporal and spatial resolution TerraSAR-X data from Svalbard that indicate a strong annual cycle in calving rates, peaking in September-October coincident with maximum fjord temperatures. This pattern is consistent for all studied glaciers irrespective of glacier activity (fast, slow, surging or quiescent), and we conclude that in Svalbard calving is paced by melt-undercutting followed by mechanical destabilization of the ice tongue.

Although parameterizations of melt undercutting are included in many models employing the crevasse-depth calving criterion, amplification of calving by melt undercutting (the ‘O’Leary Effect’) has not been rigorously analyzed or tested against observations. We take a novel approach to this problem, and couple the finite element model Elmer-Ice with a discrete particle model (DPM) to explore in detail the links between melt undercutting and failure of the ice tongue. Employing glacier front geometries representative of Kronebreen (Svalbard), Columbia Glacier (Alaska) and Helheim Glacier (Greenland), we use Elmer-Ice to simulate progressive undercutting of the ice front by melting. At selected time steps, the model geometry was exported into the DPM, and runs conducted to study fracturing and calving behavior using different values of the fracture stress. We quantify the O’Leary Effect for different geometries, and propose a modified calving law incorporating the effects of melt-undercutting. The results highlight the importance of accounting for the impact of melt undercutting on calving losses in dynamic calving models.