Contribution of Crevasse Advection and Mixed Mode Calving to Glacier Dynamics

Abstract

Modeling of the climate system including, but not limited to, atmospheric dynamics, ocean dynamics, and ice dynamics, is one of the crucial scientific problems of the 21st century. This work focuses on modeling of iceberg calving, the process by which high stresses within ice cause fractures and eventual detachment of icebergs from glaciers. Iceberg calving causes approximately half of mass loss of the world's glaciers, but remains poorly understood and implemented in climate models. Existing model implementations of calving exist at a wide range of spatial and temporal scales, ranging from models that seek to resolve crevasse propagation on the scale of seconds or minutes to broad parameterizations of calving implemented in ice sheet scale climate models. In this work, we seek to develop an intermediate model that can run for year to decade timescales and determines calving based on the internal stresses within the ice. We first focus on crevasse advection, the memory of previously formed crevasses within the ice and their impact on calving behavior. Second, we implement the possibility for mixed mode calving, where ice can fail either via high tensile stresses, high shear stresses, or a combination of the two. Lastly, we include submarine melt to see the combined effect of melt and mixed mode calving on glacier stability. This model is developed using the highly flexible Python finite element library FEniCS and LEoPart, a particle tracking library developed for use with FEniCS. Our novel use of particles to track previously crevassed ice provides a computationally efficient method to track glacier parameters that does not diffuse over time. In initial model development, we identified a key numerical consideration related to the buoyant boundary condition on ice that can create unphysical results if not careful managed in a calving model. Once this issue was documented and addressed through a simple addition to the glacier momentum balance, including crevasse advection in the model, which should increase calving rates, reduces overall rate of glacier advance but is incapable of causing retreat in most cases. This indicates that other mass loss mechanisms, such as shear calving and submarine melt, are crucial to mass balance and should be included in future models whenever possible. When mixed-mode calving is included, we find that glacier behavior is highly unstable with regards to the shear strength of ice. Sharp transitions exists at shear strengths where the ice transitions from slow advance or retreat to rapid, catastrophic collapse. Including submarine melt causes further steady mass loss, but does not significantly alter the shear strength threshold necessary for rapid collapse. This shows that if models seek to model potential catastrophic glacier collapse, which is a point of current scientific contention, mixed mode failure including shear localization should be a key focus of modeling efforts.