Sharp contrasts in observed and modeled crevasse patterns at Greenland's marine terminating glaciers

Crevasses are affected by and affect both the stresses and the surface mass balance of glaciers. These effects are brought on through potentially important controls on meltwater routing, glacier viscosity, and iceberg calving, yet there are few direct observations of crevasse sizes and locations to inform our understanding of these interactions. Here we extract depth estimates for the visible portion of crevasses from high-resolution surface elevation observations for 52 644 crevasses from 19 Greenland glaciers. We then compare our observed depths with those calculated using two popular models that assume crevasse depths are functions of local stresses: the Nye and linear elastic fracture mechanics (LEFM) formulations. When informed by the observed crevasse depths, the LEFM formulation produces kilometer-scale variations in crevasse depth, in decent agreement with observations. However, neither formulation accurately captures smaller-scale variations in the observed crevasse depths. Critically, we find that along-flow patterns in crevasse depths are unrelated to along-flow patterns in strain rates (and therefore stresses). Cumulative strain rate is moderately more predictive of crevasse depths at the majority of glaciers. Our reliance on lidar limits the inference we can make regarding fracture depths. However, given the discordant patterns in observed and modeled crevasses, we recommend additional in situ and remote sensing analyses before Nye and LEFM models are considered predictive. Such analyses should span extensional and compressive regions to better understand the influence of advection on crevasse geometry. Ultimately, such additional study will enable more reliable projection of terminus position change and supraglacial meltwater routing that relies on accurate modeling of crevasse occurrence.