4.2 Evaluating Crane Retreat against Process Models
Models of ice cliff failure each have a stability threshold, defining the boundary between configurations that naturally fail and those in which cliffs are stable and glaciers will thin and evolve by viscous processes. We evaluate the behavior of Crane against published ice cliff process models (Bassis et al., 2021; Bassis & Ultee, 2019; Bassis & Walker, 2012; Clerc et al., 2019; Crawford et al., 2021; Parizek et al., 2019; Ultee & Bassis, 2016).
A number of mechanisms exist under the “cliff failure” umbrella, including: fracture-induced full-thickness iceberg detachment (Bassis et al., 2021); initial slumping followed by buoyancy-driven retrogressive block rotation (Parizek et al., 2019; Bassis et al., 2021; Crawford et al., 2021); and forward block rotation resulting from surface crevassing (Crawford et al., 2021). Each model is built from a different physical framework, and therefore requires as input a different quantification of the material strength of ice. Ice strength appears in models through terms like the fracture toughness (Clerc et al., 2019; Crawford et al., 2021; Parizek et al., 2019), the yield strength (Bassis et al., 2021; Bassis & Ultee, 2019; Crawford et al., 2021; Ultee & Bassis, 2016), or critical strain thresholds (Crawford et al., 2021). Ideally, these terms can be constrained by laboratory measurements using physics first-prinicples (Druez et al., 1989; Petrovic, 2003; Xian et al., 1989), but in practice, they remain uncertain, as they are affected by the uncertain history of ice damage that modifies real glacier systems (Borstad et al., 2012; Lhermitte et al., 2020; Mobasher et al., 2016).
Because of this parametric uncertainty, the models we compare against observations at Crane produce a range of possible cliff-failure thresholds, assuming strength values ranging from intact or undamaged ice (strong) to ice with a lower yield strength (Ultee & Bassis, 2016), lower fracture toughness (Clerc et al. 2019; Parizek et al. 2019), or ice including starter cracks (Bassis et al., 2021) (weak). We proceed using each model’s self-described optimum parameters, but present alternative “strong” or “weak” failure thresholds in Figure 3 if available (see the Supplementary Materials for a full description of model thresholds and strength categorization).
Using nominal (strong) ice, process models require greater cliff heights (Bassis & Ultee, 2019; Bassis & Walker, 2012; Clerc et al., 2019; Crawford et al., 2021; Parizek et al., 2019; Ultee & Bassis, 2016) and faster timescales of ice shelf collapse (Clerc et al., 2019) than we observe at Crane Glacier to initiate retreat by cliff failure (Figure 3). When ice is treated as weak or damaged, process models predict that cliff failure could initiate at Crane with a terminus height of 111 m. This suggests that it is possible for accelerated calving to have occurred by cliff failure, and if so, models might more accurately predict MICI initiation if they consider weak ice a realistic scenario, rather than an extreme one.