Figure 6 . Second principal stresses and compressive arch locations for different front geometry conditions (non-rift scenario). F1–F4 are the front locations in 1963, 2000, 2013, and 2017, respectively. H1–H5 are the hypothetical front locations under future retreat scenarios.
5.2 Dynamic responses to front retreat (non-rift scenario)
The retreat events during 1963–2020 neither exceeded nor modified the compressive arch (Figures 6a–d). Under future retreat scenarios, successive retreat would allow the compressive arch to migrate upstream until H3 (Figures 6e, f). The front geometry change due to retreat can result in flow accelerations and backstress loss (Figure 7), with the magnitude and extent depending on the location of the retreat. The two retreat cycles reduced the backstress at the areas adjacent to the ice rises and the central downstream portion (Figure 7i–k). Flow accelerations (greater than 20 m/year) caused by front geometry change are limited to the central downstream portion and are less than 50 m/year for the majority of Larsen C (Figures 7a–c). Therefore, the front geometry change due to the past retreat events is insufficient by itself to explain the observed velocity increase. We compared the derived flow velocities with the modeled flow velocities of corresponding front geometry. Over the downstream portion of the ice shelf, the averaged difference increases over time with the propogation of R1 and R2 from 9.8±8.5% (2000–2002, F2 condition), to 16.3±8.6% (2008–2010, F3 condition), to 17±12.9% (2013–2015, F3 condition), and to 23.4±24.5% (2015–2016, F3 condition). This suggests that, to make a realistic prediction of flow velocity change due to ice shelf retreat, it is necessary to first consider rift development in response to front retreat. Nevertheless, the modeling scenario considering only the front geometry condition can be used with the observational data as a diagnostic indicator. The increasing difference between observed and modeled flow velocities over time may be an indicator for future calving events.
However, even without considering rifts, once the ice front reaches the current compressive arch, widespread flow accelerations (~25%) would be triggered (Figure 7d), as a result of the dramatic backstress loss (Figure 7l) from Bawden Ice Rise and Gipps Ice Rise. This emphasizes the important role of ice rises in stabilizing the ice shelf. Complete buttressing loss from ice rises would cause ice-shelf-wide flow accelerations, and the backstress-reducing effect would extend to the grounding line. After the compressive arch ceases to exist, even small perturbations in the ice front would cause large-scale backstress loss and flow accelerations (Figures 7g–h, o–p), particularly over the C2 and C6 units. This effect would reach the grounding line, and the upstream glaciers feeding the C2 and C6 units would speed up in response to the buttressing loss caused by a small retreat at the ice front. This scenario resembles the pre-collapsed Larsen B, for which the compressive arch was removed and flow accelerations occurred across the ice shelf. These modeling experiments suggest that ice rises are crucial for determining the tipping point of ice shelf stability. The compressive arch is the critical boundary, beyond which the ice shelf flow velocities and stresses are highly sensitive to front retreat. Once the compressive arch is absent, front geometry change will become an important control on ice shelf flow dynamics and stress conditions.