Figure 4 . Flow velocity maps (a–f ) of six selected periods with ice front changes, and along-flow velocity profiles for C4 (g ) and C6 (h ). The locations of the C4 and C6 centerlines are shown in panel a .
5 Modeling experiments
The observational data have indicated a clear but also complex linkage between ice shelf retreat, flow acceleration, and rift development. Both front geometry change and rift development modulate the stress conditions across the ice shelf, which in turn change flow velocities and strain rates, further affecting rifting and calving. We conducted diagnostic modeling experiments to investigate how varying front geometry and rift-induced weakening control the dynamic responses of ice shelf flow velocities and stress fields.
5.1 Ice rigidity inversion and modeling scenarios
Before the diagnostic experiments, we first solved for the ice rigidity\(\overset{\overline{}}{B}\) using the inversion procedure in ISSM. We used the 450-m resolution MEaSUREs phase-based Antarctica ice velocity data product (Mouginot et al., 2019) to solve for\(\overset{\overline{}}{B}\). We used this velocity product to avoid uncertainties induced by velocity interpolation errors. This velocity product agrees best with the derived 2008–2010 flow velocities, with an average difference of 10±25 m/year. Figure 5a shows the optimized ice rigidity \(\overset{\overline{}}{B}\). This paramater defines the relationship between strain rate and stress, and is affected by ice temperature, fabric, water content, impurities and presence of cracks. Low values of \(\overset{\overline{}}{B}\) indicate “soft” ice and high values indicate “stiff” ice. The spatial pattern of\(\overset{\overline{}}{B}\) (Figure 5a) shows a good agreement with the suture zones and the zones of rapid shearing where outlet glaciers are flowing past promontories near the upglacier end of the ice shelf. The low values of \(\overset{\overline{}}{B}\) near Gipps Ice Rise (highlighted area in Figure 5a) are due to the observed rift that mechanically weakened the ice shelf, thus making the ice appear “softer” than the surrounding ice. This results from the objective of parameter inversion, which is to minimize the difference between modeled and observed flow velocities under the continuum-mechanical modeling framework. When simulating for the years without rifting, these low values of \(\overset{\overline{}}{B}\) derived from the rifted condition could artificially influence the model simulations. We modified the\(\overset{\overline{}}{B}\) values in this rifted area (Figures 5b and c) for different modeling scenarios. We considered three scenarios as follows.