I enjoyed this case study, which looks at the development of rifts on the Brunt Ice Shelf. The authors present a very nice suite of in-situ and satellite observations to track rift development. There is perhaps a missed opportunity to explain rift propagation through space and time beyond being generally dependent on “ice shelf geometry and degree of contact with a pinning point”. For example, how can understanding the Halloween Crack formation be applied to rifting at other ice shelves?

I have difficulty appreciating high value in the inverse modelling time slices. Given the excellent observational package, the additional modelling neither expands the time/space coverage of the study, nor yields additional process-level insight. Some specific challenges with the modelling:

• Using shallow continuum mechanics while ignoring fracture mechanics. There is clearly a lot of energy going into fracture rather than deformation here. Stresses are also changing over short length scales, meaning non-trivial coupling stresses.
• Characterizing the rheology of a floating ice shelf: With no basal drag, is it possible that ice shelves are rather low deviatoric stress environments? (Pettit2003; http://doi.org/10.3189/172756503781830584).
• By prescribing rift locations, it is difficult for the model to provide independent insight on rift processes. The inferred changes in fluidity surrounding the prescribed crevasses do not seem physically based. Or are the authors suggesting ice properties like viscosity have actually changed ~10 km from the rifts?

Where diagnostic modelling could be helpful is assessing local strain rates and principle stresses. This could provide insight on whether the initial fracture was flow perpendicular (i.e. pure Mode 1 opening) or not flow perpendicular (i.e. additional Mode 2/3 thrust/shear fracture). With the ice rumple in play, virtually any combination of mixed mode fracture is conceivable (Colgan2016; https://doi.org/10.1002/2015RG000504). The offset between principle stresses from rift orientation that would provide this insight, which might be the most applicable diagnostic modelling pursuit.

Line 29: A range of factors influencing rift propagation are mentioned, but ice properties (i.e. meteoric versus marine ice and/or damage history) seem overlooked in this listing. Presumably both could be important for the Brunt Ice Shelf.

Line 105: The reader would benefit from seeing the ATL03 product plotted along the ATL06 product for an example rift. It remains somewhat unclear why the algorithm looks for elevation gradient inflections in the ~200 m spatially averaged ATL06 product instead of elevation thresholds in the ATL03 product.

Line 120: It is unclear what “below 50% of this” means in terms of an elevation. If the mean ice shelf elevation is 200 m, for example, does this mean 100 m elevation threshold?

Figure 2a: The inset is too small, perhaps it should be its own figure? More generally on Figure 2, there should probably be a scale bar in each subfigure, given the number of spatial scales. I think that 20 sub figures are too many sub figures for a single figure. I’m also not sure if both the red and blue subset areas are needed, given their overlap.

Figure 3: It is unclear how the rift centerlines are determined. It is not precisely centered in this figure.

Line 131: A sentence is needed introducing what block-bisected rift is.

Line 159: It should be explicitly stated how velocity azimuth is determined in the Gardner2020 (https://doi.org/10.5067/6II6VW8LLWJ7) product. Many satellite-derived products are simply displacements projected down the direction of steepest surface slope of a DEM, which can make azimuths dependent on DEM choice rather than an independent 3D solution.

Line 192: This paragraph sounds more like methods than results.

Line 208: Finding that the Halloween Crack formed in the same locations in the 1968 and 2016 is perhaps a very important, but currently downplayed, finding of this study.

Table 1: What is the “Stancomb-Wills Ice Tongue” header meant for here? Also Line 219 says the total ICESat-2 rifts is 375, not 380, as shown here.

Line 231: If the reader has perhaps forgotten which are the high/low power lasers, an explicit statement here saying whether laser power influences retrieval ability would be helpful.

Figure 4: It seems unnecessary to include the ICESat-2 launch as a vertical dash in all sub figures. It seems asymmetrical as Worldview and Landsat satellite launches are not highlighted.

Line 270a: I understand these opening rates are given relative to ICESat ground tracks, but I would appreciate a clear statement on the minimal (?) of rift advection on apparent opening rates. For example, a rift with non-uniform width – even if it is not opening or closing – could still yield width changes across a given ground track as it is advected across that given ground track.

Line 270b: The observation that rift opening rate appears to briefly slow down after calving is interesting. Does this imply that iceberg A-74, while still attached to the Brunt Ice Shelf, exerted a net extensional stress on the ice shelf? Or is there possibility a kinematic wave at play?

Figure 6: It is difficult to discern the different shapes of the small markers.

Line 285: The motivation and framework of the simulations should be described earlier, in methods. Indeed, much of the model description should probably be moved from supplementary to the methods, including ice temperature/viscosity assumptions and how they interact with fluidity.

The supplementary material contains many great figures, but I wonder if much of the supplementary text could be worked into the main methods for ease of the reader? At 484 Lines, this manuscript seems to have space within the word limit.