Review of “Melt sensitivity of irreversible retreat of Pine Island Glacier”

The manuscript “Melt sensitivity of irreversible retreat of Pine Island Glacier” by B. Reed and co-authors investigates possible conditions that caused the retreat of Pine Island Glacier from a ridge about 40 km from its present-day conditions where the glacier was likely grounded until the 1940s. Using the Úa ice flow model and starting from an initial state close to steady-state for a cold ice shelf cavity and with an advanced grounding position on the ridge, they test the impact of warming conditions happening over a given period (about a decade) before the cavity becomes cold again. They found that once the glacier is destabilized from its ridge because of warmer conditions, it cannot restabilize on this ridge and continues retreating even if the cavity becomes cold again. They further investigate the time needed to destabilize the glacier depending on the degree of warming and its duration period.

This paper is well written, clear and very interesting. It addresses the important question of understanding the conditions that triggered past grounding line retreats. My main comment concerns the role of uncertainties in geometry and other parameters that are unknown for the 1940s should be better acknowledged and their impact discussed. The other ones are relatively minor, including a few sections that could be clarified for example to explain better what adjustments were made to the ice shelf thickness or how the friction was calibrated under the ice part that is currently floating but is grounded in the initial configuration since this cannot be inferred with the inversion. Finally, the text generalizes some conclusions without demonstrating them in a few cases.

Main comments:

l.101: What adjustments where made to the ice shelf thickness? What part of the ice shelf was impacted? And what was the magnitude of this correction? Some overall numbers should be added to the text.

l.104: Additional information should be added here to understand what is done to the friction in the part of the domain that was grounded in 1940 (and therefore a friction coefficient is needed) but is not grounded during the satellite area (and thefore cannot be inferred with observations). How was this friction chosen? Also, given there is a large uncertainty in these values, what is the impact on the results presented?

l.133: why was a period of 1000 years chosen to simulated the grounding line advance? And why was the melt set-up to zero instead of cold conditions? I am sure a number of conditions could lead to a more or less similarly advanced position, so I am curious why such conditions were chosen?

l.143: similarly to my question about the grounding line advance, why was a period of 100 years chosen for the initial grounding line retreat?

Fig.6: the small inset showing the thermocline depth and its timing is very useful to understand the experiments and should be provided earlier (for example in the description of experiments) for improved clarity.

l.267: the experiments performed show this behavior for Pine Island Glacier but not that it could happen to other places at the same time, so this statement should be toned down.

l.321: How does the retreat and possible readvance from this ridge compare to the present day conditions investigated in previous studies such as Favier et al. 2014 or Seroussi et al. 2014?

Technical comments:

l.40: there was not much slow down reported in Mouginot et al. 2014 outside of the Eastern Thwaites ice shelf, maybe less acceleration or relatively stagnant conditions, but not really a sector wide slowdown and the discharge kept acceleration at least remained constant.

l.83: why not use the actual velocity at the divide instead of zero? It is unlikely the velocity changed much during this period.

l.92: What is the refinement? It would be good to put an actual number to get at least the order of magnitude in the text.

Fig.1: the blue line for A-B on panel 1 is hard to see

Fig.2 caption: Why grounding line is displayed on panel c?

Table 1 and text lines 172-179 : it would be great to add the total number of experiments performed as part of the WARMvar and CODLvar cases. Were all the possible combinations tested? If not which ones were tested and how was that decided?

Fig.3: it would be good to add the years at the top of the corresponding columns

Fig.4 caption: for which experiement are the vertical black dashed lines?

l.220: “there is A continued retreat”

l.300: It looks like the glacier continues to lose mass, so what does “stabilized” mean in this context?

Review of Melt sensitivity of irreversible retreat of Pine Island glacier, Reed at al.

This study explores the sensitivity of Pine Island Glacier in the Amundsen Sea Embayment of West Antarctica to basal melt. Specifically, the authors investigate potential retreat from a ridge ~40km downstream of the current grounding line that occurred after an unusually warm anomaly in the 1940s. Using an ice sheet model (Ua) and a depth-dependent basal melt parameterisation, they create an historical state by initialising the model into the present day and the turning off melt to enable readvance to the downstream subglacial ridge. They then run forward ("relax")  for another 100 years with a cold ocean forcing. Starting from this initial state, they find that with a warm forcing of 12 years of more, the ice continues to retreat and doesn’t restabilize on the ridge even if conditions are subsequently returned to the cold conditions from the original steady state. They further investigate the duration and strength of melt that is required to lead to unstable and irreversible retreat, and find that even short periods of high basal melt rates are sufficient to force such a retreat, with near zero basal melt required to stop this and lead to readvance towards the ridge.

The manuscript is clear and well written, and the experiments performed are appropriate and insightful. The study compliments the recent Reed at al, 2023, and allows a more thorough exploration of the basal melt regimes (timing and duration) that lead to irreversible retreat in this region. The research questions highlighted in the last paragraph of the introduction are clearly stated and answered by the study.

I don’t see any major issues but feel the paper would benefit from a more thorough discussion of both some of the modelling choices and implications, and of the viability of some of the basal melt regimes (in terms of thermocline variation) that are explored.

Main points:

• Would results have been different if you had readvanced the grounding line under cold conditions, as used for the relaxation 100 years, rather than the no melt? (described in methods L95-100, and L130-145). Would the grounding line still have advanced, and would the steady state have been the same? Do you think the position of the steady state has any impact on the retreat i.e. would results have been any different if you’d initially advanced using cold forcing only?
• I think the paper would benefit from more details/discussion of the modelling choices made in the initial state. In particular:
  • There are downstream basal drag values that cannot be obtained from the initialisation because they are under ice that is floating in the present day. The authors state that these are set to a constant value (L392 and Figure C1) but this value doesn’t seem to be stated. What is the value and why/how was it chosen? And how does, or might, it affect the future simulations? How can you be confident the results aren’t highly dependent on this choice?
  • On L138 the authors state that the lack of advancement beyond the subglacial ridge is “aided by the fixed calving front” – can you speculate on what you’d expect if the calving front advanced? Would you expect a different steady state/more advance of the grounding line? How much of limitation is this fixed calving front for the study?
• How realistic are the thermocline heights and prescribed melt rates?
  • Can you give more context, either in “2.6 Perturbation experiments” or in the Discussion, for the thermocline profiles chosen? Can you comment on how extreme are some of these e.g. the standard warm profile using in WARM25 etc, and the really high melt profiles that allows unstable retreat after just 5 or so years of forcing? (introduced in 3.4 Mapping the stability regime). It is stated that thermocline depths of 1100-1200m are unrealistic, so discussing the realistic range of warm forcings would also help, maybe around L315-320.
  • In addition, on L38-40, you mention “shallow” and “deep” thermoclines– how shallow/deep? What was the depth here compared to those explored in this study?
  • L117-118: thermocline depth varies, but not this melt rate, correct? Can you put this choice of 100 m/a into context here? What do you think would happen if you explored changes to this melt rate as well as thermocline depth?

Minor points:

• L60-61: “finished when the glacier stabilized at an ice plain in the early 1990s”. How do the suggest timings fit together here, when put together with L45-50: “The glacier has been retreating across an ice plain since the early 1990s” and “the subsequent ice loss was unaffected by the reduced basal melt rate in the early 2000s”? It seems from this text that the glacier stabilized in the early 1990s but was also trigged to unstably retreat at that time too?
• Can you deduce anything from your results about whether the unstable retreat from the 1940s can be attributed to anthropogenic change or natural variability alone? It seems to me that no trend in warming is required to sustain the retreat – quite the opposite, in fact, because in many cases the ocean has to become colder than it originally was to halt the retreat. So does this lead you to conclude that this unstable retreat could be due to natural variability alone?
• L119-122: it may help the reader if you link to the thermocline plots in Figure 2b here.
• L138-140: The initial state from the 1000 years of no melting is “not far from the 1940s position” and after relaxation for 100 years “the new state represents the approximate situation prior to the warm anomaly in the 1940s”. I’m curious how well defined the 1940s state is in Smith et al. 2017, and whether it is clear that the relaxed state matches it more closely than the unrelaxed steady state?
• Figure 3, middle row: these melt rates look a bit stripey here, why is that?
• Line 200: can you comment on the timescale of the retreat here? Is it consistent with what is observed in the 1970s, or a bit slower? Would you expect it to capture the timescale?
• Figure 4, black dashed line indicates time of melting starting upstream of the ridge – but presumably just for cases WARM12and the cold cases? Please clarify.
• Figure 6: can’t tell that the dashed line is dashed.
• L290-295: you note that the stabilisation on the prograde slope might have coincided with cold ocean conditions – but would they be necessary for stabilisation in your model, or does it stabilise there even in warm conditions?
• L332: Bett et al, 2024, also use a coupled ice-ocean model and find that ocean melt around pinning points is a key control on the retreat.
• Figure D1: Lines for t=12 years and t=25 years hard to distinguish.
• Figure E1: For the reversible cases the final state grows compared to the initial state – is the final grounding line position similar to the steady state or significantly more advance? Presumably not more advanced that the no melt case from the first steady state (after 1000 years of no melt)?

References:

 Bett, D. T., Bradley, A. T., Williams, C. R., Holland, P. R., Arthern, R. J., and Goldberg, D. N.: Coupled ice/ocean interactions during the future retreat of West Antarctic ice streams, The Cryosphere Discuss. [preprint], https://doi.org/10.5194/tc-2023-77, in review, 2023.