This study poses a novel mechanism underlying seasonal variations in the melt of Ekström Ice Shelf, linking it to seasonal variations in sea ice production, densification and destratification of the waters in the ice shelf cavity, and subsequent acceleration of the meltwater plume. To achieve this, the authors present a novel estimate of melt rates in the center of Ekström Ice Shelf with fine temporal resolution using an autonomous phase-sensitive radar system. They compare this with time series of sea ice formation rates estimated using satellite derived open water area, a reanalysis product, and bulk formulae to estimate ocean-air heat fluxes. This resulting seasonal variations in sea ice production and ice shelf melt rates approximately co-vary, with sea ice formation appearing to slightly lead the melt rates. This leads the authors to pose the hypothesis that variations in sea ice lead to changes in the density of the waters in the EIS cavity, with a higher density contrast in winter that accelerates the melt plume and thus increase the melt rates.

This topic is certainly relevant to, and of interest to readers of, The Cryosphere, and the manuscript is well-written and clear (if rather terse in places). The portion of the manuscript describing the melt rate time series is detailed, with novel results that are a valuable contribution to ongoing community efforts to quantify and understand basal melt of ice shelves. However, the portions of the manuscript that link the melt rates to sea ice production and melt plume dynamics are less convincing, and remain plausible but somewhat speculative, despite the evidence presented. To elaborate on this point:

1. The link between the plume velocity and sea ice formation is emphasized in the abstract and discussion, but is not strongly supported by the evidence presented in the manuscript. The seasonal cycles of sea ice formation and ice shelf melt rates line up nicely, but does not imply a causative relationship. Furthermore, on shorter time scales there is no clear link between sea ice formation and melt rates. The authors’ plume simulations are certainly consistent with the idea that variations in sea ice formation could lead to changes in melt plume velocity, but involve the assumption that the changes in cavity stratification result only from changes in sea ice production (rather than lateral advection, for example), in addition to various other idealizations in the plume model that the authors have noted.

2. The plume modeling comprises a relatively minor portion of the manuscript, yet seems to have developed into the central focus of the paper: for example, the title focuses specifically on the meltwater plume as the driver of the melt. This contrasts with the focus of the manuscript itself, which is centered on observations of melt rates.

In summary, the authors have produced a manuscript whose primary result is a novel observational estimate of basal melt rates, complemented by a plausible hypothesis that the seasonal cycle of the melt rates is driven by sea ice formation-modulated variations in melt plume velocity. Yet somehow the plausible hypothesis has acquired central emphasis in the title, abstract and conclusion, despite limited evidence. I recommend that the manuscript be returned to the authors for major revisions so that they can either re-frame their findings or more strongly support their claims.

Other comments/questions:

L1-2: This statement of the gap in understanding is overly vague and general: there is some previously-established understanding of ocean-driven melting in this region, and the abstract could be more specific about placing the work in this context.

L91-94: These previous estimates are only given a cursory mention in the manuscript, yet the authors’ novel time series of basal melt is one of their central results, and so warrants more extensive discussion in the context of previous work. Here, or in the discussion section, the authors should discuss the differences between their melt rate estimates vs. previously published estimates, and possible reasons for the differences between them. A key question on readers’ minds is going to be how much confidence to place in these versus previous estimates, and any information the authors can provide to help readers make that judgement will improve the manuscript.

L101: I found “mixed with a replica of the transmitted signal” to be a difficult to interpret: could the authors please explain this procedure more precisely?

Also I was also confused by the 40kHz sampling. This may simply stem from a fundamental misunderstanding of the analysis on my part, but isn’t this sampling period much longer than the two-way travel time through the ice (based on the propagation speed given later)? If so, how do the authors resolve distances down to 1mm?

Fig. 1: Might it be helpful to include a map of previously-reported melt rates, e.g. those of Adusumilli et al.? In addition to providing a clearer point of comparison, this would also help readers to judge how representative the melt variations at ApRES are of other parts of the ice shelf.

L97-115: I found the description of the methodology rather hard to follow. I think the issue is that it is written with enough detail to summarize the key steps, but not enough detail for a reader to completely reproduce all of those steps. If this methodology is standard then some of the detail could be omitted and replaced with appropriate references. If not then the authors need to be more precise here; I would recommend expressing some of the key concepts as equations, and perhaps including a figure illustrating the analysis procedure.

L109-110: “Spectral analysis” is rather vague. It sounds like the authors may be computing overlapping windowed spectra, but I am really not sure and would appreciate a more thorough description of the approach here.

L114: Please provide citations for the propagation speed and relative permittivity.

L125-133: This portion of the text suffers from a similar issue as above: the ideas are well articulated, but are difficult to follow without equations that precisely express the calculations and figures that illustrate the process.

L144: How does surface melt factor in to the ice thickness change budget?

L155: Please briefly explain how this uncertainty was computed.

Eq. (4): How accurate is the assumption of constant strain close to the upper and lower surfaces of the ice shelf? (This is somewhat out of my area of expertise.)

Section 3.2: This approach is similar to that of Tamura et al. (2016), but there isn’t any discussion of how this method compares (in approach and in result) to previous estimates of sea ice formation. Please expand this section (and perhaps the discussion section) to place this part of the work in the context of previous studies.

L216-217: This is an intriguing possibility, but I would expect tidal lateral excursions to be small relative to the size of the coastal polynya. Perhaps the authors could provide a more quantitative estimate to support their speculation?

Fig. 4: What is the time scale for the establishment of the plume circulation? Based on the ice shelf length and the plume velocity I would estimate O(10 days), which is much shorter than a seasonal time scale. Thus it is reasonable for seasonal variations in sea ice production (but perhaps not weekly?) to influence melt rates via changes in the plume velocity.

Also, the authors discuss the density contrast between the plume and the ambient water, but do not show profiles of the plume density. I suggest including these in the plot to aid the discussion.

Have the authors considered running simulations with spring-like and autumn-like thermohaline stratification? It is surprising that spring does not receive particular focus here, despite exhibiting the highest melt rates. Is there insufficient hydrographic data to construct a full seasonal cycle of ambient stratification profiles in the cavity?

Section 4.3: How sensitive are the plume model results to the particular geometry or particular stratification used as inputs? The authors have selected a particular track along the ice shelf, and have made a specific choice in idealizing the summer/winter stratification, and I think it is important to assess whether these (somewhat arbitrary) choices have influenced the results. I would not expect a strong sensitivity to either choice, but I think the authors should check and discuss this.

L228-234: How large are the density contrasts between the plume and ambient waters in the winter and summer cases?

Also, this is an appealing dynamical explanation, but it is difficult for a reader to see exactly how the changes in ambient stratification influence the plume velocity without appropriate equations that express the plume dynamics. I recommend including at least the plume momentum equation to help clarify this mechanism. A diagnosis of the terms in the plume momentum balance in the summer and winter cases would provide quantitative backing for this conclusion.

Zeising et al. present a new, almost 4-years long time series of basal melt rates from a site at Ekstrom Ice Shelf. The melt rates are inferred from an in-situ, phase-sensitive radar. It is pointed out that melting is higher in the winter months than in the summer months. The provided explanation of this seasonal difference is based on an assumed seasonal stratification change on the continental shelf. A meltwater plume model forced by two idealized profiles is used to support this claim. The plume model output is used to attribute melt rate change to either temperature or flow speed change.

The melt rate data the authors acquired and present are unique and very intriguing and I am excited by their appearance. On the oceanographic interpretation side of things, I think there are quite a few gaps and perhaps misunderstandings, furthermore, the most interesting aspects of the data don't seem to be addressed at all. I think with some more careful attention to the oceanographic interpretation this paper will be a nice contribution to the literature.

Some comments are below, other are included in the attached pdf.

* The focus of the paper is on seasonal melt rate variability, which the authors grossly simplify to higher melt in the winter and lower melt in the summer. The interpretation provided would be appropriate if the authors only possessed two melt rate measurements, one from each season and nothing else. However, the more complete data set, requires more elaborate explanations.

Some questions that aren't addressed:

1) What are the approximately monthly oscillations in the melt rates?

2) In the winter, there are times when melt rate is as low as in the summer - how do you explain that, if the winter stratification is low - presumably AASW layer is formed in the summer only?

3) While the monthly oscillations are also present in the sea ice production estimates, they do not seem to correlate well with the melt rate time series - why could that be?

4) Although melt rate and sea ice production both show seasonal signal, there does not seem to be any interannual or intraseasonal correlation there. There are a lot of seasonal signals in the system, and it is not clear to me, or it hasn't been discussed why the sea ice signal is the main driver of the seasonal melting. Also, shouldn't there be a time delay in seasonal melt increase after the seasonal sea ice production start corresponding to the dense water propagation to the grounding line?

* In general the argumentation should be a bit more specific, detailed, and supported. The following sentence, one of the few addressing the monthly oscillations I refer to above, is a good example of the vagueness of the explanations:

'It is easy to imagine that the varying intensity of the tides modulate

the efficiency of the sea ice formation in coastal polynyas, as well as the exchange of water masses on the continental shelf

with the ice shelf cavity, explaining the characteristic time scale of the observed variability.'

* The plume model gives a steady state solution. The authors force it with T/S profiles at the calving front (if I understand well). It will take a while for a change in continental shelf properties to propagate into the cavity - what is the time scale of that, and do the monthly oscillations interfere with that timescale, and therefore the applicability of the plume model to explain temporal variation?

* Some more justification of the realism of the idealized TS profiles would be useful. It looks to me that the summer profiles from Smith 2020 actually look more like your winter profiles - why is that?

Also, since the results will be sensitive to stratification, can you consider a range of plausible stratification profiles (also non linear ones) to capture some uncertainty in that, and therefore some sense of how robust the results are?

Presumably AASW will not necessarily produce a linear profile if it restratifies the water column primarily near surface?

* If the point of the paper is to distinguish between the effect of temperature vs speed on melt rate change, than that should be quantified somewhere.

* It feels in the discussion as if the authors are trying to say their observed case doesn't fit standard understanding of sub ice shelf circulation. They say that what they observe is not Mode 1 circulation and they argument with variability and importance of flow speed changes (around line 238). Mode 1 circulation is a conceptualization of circulation that fits Ekstrom well, from the information provided. Mode 1 conceptualization does not distinguish relative importance of temperature vs speed changes on melt rate change. The idea of melt rate decrease as a consequence of circulation shut down due low sea ice formation has also been suggested before (Nicholls 1997), however was subsequently shown to be too simplistic (e.g. Hellmer 2012 and Naughten 2021). So it is probably worth highlighting similar concerns somewhere in this paper and not be overly conclusive.

* Comparison and interpretation of similar measurements at Nivlisen are a bit odd/wrong. There is no shown basis for why the plume at Nivlisen should be less strong than at Ekstrom.

* Some indication of horizontal circulation beneath Ekstorm would be useful, since the authors use related platelet observations to support hypothetical ISW outflows. Related to that, some arguments why a simple 1D plume model is appropriate to model the 2D melt rates would be useful too. Are there perhaps any reliable mean satellite melt rate estimates that would support an across shelf uniform melt rate pattern?