Johnson et al describe an apparatus with which it is possible to obtain measurements within 0.5 m of the ice-water interface on growlers. They demonstrate the utility of this apparatus by obtaining short (less than 3 minute) time-series of water temperature and sound, as well as detailed images of the ice face, from some growlers in Hornsund Fjord, Svalbard. This apparatus permits measurements in the proximal boundary, which have previously mostly been limited to the quasi-horizontal ice-ocean interfaces or to snapshot observations further from quasi-vertical ice-ocean boundaries. The exception to this are recent observations obtained using IceFin, but such deployments are costly. As such, measurements obtained using this apparatus have the potential to improve our understanding of ice-ocean interaction in the vertical regime distal to subglacial discharge plumes, where existing parameterisations seem to perform poorly.

Whilst the apparatus has the potential to deliver important measurements, I question whether it is necessary to publish a paper dedicated to describing the apparatus and a test deployment. A lot of the manuscript is dedicated to a lengthy description of the deployment of the apparatus, which I’m sure will be useful information to anyone who uses the apparatus in future, but I’m not sure a paper is the best medium to disseminate that information. Instead, this information (and the rest of the description of the ice frame) might be better given in a users manual, which could be provided alongside a manuscript describing the important measurements and findings that the apparatus has been used to obtain (or could be used to obtain).

Whether or not this information remains in a paper or a user guide, I have some minor suggestions:

Line 1 and elsewhere: I think “apparatus” or similar would be more appropriate than “method”. Further in the abstract the framing of the sentence around a “method” means that the ice frame doesn’t really get described in the abstract and (as written) it just reads like you screwed the instruments directly to the glacier ice.

Line 27: “such processes” is a little ambiguous. Can you expand, or specify that you mean processes related to pressurized bubbles in glacier ice

Line 28: “general agreement” – can you be more specific regarding the variable or behaviour here?

Line 29: “models also typically neglect the effects of bubbles” – consider rewording to “but neither the laboratory studies or numerical simulations include the effect of pressurized bubbles” (or similar).

Line 31: “submarine melting” should I think be “submarine melt rates”.

Line 43: perhaps specify that you mean the vertical part of icebergs here, because several studies have very prolonged measurements of horizontal ice faces. Also try to be more specific than “prolonged”, given that the time-series presented here are limited to a few minutes in duration.

Section 2.1 or elsewhere: can you provide the weight of the frame somewhere?

Section 2.1: can you provide the cost of the frame materials? (even just the total cost)

Section 2.1: I think the ice screw depth should be given here as well as an estimate of the melt out time during the field deployment (along with the weather conditions at the time) – could you have deployed the instruments for much longer under cooler conditions with larger ice screws?

Section 2.1: I think you should include how many attachment points for instruments are there on the current configuration of the frame? Could that be adjusted within reasonable weight constraints?

Line 114: related to the above, “Once enough data had been gathered”, is quite ambiguous. Please be more specific to your setup during this deployment.

Section 3.4: “in principle other properties of the water…” – could you also measure the distance from some point on the frame to the ice face? Could that give you a direct measure of melt rates? (similar approaches are used to measure accumulation and ablation on glacier surfaces)

Line 207: “rods would rest against the submerged face” – I can see how this works if the iceberg tilts towards the ice frame. Does it also work if the iceberg tilts away? Can you make that clear either way in the text?

Figure 2 or a separate diagram: provide the measurements of the ice frame struts. Perhaps also provide the packed dimensions.  

Figure 3: there are some interesting fluctuations in the temperature data with a wavelength of approximately 50 seconds. Are you able to determine how they come about? Or at least demonstrate that they are not caused by movements of the growler relative to the sensors?

General comments:

Johnson et al. present a novel technique for collecting in situ measurements of the near ice-water interface of glacier ice using a frame that can be equipped with a suite of instruments and attached to the ice. Understanding heat transport through the near ice-water interface of glacier ice is key to accurately estimating submarine melting, which can be underestimated using current theory according to some observational studies. The technique of Johnson et al. seeks address a current lack of direct observations with scope to improve glacial melt estimates in the future. The authors tested their apparatus, fitted with a hydrophone, underwater camera and thermistor string, on growlers in Hornsund Fjord, Svalbard. I think this is an exciting technique and a nice, if very brief, showcase of some measurement capabilities. I think that, with some adjustments, a paper is a nice way to present this research as a proof-of-concept study.

Specific comments:

I agree with Referee #1’s comment that the deployment description is quite lengthy, for example, details about boating safety and frame redeployment after unsuccessful attempts could be omitted. I also wonder if further data exploration might be feasible/possible. Temperature measurements spanning 20 minutes are mentioned in section 3.4, but only 2.5 minutes worth of data are shown. If images and audio were also collected for the 20 minutes, could you present coincident data and look for, e.g., images of bubbles being released (learning more about this seems to be a major part of the justification) corresponding to audio pulses and maybe being seen somehow in the temperature data? Some kind of quantified results from what I’ve mentioned above or even something like bubble release rate from the audio data that you could compare with other literature to show that your apparatus is providing novel observations would really help the discussion which is mostly rationale and outlook at the moment. The outlook is good though and proposes some interesting future applications for your system.

Briefly on the frame design, could the pair of flotation spheres collide and interfere with audio data? Perhaps something like a large, sealed PVC pipe could be attached parallel to and above the crossbeam instead of the spheres (or the aluminium crossbeam could even be replaced by a larger floating one).

Technical corrections:

Line 15: Consider moving details of salt and gas to the third sentence where they’re mentioned and then combining the first two sentences.

Line 19: Replace “this interface” with “the proximal boundary”.

Line 35: Poorly how?

Line 51: Please check style guide regarding capitalising figure # and appendix #.

Line 107: Remove “then” after “crossbeam was”.

Line 121: How was the size range 3-5 m determined? Also, spaces should be removed either side of dashes in this paragraph or added elsewhere.

Line 133: Quantify? “Enabling calculation of a rise velocity of …”.

Line 141: “figure 3e-f”.

Line 156: “…ice face of a floating growler about 30 cm below the sea surface…”.

Line 183: What was the wall thickness of the tube?

Line 185: What material was the threaded rod?

Line 209: “remain”

Figure 2: It would be nice to show the thermistor array and the optional standoff beam. It would also be nice to label the instruments with what they actually are.

Figure 3: As mentioned above, it would be nice if more data could be included here. If you have images of a bubble being released and rising, I think it might be very nice to include multiple images showing this. If the ice-face images could be linked to the audio, I would be great to timestamp the images so that they could be linked to a longer time series of audio data. I understand that this may not be possible due to distance between instruments, but that distance may be an important consideration for future deployments if the goal is to link data from all the different instruments.

Figure 3a: This may be easier to relate to panels c-f if time were put on the x axis (though I also understand the logic of having distance on the x axis). It might be interesting to include a black line at some specific temperature contour to better see patterns.

Figure 3d: In the caption, “A blown-up view of a bubble-release pulse recorded from the ice frame. e…”

I hope these comments are helpful and I look forward to seeing future development of the ice frame.

Thanks