Paper is on a topic of interest - detection and characterisation of subsurface eddies in the Arctic Ocean. These eddies potentially play a key role in mixing and so are of general interest to Arctic oceanography.  Overall, I find the paper easy to follow, with a rigorous analysis of the data used in picking out of the eddies. Accordingly, I think the paper is a valuable contribution and will be of general interest.

However, at this stage the paper reads more like a data report than a scientific paper. For example, in the intro there is a detailed review of Arctic eddies, but there is virtually no discussion of what the eddies "do" within the context of the Arctic ocean system. Such comments would greatly strengthen the justification for the paper by linking to broader Arctic Ocean issues. This could easily be added by referring to recent reviews of the topic (eg. van Appen et al, 2022 - which is already referred to a different context, and Lenn et al (2021). Ocean Mixing: Drivers, Mechanisms, Impacts. Meredith, M. & Garabato, A. (eds.). Elsevier, p. 275-299).

With regard to the discussion on generation mechanisms I believe this discussion could also be usefully widened in an Arctic context. (eg.  see MacKinnon et al (2021).  A warm Jet in a cold ocean. Nature Comms, 12, 1, p. 2418; Shultz et al (2021). Turbulent mixing and the formation of an intermediate nepheloid layer above the Siberian continental shelf break. Geophys Res Letts).

The manuscript "Intrahalocline eddies in the Amundsen Basin observed in the distributed network from the MOSAiC expedition" presents a valuable and well-documented observational study of IHEs in the central Arctic Ocean. The topic is highly relevant and timely, addressing a clear knowledge gap regarding the occurrence, structure, and dynamics of (sub)mesoscale eddies beneath sea ice during winter conditions. The work provides novel insights based on a unique multi-platform dataset from the MOSAiC DN, and the methodological approach, including the use of the Maximum Swirl Velocity (MSV) technique, is both interesting and appropriate. Overall, the study meets the scientific standards expected for publication and should be accepted after revision. The revisions required mainly concern clarity, internal consistency, figure-text integration, and transparency for reproducibility, rather than the scientific validity of the results. 

Here are my detailed observations and suggestions regarding the manuscript:

- General: 1) Include figures similar to Figures 4 and 6 for all detected eddies as supplementary material. This would provide a complete visual reference for each identified IHE and enhance the reproducibility and transparency of the analysis; 2) The statement that eddy generation and trajectories cannot be directly observed under Arctic sea ice is repeated several times throughout the manuscript. While this is an important contextual limitation, mentioning it once—ideally in the Introduction or early in the Discussion—would be sufficient.

- L6-7: Consider removing the explanation of the ADCP acronym, as it is a widely known instrument within the physical oceanography community.

- L12-14: “…, placing them in the (sub)mesoscale dynamical regime, consistent with quasi-geostrophic behaviour.”. Later sections report Rossby numbers up to 0.62, implying partial cyclogeostrophic balance. Suggest rewording to reflect a transitional dynamical regime where both geostrophic and cyclogeostrophic effects may be relevant.

- L18: “…suggest that local convection…” Replace with "shallow local convection" for clarity and to avoid implying deep convection processes.

- L34: “...which may exhibit diameters close to or below this threshold.” The use of "this threshold" is confusing, as the Rossby radius has not been explicitly introduced as a limiting value. Suggest replacing with "this scale" or "this length scale" for clarity.

- L38: “The Amundsen Basin is the deepest part of the Arctic Ocean...”. Consider referencing Fig. 1a here, since this figure displays the bathymetry and general setting described in lines 38–44.

- L46: "...where horizontal velocities are minimal at the eddy core..." This statement is not strictly accurate as written. Velocities are minimal at the centre of the core, not uniformly “at the core.” Please clarify this radial structure.

- L47–49: The manuscript presents the contrasting vertical isopycnal structures of anticyclonic vs. cyclonic IHEs and cites McGillicuddy Jr. (2015). Is there observational support for cyclonic IHEs in the Amundsen / Eurasian Basin, or elsewhere in the Arctic, beyond this reference? If McGillicuddy Jr. (2015) is being used conceptually rather than as direct observational evidence under sea ice, consider softening the phrasing.

- L50: The discussion of prior studies on IHEs/ITEs would benefit from one or two additional key references, both theoretical and observational, to frame IHEs in the broader context of intrathermocline eddies and subsurface lenses. I suggest including reference such as: McWilliams (1985, RGO, doi: 10.1029/RG023i002p00165; 1988, JPO, doi: 10.1175/1520-0485(1988)018<1178:VGTBA>2.0.CO;2); Chaigneau et al. (2011, JGR, doi: 10.1029/2011JC007134); Dilmahamod et al. (2018, JGR, doi: 10.1029/2018JC013828); among others. These help connect Arctic IHEs to the global literature on sub-surface, density-trapped vortices.

- L50–57: The paragraph lists previous studies (Aagaard and Carmack, 1989; Manley and Hunkins, 1985; Timmermans et al. 2008; Zhao et al. 2014; Polyakov et al. 2012; Woodgate et al. 2001) but does not synthesize their key dynamical/structural findings. I strongly recommend adding 1–2 summary sentences describing what those studies actually found in terms of typical radii, vertical extent / core depth, maximum azimuthal velocity, water mass properties (e.g. Canadian vs. Eurasian halocline waters), Rossby number / dynamical regime if available. This would give the reader a quantitative baseline of “what an Arctic IHE usually looks like,” against which the present MOSAiC eddies can be evaluated as similar, distinct, or extreme.

- Figure 1a: The bathymetric colour scale should be inverted, this would align with conventional bathymetric visualisation practices and improve intuitive interpretation; add labels for the main bathymetric and geographic features visible in the panel, such as Lomonosov Ridge, Gakkel Ridge, Fram Strait, and the East Siberian Shelf; include a simple schematic arrow indicating the Transpolar Drift, Artic Ocean Boundary Current, and the entrance of Atlantic Water Boundary Current.

- Figure 1b: Here, the spatial configuration of the L-sites around the CO on 10 October 2019 is shown. It is unclear why this specific date was chosen, since the study period spans 19 October 2019 to 15 March 2020. Was this the configuration at deployment, or does it represent the average geometry during the drift? Please clarify the rationale for selecting this date. If the configuration changed noticeably during the winter drift, consider showing the mean position with a range of variation (e.g., using error bars, semi-transparent ellipses, dashed outlines, or a numerical range such as mean ± interval) to illustrate the spatial evolution of the DN. Relatedly, the DN likely rotated or sheared around its central axis during the drift. Quantifying this rotational or deformation rate (if available) would help readers evaluate whether such motion could affect the quasi-synoptic assumption or the relative sampling geometry used in the eddy detection analysis. As a visual aid, you could optionally include a transparent ellipse indicating the approximate range of motion or orientation of the L-sites around the CO, aligned with the mean propagation axis. This would communicate both spatial uncertainty and the general drift direction in a single glance. Rabe et al. (2024) includes snapshots and maps showing DN behaviour and evolution during MOSAiC. If these aspects are already documented there, please explicitly refer to that study in the main text when describing the DN configuration adding its temporal variability.

- Figure 1c: Indicate on the time series the specific moments when the IHEs were detected (e.g., using coloured markers or shaded bands). This would help the reader relate the eddy occurrences to variations in drift and current speed. Consider adding a new complementary panel (e.g., Fig. 1d) showing the "mean vertical profile of current magnitude", one including all and others separated into periods when the instruments were inside IHEs and when they were outside IHEs. This comparison would visually demonstrate the influence and characteristic velocity enhancement or structure associated with eddy passages. The caption currently refers to “mean current speed averaged from the mixed layer to 100 m,” but not all ADCP records reached 100 m depth. Please revise this depth range to reflect the actual valid measurement interval (e.g., “averaged over the available ADCP depth range” or specify the precise depth range used).

- General for "Data" subsection (end of first paragraph, L70–84): The paragraph effectively introduces the dataset and deployment configuration, but it ends abruptly without describing how the data were processed. It would strengthen transparency and reproducibility to include a brief description of the processing workflow — either as a step-by-step summary (e.g., sampling frequency, sensor type, software used, averaging, interpolation, filtering, de-tiding, etc.) or by citing an established data-processing scheme from previous works (e.g., Krishfield et al., 2008, JTECHO, doi: 10.1175/2008JTECHO587.1; Timmermans et al., 2010, JTECHO, doi: 10.1175/2010JTECHO772.1; Toole et. 2011) This would also clarify whether identical methods were applied to all ITPs and ADCPs or adapted for specific platforms.

- L85–88 and Table 1: The description of the ITP and ADCP sampling configurations in lines 85–88 is difficult to follow, and the same information in Table 1 is not immediately clear without reading the accompanying text. The table header should be improved to make it self-contained. Consider revising the table caption and column headings to explicitly state what each variable represents (e.g., “Profiling frequency [h]” = interval between consecutive profiles, “Vertical resolution [m]” = bin size of the instrument). Include units directly in the header rows and, if possible, a short note indicating the meaning of abbreviations such as AOFB, ITP, and CO-PS. The text in lines 85–88 could also be simplified to describe, in plain terms, that each site had different profiling intervals and vertical ranges, which can be cross-checked directly in the table. This would make Table 1 more intuitive and reduce the need for readers to cross-reference multiple parts of the manuscript to interpret the experimental setup.

- L88–89: The sentence "Given the drift velocity of the sea ice,..., where L3 has the highest horizontal resolution and L2 the lowest" is somewhat confusing as written. The expression “horizontal resolution” may be misleading here, since the authors refer to the horizontal spacing between consecutive profiles, which results from the product of the ice-drift speed and the profiling interval. I suggest rephrasing for clarity, for example: "Because the ITPs drifted with the sea ice, the horizontal separation between consecutive profiles depended on the drift speed and profiling frequency, ranging from about 1 km (at L3) to 10 km (at L2)". In addition, since the drift velocities and profiling intervals of the L-sites are known, it should be possible to compute and visualise the variability of the network geometry throughout the experiment. As mentioned above, these data could be used to support Figure 1b, quantifying how the spatial configuration of the DN evolved over time.

- L90: Replace "ridging"with "ice-ridging" for clarity.

- L92–95: The description of the ADCP configurations and measurement ranges in these lines does not fully match the information presented in Table 1 (e.g., time coverage, vertical range, and resolution). Please revise to ensure complete consistency between the main text and the table.

- L97–103: The assumption of quasi-synoptic conditions is reasonable, yet it would be useful to discuss possible biases introduced by the relative motion between the drifting DN array and the eddies. Specifically, the direction of eddy propagation, the spatial separation between L-sites, and any rotation or deformation of the DN configuration could influence how individual profiles sample the evolving eddy field. These effects might alter the apparent azimuthal velocity structure or introduce Doppler-like distortions in the reconstructed sections. I recommend acknowledging these potential limitations and, if possible, estimating their magnitude (see Allen et al., 2001, Deep-Sea Research, doi: 10.1016/S0967-0637(00)00035-2) or citing supporting work that discusses how sampling geometry and motion relative to a propagating eddy can affect quasi-synoptic interpretations.

- L104–113: The description of how isopycnal displacements were used to identify eddies is somewhat unclear. From the context, the method appears to rely on the temporal evolution of vertical density structure, interpreted as spatial variability because the DN drifted fast enough to cross an IHE during successive profiles. This implies that the identification is primarily visual, based on isopycnal tilting patterns assumed to correspond to eddy structures. The explanation should focus explicitly on how IHEs were recognised in the ITP data. If all detected features are anticyclonic, this should be clearly stated. Additionally, note that in IHEs, the isopycnal slopes are opposite at the upper and lower limits of the eddy core. Making this explicit would help readers visualise the identification criteria more accurately.

- L107–109: It is unclear whether the criterion requiring “at least two profiles within the speed anomaly” refers to two profiles on each side of the maximum isopycnal displacement or one profile on each side. Please clarify this point explicitly.

- L110: Please clarify what this “additional” refers to.

- L114: Replace “defines the radius of maximum velocity” with “is defined by the radius of maximum velocity” to better reflect the physical relationship being described.

- L117: When mentioning “other features such as meanders or fronts,” please include relevant references documenting their presence in the region. For example, studies by von Appen et al. (2018; 2022) describe submesoscale filaments, meanders, and frontal activity in the marginal-ice and central Arctic.

- L117–124: The methodology implicitly assumes that each L-site transected the IHE, yet it is not explained how this crossing was verified. There seems to be a missing step in the identification procedure: before testing for solid-body rotation, one should check for a change in the sign of the azimuthal (or cross-track) velocity, indicating opposite flow directions on either side of the eddy centre. This reversal is a key diagnostic confirming that the profiler indeed crossed through (or near) the eddy core rather than only sampling one flank. Please clarify whether such a criterion was applied.

- L128: The citation appears to be misattributed. The approach referenced here corresponds to Castelão et al. (2013), whereas Castelão & Johns (2011) focused on the North Brazil Current rings that informed this method’s development.

- L135: The text refers to “Equation 1” when describing the computation of azimuthal velocity, but based on the context, it should correspond to Equation (2). Please verify and correct the equation reference to maintain consistency with the definitions given earlier.

- L135–137: A short transition sentence is needed to connect this paragraph with the following subsection. As written, the discussion around the Rankine vortex model appears self-contained, but Figure 3 shows results refined through the Maximum Swirl Velocity (MSV) method explained in the next subsection. Without this clarification, readers may interpret the point of maximum velocity as being chosen arbitrarily for the Rankine fit.

- L147: It is unclear how is obtained at this stage.

- L149: Change “Equation 1” to “Equations (1–3)”, as all three are used in the velocity decomposition.

- L151: Specify which optimisation method was used to minimise the cost function J.

- L155: The variable Rmax mentioned here appears to correspond to the refined radius obtained from the MSV method, which is later denoted as Rm in the text and in Table 2. Please standardise the notation throughout the manuscript to avoid confusion.

- L169: It seems that the reported Coriolis parameter contains both a unit error and a missing exponent in the numerical value.

- L172–173: The statement “This area is distant from major bathymetric features…” is not entirely accurate, since the Amundsen Basin itself is bounded by prominent topographic features such as the Lomonosov and Gakkel Ridges. Please clarify how the study region can nonetheless be approximated as laterally homogeneous and vertically unaffected by topography. Explaining this assumption would help justify that, given the small Rossby deformation radii in this area, such an approximation remains appropriate for the present analysis.

- Figure 2: Include position of CTD and ADCP profiles as in Fig. 6

- L184–185: Please clarify the distinction between the solid-body rotation region (core) and the inner core.

- L187: The phrase “best-sampled eddies” is vague. Specify the criterion used. Including in Table 2 the number of CTD and ADCP profiles used for each eddy (or within each eddy’s core) would make this classification transparent.

- L197–200: It is not clear how the mixed-layer depth (MLD) was determined. Please clarify the method in the Data subsection, alongside the data-processing description and the use of TEOS-10 for thermodynamic variable computation.

- L200–201: The description is somewhat confusing, as the vertical boundaries of the IHEs were defined using local maxima in buoyancy frequency (N). Are these N maxima associated with specific isopycnal surfaces? If so, please indicate whether the corresponding isopycnal densities were consistent among the observed IHEs or varied from case to case.

- Figure 6: Please include absolute salinity (SA) sections and profiles in this figure to complement the conservative temperature and density fields. If adding them here would overcrowd the panel, consider including the SA plots as supplementary material for completeness. Including this data is essential to show how each eddy is embedded in the vertical salinity structure and to visualise the associated halocline displacement.

- L221: The refined radii obtained through the MSV method are denoted as Rm. Please ensure this notation is used consistently throughout the text and figures, replacing any remaining instances of where appropriate.

- L225–226: Consider whether geostrophic currents could be estimated from the thermal-wind balance, at least for the well-resolved IHEs at the L3 site. This would allow comparison with directly measured velocities and provide insight into the relative contribution of horizontal ageostrophic velocity components. It would also be valuable to discuss this in the context of cyclogeostrophic balance, possibly citing Shakespeare (2016, JPO, doi:10.1175/JPO-D-16-0137.1) to strengthen the argument regarding the transitional regime described later. Additionally, consider computing or at least discussing the length-scale Burger number, as this parameter could clarify the dynamical interpretation presented in lines 235–240 and Figure 8.

- L230-231, 237, Figure 7 and 8: Please define explicitly how the hydrographic "core properties" were calculated. 

- L238 and Figure 8c: The claimed linear relationship in Figure 8c is not clearly visible. However, this lack of correlation is still an interesting result, it could reflect the degree to which centripetal forces contribute to the eddy’s dynamical balance or the limited resolution of some eddies due to fewer available profiles. You may consider highlighting the best-sampled eddies (e.g., with different symbols) and performing a linear fit only for those cases, or alternatively, revising the interpretation to acknowledge the absence of a clear linear trend.

- L243: This idea could be expanded further. Consider performing a non-parametric statistical analysis (e.g., clustering or rank-based test) to assess whether the Rossby number (Ro) values support a natural segmentation of the eddy population. Such an analysis would help justify the chosen Ro thresholds shown in Figure 8b and strengthen the interpretation of dynamical regimes.

- L245: Correct the figure reference; the text refers to Figure 8d, not Figure 7d.

- Table 2: Add a brief definition of what "core values" represent. Also specify what stands for Rm in this table and include its units. I assume these correspond to the refined radii, but this should be stated explicitly.

- General (Section 4.1): It is not clear why the analysis of potential duplicate detections focuses only on L1, L3, and CO, while L2 is excluded. Since the text repeatedly notes that duplicate sampling is most likely between L2 and CO, it would be appropriate to explain why these observations were not analysed or illustrated. A brief justification (e.g., data gaps, unreliable velocity records, or spatial separation) would clarify the rationale for this choice and ensure consistency with the earlier statements in the manuscript.

- L256-258: Consider discussing potential rotational effects of the DN during the drift, and please also indicate the expected translational velocity of the IHEs.

- L260: Please include the range of variation for this distance.

- L261: Include the distance between L2 and the CO for completeness and consistency with the other site descriptions.

- L269: The text incorrectly states that L3 detected IHE E4 on 29 October, whereas Figure 9 clearly shows that L3 detected E2, not E4. Please correct this reference. In addition, the description should explicitly refer to the corresponding panels in Figure 9 that illustrate these detections to help readers follow the discussion.

- L270-271: This sentence should be rewritten for clarity. Visually, the two eddies appear well separated, and the azimuthal velocity time series at the CO site confirms this by showing a circulation pattern inconsistent with either IHE. Therefore, the text should clarify how the authors define when a site passes “near the location of a previously detected eddy”. The solid line along the CO trajectory and the dashed lines in the Hovmöller diagram (Figure 9b) are misleading based on the data shown. Removing or redefining these visual cues would help reduce confusion.

- L278-283: The authors conclude that E5 and E6 correspond to the same eddy sampled about one week apart. This interpretation should be developed further, either in the Discussion or as a perspective for future work. Questions naturally arise: To what extent did the eddy evolve during this period? What mechanisms could have influenced or forced such changes (e.g., erosion, intensification, or mixing)? In L280–282, there is also an apparent contradiction. The text first suggests that differences between E5 and E6 arise because different water masses of the eddy were sampled, yet immediately afterward states that Figure 10 shows both structures share the same water-mass properties. Please clarify this point or rephrase to resolve the inconsistency. Finally, note that besides L3, the CO site appears to have crossed these structures—possibly earlier than L3—yet the azimuthal velocity Hovmöller does not clearly display the eddy signal. This aspect could be acknowledged or briefly discussed, as it adds context to the spatial and temporal sampling coverage.

- L290: The text states that the observed IHEs are “saltier and warmer” than those described by Zhao et al. (2014). However, Figure 10b does not clearly show them as warmer—rather, they appear to deviate from the T–S relationship established in Zhao et al. (2014). Please clarify this interpretation: are the eddies indeed warmer in absolute terms, or do they simply depart from the previously reported temperature–salinity trend?

- L290-292: When describing the hydrographic structures, it would be helpful to guide the reader through the specific region of the T–S diagram being discussed—indicating the approximate temperature and salinity ranges where they are located. Additionally, please expand qualitatively on the meaning of the term “wedge shape.”

- L321-322: The statement "given the lack of strong horizontal velocity shear... barotropic instability is unlikely to be a dominant mechanism in this region" is not explicitly supported by the presented observations nor previous literatrure. No quantitative evidence of horizontal velocity shear is shown, and the match between eddy scale and alone is insufficient to exclude barotropic processes. If this conclusion cannot be demonstrated directly, I suggest either removing the sentence or reframing it as a hypothesis or qualitative inference to be tested in future studies.

- Figure 10: I recommend dividing this figure into two separate figures—one containing panels (a) and (b), and another with panels (c–e). This separation would improve readability and allow each set of results (hydrographic vs. dynamical) to be interpreted more clearly.

- L325: Specify that eight IHEs were identified in total, one of which was characterised twice, to make the count explicit and consistent with the preceding description.

- L338: The proposed observational resolution of ∼5 km seems inconsistent with the reported eddy radii (6 km app). If the goal is to resolve the smaller IHEs, a finer spatial resolution would be required.

Thank you for the opportunity to review this work, and looking forward to the revised version,

Kind regards