This study explores the sensitivity of model representation of atmospheric cloud radiative heating profiles over the North Atlantic to changes in grid resolution, atmospheric convection (explicit or parameterized), and microphysics scheme within the ICON model.  While grid resolution is only found to play a small role, cloud radiative heating profiles are highly sensitive to the model representation of convection and microphysics.  In particular, the role of cloud ice mass mixing ratio appears to play a critical role.

This manuscript provides a good discussion of the factors governing atmospheric cloud radiative heating profiles at midlatitudes in the ICON model.  Technically, the manuscript is sound and just needs some minor corrections/clarifications (detailed below).  However, my general impression in reading this paper is that I’m not sure GMD is really the right journal for this work.  The manuscript is using the model output data set from a previous study (Senf et al. 2020) and really not describing fundamentally new methods, rather just extending the authors’ previous work from the tropics to the midlatitudes.   I'll leave it to the editor to decide whether GMD is the appropriate venue for this work.

Lines 9-10:  Isn’t this point (coupling of microphysics and convection schemes) just a hypothesis provided at the end of the paper (Line 352-360)?  If so, it doesn’t belong in the abstract as a statement of certainty.  I don’t see any formal evidence presented to support this conjecture.

Lines 23-25:  Lu et al. (2007) do not discuss cloud-radiative impacts, and models do not agree on whether the presence of cloud radiative effects drive a poleward circulation shift (see discussion in Voigt et al. 2020 review).  For example, Li et al. (2015) do not find a poleward expansion of the circulation due to the presence of cloud radiative effects, and they actually show that cloud-radiative effects decrease the static stability in the tropics.

Line 32: The intensification of ENSO due to cloud radiative effects is again a model dependent result.  Middlemas et al. (2019) found a differing effect on ENSO.

Lines 136-137:  More detail probably needs to be provided here to explain this conclusion, as the numbers in Table S1 do in fact look quite sensitive to the particular thresholds used.

Lines 146-155:  It also seems important to note/discuss here that the altitude of the lower and upper tropospheric cooling peaks differs fairly significantly by model.

Lines 169-170: Also convective heating rates appear to be important in this layer

Lines 172-173: Also, the cooling peak appears to be slightly higher in altitude in the one-moment scheme

Line 180, typo: Change “Then” to “The”

Lines 184-185:  Also, a large heating peak develops at lower altitudes, which is not present in the simulations with the deep convective parameterization

Lines 221-227: Good to double check the percentage values quoted in this paragraph.  They appear to match what is shown in Fig. S3, not Fig. 7.

Lines 230-232: Can you provide a physical explanation for why the isolated high clouds warm and the deeper clouds cool?

Line 243 (and hereafter):  The term “higher grid spacing” could be confusing and could imply coarser resolution to some readers.  I would either say “higher resolution” or “finer grid spacing”.

Line 269:  It doesn’t look like a factor of four.  At best, it looks like a factor of two.

Line 271: The relative increase actually appears stronger in the thin cloud layers.

Lines 317, 325:  This citation structure is confusing.  Initially, I was looking for Fig. 10a and Table 2 in this paper.  Please clarify that this figure and table are in the Sullivan et al. (2022) paper, and not this paper.

Line 328:  I think you need to elaborate more on why you choose “supersaturation generated by vertical velocity” as one of your cloud controlling factors.  The other two are obvious from the above equations, but this one is less obvious.

Line 384, typo: boundary

Figure 1 caption: North Africa, as well

Code and data availability: available is misspelled.

References:

Li, Y., Thompson, D. W. J., & Bony, S. (2015). The influence of atmospheric cloud radiative effects on the large-scale atmospheric circulation.

Journal of Climate, 8, 7263–7278. https://doi.org/10.1175/JCLI-D-14-00825.1

Middlemas, E. A., Clement, A. C., Medeiros, B., & Kirtman, B. (2019). Cloud radiative feedbacks and El Niño–southern oscillation. Journal

of Climate, 32(15), 4661–4680. https://doi.org/10.1175/JCLI-D-18-0842.1

Sullivan et al. (2023) analyze cloud radiative heating (CRH) in the upper troposphere over the North Atlantic Ocean as simulated by the ICON model. They provide a comprehensive analysis of dependencies on model resolution and ice-cloud microphysics parameterizations. They also attribute model differences to different classes of clouds with different vertical structure and to different cloud-controlling environmental factors, which I found useful for understanding the model differences. The paper is well written and logically organized, and the figures are clear. I have one suggestion for generalizing the results and several minor comments. I recommend that the manuscript be accepted if these minor issues are addressed.

General Comments:

The authors comprehensively discuss CRH in the upper troposphere, but they barely mention CRH in the lower troposphere. However, previous studies have shown that the CRH climatology and relationship of CRH to prominent modes of natural variability both have a peak magnitude in the lower troposphere (Haynes et al., 2013; Papavasileiou et al. 2020; Wall et al., 2022). I realize that the authors want to focus on ice microphysics, for which it makes sense to analyze upper tropospheric CRH. However, I think neglecting lower-tropospheric CRH from the discussion gives the (unintentionally) misleading impression that it is less important for the overall CRH throughout the troposphere. Could the authors add a section that analyzes CRH for the low-level cloud classes and some discussion that compares this analysis with the CRH for the upper-level cloud classes? I think this would generalize the findings and add value to the paper.

Specific Comments:

(1) Fig. 12: It would help to plot all of the panels on the bottom row with the same range of vertical-velocity values.

(2) Fig. 13: Has the vertical velocity data been horizontally averaged to a common scale prior to computing these histograms? If not, then is such a comparison meaningful? I would expect finer horizontal resolution to have larger variance of resolved vertical velocity but perhaps a different treatment of unresolved subgrid-scale vertical velocity by the convective parameterization scheme. It might help to discuss this around line 350.

(3) Line 376 states “Strong microphysical and convective sensitivity and weaker grid spacing sensitivity in the CRH profiles do not appear in distributions of cloud class occurrence and appear only weakly in cloud fraction profiles.” Is this result a consequence of the fact that midlatitude synoptic weather systems are mostly resolved by all resolutions in the study? I’m just wondering if this result is specific to the midlatitudes or if it might generalize to the tropics. It would help to clarify this around line 376.

References:

(1) Haynes, J. M., Vonder Haar, T. H., L’Ecuyer, T., & Henderson, D. (2013). Radiative heating characteristics of Earth’s cloudy atmosphere from vertically resolved active sensors. Geophysical Research Letters, 40(3), 624–630. https://doi.org/10.1002/grl.50145

(2) Papavasileiou, G., Voigt, A., & Knippertz, P. (2020). The role of observed cloud-radiative anomalies for the dynamics of the North Atlantic Oscillation on synoptic time-scales. Quarterly Journal of the Royal Meteorological Society, 146(729), 1822–1841. https://doi.org/10.1002/qj.3768

(3) Wall, C. J., Lutsko, N. J., & Vishny, D. N. (2022). Revisiting cloud radiative heating and the Southern Annular Mode. Geophysical Research Letters, 49, e2022GL100463. https://doi.org/10.1029/2022GL100463