Review of “Estimation of the radiation budget during MOSAiC based on ground-based and satellite remote sensing observations” by Barrientos-Velasco et al.

A dearth of comprehensive in situ observations of the radiative energy budget over Arctic sea ice has long impeded our understanding of the atmosphere/cloud-ocean-sea ice processes. This manuscript seeks to address this gap though analysis of the year-long MOSAiC expedition data combined with CERES satellite observations of the in situ collection area.

It is clear that the manuscript represents a very substantial body of work carried out during the measurement and analysis phases. The amount and variety of data treated in the manuscript is very impressive, and the methods are well-referenced and appear to produce very logical results. However, the sheer scale of the treated material combined with some presentation choices does make the manuscript quite cumbersome at times. Thus, my main comments and improvement recommendations mainly deal with presentation rather than the analysis, which seems sound. I trust that, with revisions, the manuscript shall become a valuable contribution to the body of literature dealing with the Arctic radiative energy budget and its measurements.

Main comments:

1. The algorithm components are clearly described, but the overall scheme not quite so clearly. I gathered from section 3 that the MOSAiC measurement data is first used as input for the ShupeTurner cloud retrievals, which are in turn used as input for TCARS, correct? A simple flow chart could bring clarity here, though the number of display elements in the manuscript is already very large. Please consider how to make the processing flow fully unambiguous to the reader.
2. I may have missed this, but it felt that the uncertainty of the derived flux components was not very clearly discussed. If my assumption of the processing chain in previous point is correct, what is the sensitivity of the flux components to original MOSAiC measurement uncertainties plus whatever the ShupeTurner + TCARS add on top? I was glad to see unambiguous targets for closure of the downwelling components, but the Lanconelli reference is old and thus would not correspond directly with the observational uncertainty, right? And what were the closure targets for the upwelling fluxes?
3. Section 4 was, for me, challenging to unravel. Besides the sheer length and scope of treated material, there were items I suggest the authors pay attention to:
   1. Subsectioning and reuse of tabular content: The present text, particularly in 4.3 and 4.5 is composed of lots and lots of numbers in watts per sq.m with rather invisible movements from one radiative energy budget component to the next in the text. I would advise considering the use of additional subsectioning to separate flux components for clarity, and a careful review of where (and how) it is really necessary to reiterate table contents – particularly if the reiteration is not placed into further context, as is the case towards the end of 4.5 (lines 649-660 in particular).
   2. Many descriptions of display elements are to me unnecessarily wordy, often repeating large parts of caption content. This sometimes extended to other parts of the text, e.g. conclusion 2’s first half seemed like a very long way to say that the results of Rabe et al. (2024) on the spatial continuity of the longwave flux were confirmed here. Please examine where fewer words could deliver the same message.
4. In 4.3 (lines ~490 or so), the message implied from the numbers seems to be that the radiative effects of sea ice albedo changes at TOA are roughly halved by cloud presence, which is the number we’re seeing from a variety of other data too. This is a point which could use a bit of spotlight, I suggest you reference e.g. Sledd and L’Ecuyer (2019) on this.

Minor comments:

• Terminology: If the authors prefer “solar” and “terrestrial” in stead of “shortwave” and “longwave”, sure (even if “downwelling terrestrial” sounds a bit curious for me), but in any case please make the equivalence fully clear early on for those of us used to the latter option.
• Figures: Most figures in the review version seemed unnecessarily small, affecting readability somewhat. For figures 4-6, I would like to see the closure acceptance limits for the fluxes displayed as colored vertical lines or shaded regions to avoid the need for checking in the text.
• Daytime vs . daily mean albedo (lines 221-225 and Figure B3): Just to confirm since you are comparing with CERES – using the daily mean flux ratio avoids solar elevation screening, but is the resulting number then same as the daily mean albedo used in CERES? During the melting season the sea ice floe’s albedo can have a notable non-symmetric diurnal cycle, is the orbital sampling behind CERES albedo such that it's assuredly an apples to apples comparison, given the varied footprints too?

Assessing the radiation budget in the Arctic and the impact of clouds is still challenging and often limited due to missing simultaneous detailed observations of radiation, thermodynamic, and cloud properties. The comprehensive observations of the MOSAiC expedition provide thus a great opportunity to quantify the solar, terrestrial, and net fluxes for the Central Arctic over a complete year and to estimate the cloud radiative effect (CRE). With this work, the authors thus add another important puzzle piece to the Arctic radiation budget and the corresponding cloud impact. In order to estimate the radiation budget and the CRE, different data sets/products and methods are used, i.e., on the one hand, a column (1D) radiative transfer simulation using the TCARS setup and on the other hand, the satellite-based CERES SYN product. The results are presented in four main sections with 4.1 presenting an overview of the atmospheric and surface conditions, 4.2 summarizing the results of the cloudless simulations with TCARS and CERES SYN, 4.3 presenting radiative closure studies, 4.4 evaluation of the net terrestrial flux and 4.5 presenting the radiation budget and CRE during MOSAiC.

Major comments:

The paper is generally clearly written, and the methodology is sound. My major concern is the length of the manuscript, particularly the extensive sections 4.2 and 4.3. The number of figures, tables, and numbers provided is simply overwhelming, i.e., 10 figs. + 12 figs. in the appendix, 6 tables and 3 tables in the appendix. Just moving information to the appendix does not solve this problem. I think the discussion of the yearly cycle of the net terrestrial flux (4.4), the radiation budget, and the CRE (4.5) are the most interesting sections. However, with the detailed and lengthy comparisons in the sections before, it is hard to follow and keep the reader’s attention (which is really a pity since the results of 4.4 and 4.5 are really a highlight).

I recommend reorganizing the manuscript and drastically shortening/synthesizing the content. I have some recommendations but I would also leave it to the authors to decide which parts to shorten/remove.

1) It is unclear to me why you need 3 different simulation setups with 3 different surface albedo inputs. Eventually, you anyhow focus on TCARSe2. Of course, taking into account the impact of the spatial variability (of the surface albedo) is of interest, particularly since you also use a satellite product with a larger footprint. However, for the reader, it would have been much easier to follow if you had a dedicated section on this (e.g., “Impact of spatial variability”) and otherwise used one TCARS product only (which you think is best suited for your analysis).

2) Do you really need section 4.2 “Consistency of cloudless simulations”? 4.3 deals with the radiative closure assessment including also dedicated cloud-less comparisons. Any findings from 4.2 might be perhaps included in the discussion in 4.3. So I would remove this section completely.

3) For me, the separation of 4.4, particularly Fig. 6, from 4.5 (Fig.7) is unclear. I suggest having one dedicated section about the yearly cycle of the (net) fluxes (including Fig. 6-8) and one section on the yearly cycle of the CRE (including Figs. 9+10).

In my opinion, it makes sense to first present Terr-N SFC in Fig.7a and subsequently zoom in and discuss the monthly distributions of Terr-N (Fig. 6), since the bimodality can not be seen in the monthly boxplots. For me, this would be a more natural way to follow.

Specific comments:

line 36: Reference “D. and Rex”. Please check.

line 37: “Satellite observations”: Do you refer to particular ones? Of clouds?

line 42: “Additionally, Hartmann and Ceppi…”: What is the logical connection between this sentence (trends in radiation) and the one before (comparing ground-based/satellite cloud observations)? Please rewrite/motivate…

lines 48 ff: Can you also comment on the uncertainties in the trends of these studies in particular of the trends in clouds?

lines 99-100: Ebell et al. (2022) refers to the HATPRO measurements, not MiRAC-P. To be corrected in Table 1, too.

lines 122-123: verb missing

lines 149-150: “which is defined for the calculations as 20 km”. What do you mean?

lines 152-153: Are cloud properties in the CERES SYN data set provided for each cloud layer? Can you comment here on the vertical cloud profile information?

lines 172-174: So, what was the result of the closure analysis? Did the ST2015 product perform better than the ARM Microbase cloud product?

lines 179-180: The sentence is odd. Please rewrite.

line 197-198: “TCARS uses various sources of input data such as …aerosols..”: Please mention already in this section that aerosol data are actually not included in the TCARS simulations of this study. It is mentioned later, but it would be helpful here.

line 209: “driplet” should be droplet

line 253: “which subtracts the observed radiative flux from the cloudless simulation”: this should be the other way around, i.e., which subtracts the radiative flux of the cloudless simulation from the observed radiative flux

line 256: “The atmospheric CRE…”: You introduce the atmospheric CRE but don’t show any results. Why?

lines 267-268: “…into four periods:….”: but you explicitely mention only two (Oct 15-Mar13 and Mar 14-Sep20). Can you mention all four?

lines 282-283: “The stratospheric temperature also dropped below 200 K (Fig. C1).”: Fig C1 does not depict the stratospheric temperature. Please explain.

Fig. 3: Can you introduce/present Fig.3 in section 4.1? It is never presented in detail in the manuscript but it would make sense to do so in the “overview of atmospheric conditions” part.

Fig. 3: So in ShupeTurner, the effective radius of liquid cloud droplets is a fixed value, also in the vertical, right? Could be mentioned once more when presenting the reults.

line 413: Should rather be “(Fig 4d, 4l)” only since you focus on Terr-D at this stage.

line 421: Should be Fig. 4 not C5

lines 426 ff: why do you use the surface albedo from CERES at all since you later also point out that it is underestimated?

lines 438 ff: LWP and IWP are not the only values impacting the atmospheric opacity. What about the effective radii? This needs to be discussed jointly.

lines 466 ff: You analyzed hourly mean values for different single-layer cloud types. How many cases do you have for each class in the end? What if you have different single-layer cloud types within one hour? I assume that this is actually quite often the case.

lines 462-463: “For ice clouds, there is a positive bias for TCARS of about 20 Wm-2, suggesting an overestimation in cloud opacity,…”: This should be an underestimation of cloud opacity.

line 467: “as they absorb Sol-D less effectively.” Just to be precise, the clouds do not absorb, but the atmospheric gases in the cloud layer.

lines 467-477 and ff: To be sure that I understood it correctly: For the Terr comparison, differences only occur because different CERES SYN columns are used due to the different ground stations considered, right? The TCARS Terr simulations are the same for each location, right? And for the solar part, the TCARS simulations differ also because of the different surface albedos that are used for the different stations, correct?

lines 499 ff: This is about the net SURFACE terrestrial radiative flux. At least I think so…Please add this information.

lines 505-506: The sentence is redundant. “we analyse opacity by examining the net terrestrial flux as this variable (i.e. the net terrestrial flux) is related to …opacity in the terrestrial spectral range”.

lines 509-511: This could be mentioned in the introduction of this section. See my questions before.

line 252: “data limitation was based on”. Please rephrase.

line 531: “showing relatively similar distributions”, referring to what exactly?

line 586: “Figures 7c and 7b” should be 7c and 7d

line 587: “With an underestimation of surface albedo by -21.01 %” Why do you use the CERES SYN surface albedo at all in TCARS then?

lines 606-611: Regarding the higher values of Terr-N TOA for CERES SYN: are these primarily due to the lower/underestimated surface albedo values? Or is this also an effect of how clouds are represented in the data set?

line 609: “indicating that less solar radiation is absorbed…in TCARS simulations”. This should be “more solar radiation is absorbed” since the TCARS value is smaller than the CERES SYN value

line 612: “At the TOA, the radiation budget…” Since you analyze solar and terrestrial fluxes (down, up, net) and total (solar + terrestrial) fluxes in this paper, the reader can easily mix up the different components. Please check throughout the manuscript that you always use a clear naming. Sometimes, you can deduce from the context what kind of flux is meant but I would try to be as clear as possible. Here, the total net radiation budget (sum of net solar and net terrestrial) is discussed.

lins 661-662: The accuracy of the cloud micro- and macrophysical products is indirectly evaluated in terms of radiative closure studies. Maybe you can add this here.

line 693: “surface” Net-Terr flux?

line 710: “net/total” CRE?

Fig.4 : Can you add x-ticks in all subplots?

Fig.4: I would still expect one baseline TCARS setup to be shown. I find using a mixture of e1 and e2 confusing. I would simply use e2 and say the the Terr simulations are extended to cases when solar radiation calculations are not possible since the Terr flux calculations in e1 and e2 anyhow do not differ (apart from the time period being covered).

Fig. 6 This is for the surface, right?

Fig. 7 Please specify the fluxes: “net” terrestrial and “net” solar flux at the surface, net radiation budget at the surface: maybe you can introduce all terms when you introduce the CRE. Also, be consistent: on the y-axis (e) it says “Total SFC”. I know what you mean, but for clarity, just use one dedicated term for each variable throughout the manuscript.

Fig A1. Remove time periods not covered by MOSAiC for clarity.