| 10 Mar 2026
Conservation of Heat in the Coupled Arctic Prediction System (CAPS v1.1): Comprehensive model evaluation based on the MOSAiC observations
Abstract. Atmosphere–sea ice–ocean interactions are vital to understand past and future changes of Arctic climate system. It is essential to ensure the energy closure across model components to investigate these interactions with coupled models at longer timescales. Here, we present the improved version of Coupled Arctic Prediction System (CAPS) with the conservation of heat fluxes exchanged between the atmosphere (WRF) and ocean-sea ice (ROMS-CICE) components. A set of pan-Arctic simulations covering the period of Multidisciplinary drifting Observatory for the Study of Arctic Climate (MOSAiC) was conducted with the improved CAPS. The improved CAPS significantly reduces the inconsistency in heat flux exchange between WRF and ROMS-CICE, leading to better performance in simulating Arctic sea ice conditions compared with satellite observations than its predecessor showing the drifting behaviors. The model capability of CAPS to simulate the atmosphere, ocean, and sea ice conditions in the Arctic climate system were also evaluated based on the comprehensive observations obtained from the MOSAiC drift. The assessments indicate that CAPS reproduces reasonable evolutions of Arctic conditions along the track of MOSAiC observations but accompanies with biases contributed by simulated synoptic storm systems. By applying the spectral nudging technique in the upper atmospheric levels, CAPS can better replicate the observed storm systems and reduce biases shown in the free simulation. The evaluations shown in this article also highlight the key areas for further investigations in CAPS (as well as in other numerical prediction systems) including atmospheric boundary layer processes (surface turbulent heat fluxes) and cloud processes in polar regions.
General comments
The authors present an updated version of the Coupled Arctic Prediction System (CAPS) with a corrected coupling scheme. The model experiments show good agreement between satellite observations, MOSAiC observations, and the spectrally nudged model run. Additional experiments without nudging with the corrected and uncorrected coupling schemes demonstrate the importance of the update. The paper demonstrates the utility of CAPS for Arctic coupled simulations and is an important reminder of the need for consistency across model components. The paper is well written, and I have only a few general comments and line-by-line suggestions.
In the introduction, the authors present broad trends about a selection of sea-ice and upper-ocean related topics. What I don’t see clearly from the introduction is the motivation for the development of CAPS. I would like to see more of an emphasis on what the state-of-the-art is for coupled Arctic climate modeling and forecasting. For example: How is CAPS different from other systems? What need is being filled that wasn’t filled before your updates to CAPS? What aspects of climate physics for the Arctic are uncertain, and how will improving CAPS help improve our knowledge? Not all of these specific questions need to be addressed, however the introduction should do a better job providing a context for the model improvements.
The conclusions and discussion raises discussion of the temperature inversion structure. It seems to me that this topic should be raised first in the results section, along with the other discussion of the atmospheric structure. In Figure 12, the choice of the color bar limits for panels b and c demonstrates substantial improvement between the free and nudged experiments, however it is difficult to see how strong the low bias is in the profiles in the nudged experiments. To my eye it looks like the profiles are 5-10 C too cold through much of the year. Perhaps seasonal average biases plotted as profiles (e.g., like the mean profiles shown in Fig. S6) would help clarify the issue. The representation of the boundary layer is an important point, as I would expect that changing the performance of surface fluxes in the coupler would have a strong effect on this part of the atmosphere.
Line-by-line comments:
25 “showing the drifting behaviors” — It isn’t clear if “drifting” refers to sea ice drift or about the model accumulating errors, please rephrase for clarity
37 “Arctic, a climate system” — the structure of this sentence is awkward
50 “Arctic amplification related to sea ice changes” — perhaps missing a comma?
54 “Besides of the” — the “of” is unnecessary
78-83 This sentence should be broken into at least two sentences — as is, it reads as if you are saying that MOSAiC is the first expedition from 2019-2020, rather than first overall.
83 It is not clear to me that we can call MOSAiC the first when SHEBA exists. In what sense is MOSAiC the first? Largest, certainly.
242 The L-site network and CO fit the 15 km description, but the M sites are further out.
244 “is referred to” — grammar is off here, please rephrase the sentence.
325 ULSIE/OLSIE are only defined in a figure caption — please move the definitions into the main text. As these are less commonly used acronyms, I found myself needing to look them up many times — consider whether there is a more direct way to describe it.
433 “show a throughout snow depth” — rephrase this for clarity
Figure 4: The color scale for panel a and e is not print or colorblind friendly, however the blue-yellow-red color scale is good. Please update the color map for panels a and e. It is difficult to see the vertical and horizontal lines. While I recognize that the simulation domain is large, the study focus is on the central Arctic. Hence, I think that zooming in to the area north of, say, 60 or 70 latitude (as in Fig S9) would be justified, and would enable readers to see the details discussed in the text.
Figure 5: As with fig 4, look for a color scale which is perceptually uniform. See the instructions at the Copernicus guide for authors: https://publications.copernicus.org/for_authors/manuscript_preparation.html#figurestables
As with Figure 4, Figure S3 is difficult to see. You may consider changing the figure from a 2x6 plot to 4x3, as there is a lot of extra unused white space on the page.