Please see the attached PDF.

The manuscripts presents a detailed Large Eddy Simulation study where a stratus onset, maintenance ad dissipation case in Southern West Africa is designed based on observations collected during the DACCIWA campaign. After describing the reference case, a total of 6 sensitivity studies are carried out regarding the aerosol loads, and also on the specific semi-direct effect that these exert on low clouds. Largest differences are found in the CLEAN experiment. The manuscript shows interesting results and spends enough time describing the reference case as well, something that is appreciated. The match of their reference case to the observations is appreciated, since it is not always an easy task in LES studies. The manuscript is also well structured and follows a logical order in the results. However, I find some issues that need to be addresses before this mansucript can be accepted.

MAJOR issues:

-place results in context with previous studies, both when describing the reference case as well as the aerosol sensitivity studies. 

-Language-related issues, like grammatical errors and long and complex sentences make following the manuscript a difficult task. Please correct all grammatical mistakes (only some examples given along the line by line comments) and keep sentences short for the shake of clarity.

MINOR issues:

-Please clarify how exactly the tendency profiles are obtained from the radiosoundings. As it is now, I understand that the tendency applied to the LES domain over each hour is equivalent to the difference in T (and q) between two consecutive radiosoundings divided by the time passed between these. My current understanding is that such tendencies are used as proxies for large scale advection of moisture and temperature. If this is the case, I am afraid such tendencies may include not only the evolution of temperature and humidity due to large scale advection, but also the tendencies due to local thermodynamic effects such as radiative cloud top cooling or warming/moistening (after sunrise) due to surface fluxes. 

-Section 3 goes through the results of the REF simulation in high detail. Readers would find it easier to understand, however, if instead of a description of each result, a more concise section with the most relevant results is presented. This would also allow, as suggested in the fist Major issue, some room to link the relevant results to previous studies.

Line by line comments:

L44: Please introduce briefly the direct, semi-direct and indirect effects of aerosols, given they are recurrently mentioned along the manuscript.

L67: remove dot

L87: what effect?

L135-136: I see surface fluxes first, and sensible and latent heat fluxes later on. If it is referring to same measurements please delete one of the references. If they are different measurements, please clarify.

-141 analyzer instead of analyzed

L146; What is the approx spatiotemporal resolution of these measurements?

L164: On top of the surface heating, I would expect also a weaker cloud top cooling due to solar radiation being absorbed at cloud top. If this is the case, please mention it. And if it is not, please explain why.

L186 remove comma after scales

L187 transport

L190 centered

L204: remove ‘completing LIMA’.

L242: Stratus clouds are known to be very sensitive to the vertical resolution near cloud top (Stevens et al 2005). It would interesting to learn a bit more on the sensitivity of this case to the vertical level spacing (if previous numerical experiments with different vertical spacing were performed), and why 10m was decided eventually as the vertical spacing for the lower part of the domain.

L255: Please explain further what is done regarding the tendencies for horizontal wind and the presence or not of a Nocturnal Low Level jet (since it was mentioned in L150 as being closely related to cloud formation). 

-L256 Linked to minor issue n1. It would also be of interest to show the temperature and humidity profile obtained by the radiosoundings in (at least some) of the profiles in figures 7, 8 ,and 9. 

Fig4. Adding the 4 phases introduced earlier and described below (stable, jet, stratus, convective) below the time axis would help the reader to locate the phases in this specific case and come back to it when needed along the manuscript.

L277: Please add an indicative UTC time for the onset of the convective phase to better guide the reader.

L318 ‘But, as the LLSCs…’ It is difficult to understand the meaning of this sentence. Please rephrase.

L319 Any thoughts as to why the difference is reduced during the convective phase? 

L372: I would substitute ‘at the cloud’ top by ‘above the cloud top’. In fact, the strong longwave emission is a source for turbulence in the cloud layer as parcels near cloud top cool and sink along the cloud layer.

L376: I am confused as the cloud layer is now called to be ‘very stable’, while one line above it was said to be well mixed. 

L381 I am again confused by the use of ‘stable’ when the plot shows almost near-constant equivalent potential temperature.

L415: I find the concept of breakup confusing. In this line the breakup is said to happen at 16 00, but the lower Figures in Fig 5 at 16 00 suggest that, if the breakup is defined as the first moment with LWP=0 somewhere in the domain, such breakup happenned earlier (even at 12 00, one could argue looking at FIgs 5 c,d). 

L401 It is difficult to understand this sentence, please rephrase.

L417: Now 16 00 seems to be the end of the breakup. Please clearly state how the breakup is defined and keep it consistent across the manuscript.

L420: I am confused as to at what altitude I should look for a 1.25 m2 s-2 TKE in in Fig 9.Please clarify.

L421. Also in L387 vertical windspeed was assumed to be the driver fro TKE changes. I’d suggest therefore adding the profiles of windspeed, even in an appendix given the already full panel in Figures 7, 8, and 9.

L447: dividing, not deriving. 

L465: It is very challenging to understand this sentence. Please rephrase and divide into shorter and clearer sentences.

L494: I don’t think ‘inverse layer’ is the right term.

L495 increase instead of exceed

L505 I cannot follow the last part of the sentence.

L511 The explanation is interesting. To validate such hypothesis, it would not take too much effort to compute some metric for spatial SWRADSURF variability. If it turns out to be larger in POL and REF than in CLEAN, then the hypothesis of cloud-holes increasing the shortwave radiation reaching the surface would be reinforced. Further analysis of the cloud layer could also help, since all variables are present in a LES simulation.

L529 I would not call dispersion to cloud top and cloud base.

L549: I dont see how Fig12c contributes to the sentence. Please clarify.



Stevens, B., Moeng, C.-H., Ackerman, A. S., Bretherton, C. S., Chlond, A., de Roode, S., Edwards, J., Golaz, J.-C., Jiang, H., Khairoutdinov, M., Kirkpatrick, M. P., Lewellen, D. C., Lock, A., Müller, F., Stevens, D. E., Whelan, E., and Zhu, P.: Eval-

uation of Large-Eddy Simulations via Observations of Nocturnal Marine Stratocumulus, Mon. Weather Rev., 133, 1443–1462, https://doi.org/10.1175/MWR2930.1, 2005.