|The authors made major changes to this manuscript, focusing it and dropping some of the analysis (i.e Volcanic simulations) as requested by the reviewers. This is a much better presentation. I am not convinced that the authors have fully responded to the reviewers and concerns about the type of model used for this study, but this paper is getting there. |
I tried to look at the whole manuscript again, since the focus changed. And it STILL needs some substantial revisions.
The paper is better. I think the abstract still feels disjointed. The paper does not know whether it is a modeling study or an observation study. The problem is that the ‘observations’ rest on a reanalysis model, and the key ‘microphysical’ parameter is just a passive tracer of human emissions in that model. Again, it feels like something slapped together from a few projects. But it's one less project now (only 2), so that's fine.
I think it needs better organization of the results and separation between observations and models. Significantly, the manuscript should decide the order of presenting results. E.g.: section 3.1 is called:
3.1 Observed relations between meteorology, mid-latitude cyclones, and aerosol
But then immediately:
3.1.1 Aquaplanet simulated mid-latitude cyclones and their response to changes in CCN
Put this in a more logical order please. It should perhaps flow from observations to model simulations. The figures could be kept in a similar way, or separated, but I don't think the present treatment works.
Also: with the narrower focus, there should not be a need for supplementary material, and this should be brought forward.
A few broad notes:
1. I think what I fundamentally have a problem with is applying sophisticated tools like a high resolution model in coarse ways (aquaplanet, specified and simplified aerosols and microphysics). And then saying that is what climate change will look like. I think this is valid for the WCB analysis and that partitioning is fine and interesting, but I think it is difficult to then compare that to the real world.
2. Perhaps it needs to be stated that most of what is going on in mid-latitude cyclones is resolved at the scale of the model. That's not really stated, and would probably be scientifically defensible. But it does NOT look that way from the model results: convection matters, because the answers change.
From Figure 2 and Figure 6: the MERRA2 and LR results look similar, except the Low and High CDNC have opposite effects. The high resolution aquaplanet does look like observations, but this is fundamentally different than low resolution.
3. Trying to pull parts of a reanalysis (like using sulfate as a proxy for CDNC) is dangerous. Not only are uncertainties in the data swept under the rug, but the reanalysis is also then being pulled in different directions. I know this has been published before and ‘it works’, but I think that really brushes away a lot of processes.
4. The WCB part is useful, except to the extent that ice and ice indirect effects is not really treated. But I think trying to make inferences then about the real atmosphere from the coarse analysis of the MERRA data is very dangerous.
5. The analysis takes too 'high level’ a view and does not dive deep enough into what is going on in the model. What is going on to make the LR and HR simulations respond differently? You imply it is convection. But you could show that in the model, by figuring out where convection is in the cyclones, and relating that to the changes in Figure 6.
Some further detailed notes:
P5, L15, “MERRA2 sulfate mass is a good predictor of CDNC as observed by MODIS”. You say no cross talk: but if aerosols and clouds are convolved in MODIS, then there is cross talk because MODIS aerosol is assimilated to get SO4?
Figure 2: This is a mess, because the contour intervals are all different. If you make them the same, I think you are going to end up with some wildly different results. I think that makes the analysis and comparisons very difficult.
Figure 3 also seems hard to compare: Aquaplanet, MAC-LWP and MERRA2 seem very different.
P12, L17: the problem is that sulfate mass concentration in MERRA may not interact appropriately with clouds (e.g., incorporation in cloud drops, scavenging, etc) if there are no indirect effects.
P15, L10: The albedo discussion here is not convincing. The model (Figure 9) looks quite different than the other analysis in Figure 8.
P15, L20: Im confused by this paragraph, and concerned at the method. The 2nd sentence on line 23 is missing a critical clause. Also: here is another place where you have mixed up observations and simulations in the flow. Perhaps this should be with observations.
P16, L20: I'm not convinced a regression model works here based on the previous analysis and the different behavior with MERRA.
P18, L7: This first conclusion: you come up with the same results of many models that do not match observations (e.g. Mallevele et al 2017). Why should these results with bulk microphysics be any more valid? The processes in your high resolution simulations are the same for stratiform cloud microphysics as in many global models. Why would these simulations not have the same issues?
P18, L18: Please comment on the limits of the regression model.