|The manuscript is sufficiently improved that I can follow what the authors are trying to do, and attempt a review. |
Overall, within the practical constraints of the state-of-the-art, I think the work is reasonable enough, although some of the sources of uncertainty need better highlighting. I have a few comments of varying degrees of potential importance.
"Ceteris paribus" - the language of the journal is English and superfluous Latin should, in my opinion, be avoided. Not all readers have a first language that has a Latin base. I Googled the meaning, and do not understand why you don't use English here.
"Repeat" not "recapitulate"
I suspect that the 0.64 is highly uncertain and state dependent. It would be good if you gave a rough estimate of your uncertainty for this value.
On p4, you state that, to convert from S_[LI,CO2] to S^a you should multiply by 0.64. Therefore DT_G on in eqn 10 should also be multiplied by 0.64. This may solve all your problems in terms of reconciling CLIMBER and the data as, if I understand correctly, you need to include this multiplicative factor for the data based analyses, but not for the CLIMBER analysis.
"Any specific moment", "in time" ... I do not think this is what you mean. Equation 1 relates to the steady state! This also causes me to ask what time averaging is used for the CLIMBER analysis on p7-8...? (Maybe this is already stated somewhere but I do not recall seeing this mentioned in the manuscript...?)
I don't think this is really a "validation" as such - more an "illustration" perhaps? Clearly CLIMBER is very simplified and other models may have quite different result. A sentence explaining why this information is not forthcoming from GCMs would be appropriate (ie you are forced to use a simplified model because of the CPU overhead).
P7. "Note that, in the case S... is equal to... The fit further shows decreasing..." This seems silly, and confusing, especially since you then use it as a reason for referring in a confusing way to S_CO2,LI later on. Just call it S^E_[CO2]. It is surely clear to anyone who has managed to get this far through the paper what is meant.
Fig 3b looks to be less linear not more linear than 3a
Again we have an "any moment in time". Rephrase please.
The section "In principal [which should be "principle", but these days we rely on copy-editing to pick up this kind of thing] ... (Fig3c)." Should be moved to p7 where the LGM is first mentioned in this context.
Please remove all the "old approach" / "new approach" labels. Give the approaches names, or reference the equations, or other papers. Otherwise people do not know what you are talking about, and they will be even more lost in the future when there is a new new approach and your approach becomes the old approach!
ANICE is used to convert from temperature to radiation. After seeing some of the ice model inter-comparison results I suspect this is a major source of uncertainty. If you agree, it would be good to see this at least mentioned.
"Therefore, these results...similar to the of proxy data..."
I do not know what you mean. Why?
P10 and throughout. The uncertainty ranges are quite strange because they exclude an awful lot of known uncertainty. Maybe you could be more careful about stating in the Abstract and Conclusion what these ranges include and do not include in terms of known uncertainties. Also, what are the quoted ranges - 1 standard deviation? 2,3? 95%??