This paper presents an improved representation of acetone in GISS ModelE2.1 and evaluates it with observations. I don’t understand why this is submitted to GMD considering that there is no originality or particular difficulty on the model development front – the approach replicates what has routinely been used by other models to describe acetone sources and sinks, without any tangible improvement. From a model development standpoint the implementation is trivial. I could see the value of documenting in GMD a major update to GISS ModelE chemistry affecting the simulation of species relevant to chemistry-climate interactions, but acetone is not important enough to rise to that standard, and the paper does not discuss how improved simulation of acetone affects the model’s chemistry – presumably not much.

In my opinion, this paper is not a significant contribution to model development and is well below the standard of papers published in GMD. Publishing it might actually do some harm because it might be perceived as a new take on the global budget of acetone, which has not been revisited for some years, but in fact there is no innovation here aside from the evaluation with ATom measurements, and anthropogenic emission inventories are for 2000 which is dated. The evaluation with ATom is in my view the most interesting part of the paper but the interpretation is cursory.

A few specific comments:

1. Line 39: acetone is not highly water-soluble by atmospheric standards.
2. Line 40: oxidation of acetone by OH is not a net source of radicals.
3. Line 170: what spectroscopic and quantum yield data are used for acetone photolysis? There has been some work done on that recently.
4. Line 180: Where does the fixed ocean concentration of 15 nM come from? What is the justification for assuming a fixed concentration? This obviously effects the sign and geographical distribution of the air-sea acetone fluxes mentioned in the abstract.
5. Lines 228-229, 257-258: documenting the improvement over the previous GISS ModelE parameterization of acetone is of little interest considering that the previous parameterization was so crude by current model standards.
6. Line 325: why would there be non-linearities in the system? I presume that the acetone simulations use the full chemistry mechanism (although that’s not clear – it could also be done with archived OH fields and production rates) but since acetone is in general a minor player in oxidant chemistry I don’t see why there would be significant non-linearities. The asymmetry in response to doubling/halving is not necessarily a sign of non-linearity in the presence of other sources/sinks. If there is indeed significant non-linearity the authors should explain why.
7. Lines 383-384: analysis of the evaluation with ATom observation is limited to uninstructive throwaway statements. The authors should do better. Would correlations with other chemical species be insightful?

Acetone is an important molecule in troposphere chemistry cycles and to model it correctly a rather sophisticated scheme is needed because of the complex sources and sinks including biological, chemical and physical processes. Also to note is that acetone has been well studied but recent widespread measurements provide an opportunity to further refine models.

Significant effort was put into the work described in this paper and the authors made good progress toward improving their GISS model representation of acetone. If this is sufficient for publication in this journal then fine and I can provide a more thorough review. However if, to be published in this journal, (which I am not very familiar with) there is a requirement that the paper provides an important contribution to our understanding of acetone in the atmosphere then it falls short because, although significant progress was made on this model, the work does not represent an improvement over previous modeling efforts. If this is the case I would recommend rejecting the paper. The editor will have to make the call on this.

There are some things in the paper that cause concern on my part. One is that the derived chemical lifetime from the expression (burden (Tg) /sink (Tg/year)) is too short. (Also, the burden units are wrong on line 203).  The kinetics of the reaction of acetone with OH is well known and the authors have the correct expressions on line 169.  Based on the OH reaction, if one assumes that the diurnal [OH] is 0.7E * 05, then the chemical lifetime derived due to OH reaction is 95 days. The lifetime with respect to Cl is significantly longer than this and the lifetime with respect to photolysis is also longer than this. So combined the actual atmospheric chemical lifetime is most likely > 40 days. However the derived value in the paper from burden/sink is 20 days which to me indicates a problem with the formulation of either the burden or the sink terms that needs to be investigated further.

Another shortcoming is the chemical representation in the model. The chemical scheme appears to be rudimentary and also seems to have errors.

The authors state: “Initial tests using a yield of 0.72 resulted in an overestimated chemistry source, leading us to re-evaluate this yield for the specific mixture of VOCs represented in the GISS ModelE2.1. Estimated mole fractions of propane (11%), butane (22%) and pentane (21%) in anthropogenic emissions were multiplied by each compound’s acetone molar yield (0.73, 0.95, 0.63, respectively), determining that 42% of paraffin from anthropogenic sources becomes acetone”.

The authors don’t reference where they got their estimated mole fractions from anthropogenic emissions or where they got the molar yield of acetone for each compound. Furthermore the molar yields are incorrect. Although propane does in fact form acetone with a significant yield, butane and pentane do not.

In summary, I refer to my comments above in the second paragraph.