Dear authors, your manuscript will have to undergo a second round of review. But before this is initiiated, please complete the answer t the first review first. I could not find the answer to the last of remarks of reviewer #1 on the page C4 of RC1.

Also I believe, that you missed part of the first question of reviewer #2.

In my understanding, the reviewers raised critical points which you did not answer. If I am wrong, please point me to the answers to this questions.

In the process of checking your answers to the referees, please check again, if you completed the review.

Kind regards
Mathias Palm

Dear authors, thanks you for your additional work. The main points in
the latest round of review is that while you acknowledge the
limitation of your work, it is not reflected in the title and the abstract.

Review #1

The authors discussed the comments in the response letter. However,
only a few of the comments are addressed in the manuscript. I
recommend the authors go through all the comments again and at least
include most of the comments in the revised manuscript. In addition, I
still have some comments that are not well discussed in the response
letter. I think at least a major revision is needed.

Other specific comments:

1 Before inversion, the prior emissions were pre-optimized
to fit the global mean MCF mole fractions. The authors argue that the
pre-optimized MCF emissions can reduce computational costs. But the
inversion as shown by equation(1) is to estimate the emissions and OH
by combining the information from both bottom-up estimated prior
emission inventories and the observations, as well as their
errors. The pre-optimize erase the information of prior emission
inventories, and only keep the information of observations. I don’t
think this is the right way to do an inversion. Besides, the question
is not answered:” Will the pre-optimization reduce the OH variation
estimated by the inversions since the MCF emissions already fit the
observations?” In addition, from the author’s response to Fig.4, I
feel the pre-optimization is somewhat arbitrary, which makes the
inversion lost the prior information.

2 The authors explain why the small correlation coefficient between OH
and MCF variations can reflect the OH and MCF are independently
derived. But I think a better indicator should be comparing optimized
minus prior MCF emissions and optimized minus prior OH. We can see
that in Fig.3, the MCF emissions estimated by REF inversion is much
higher than prior around 2013, which is corresponding to the large
positive OH anomaly around 2012-2013. This may indicate that the
inversion system cannot separate the OH and MCF variations.

3 For the convergence problems. From the author’s discussion, the
10-year inversions can reach convergence since they require less time
per iteration. One problem is that the 10 years inversions are for
1998-2008 when the MCF emissions are higher than 2009-2018 and the
corresponding errors are much lower than 2009-2018. If the 10 years
inversions focus on 2009-2018, it will be hard to say if the 10-year
inversion coverage to similar OH variations since the uncertainties in
MCF emissions (reach 200%) are much larger OH during 2009-2018. So I
don’t think the 10-year inversion for 1998-2008 can prove the
robustness of the 20-year inversions, as the author mentioned when
discussed my last comment.

4 “L352: Why further convergence will result in less realistic OH
variations?” The authors answer this question by adding Figure S8
(but show nothing in the manuscript) which showed that the inversions
are overfitting. Is this because the inversions use too small
observational error?

5 “L372: “Firstly...we generally identified similar tendencies in
each.” Figure 1 has shown the variations of OH estimated by three
inversions are quite different.” The authors answer this by showing
the 10 years inversions are similar. But as aforementioned, the three
10-year inversions are similar may not prove the inversions for
2009-2018 can also reach similar results. Here the only thing we see
in the main text is that the three 20-year inversions are quite
different.

6 “L374-379: Here the manuscript tried to prove the robustness of the
OH interannual by an additional inversion and a forward
simulation. But the details of the two experiments and the results are
not given. I suggest include some details in the supplements. E.g. how
the one global scaling factors compare with the REF, POP, and TM5OH?
Is the forward simulation use the prior or optimized MCF emissions? I
cannot understand the logic here, can you clarify why the two
experiments can indicate the robustness of the derived OH variations?”

It is still unclear how the authors conduct the two experiments. I
think every model experiment established should be introduced in the
manuscript or supplements. Since the two model experiments are not
shown clearly, the role of the two experiments is certainly unclear.




Review #2

I thank the authors for giving the reviewer comments careful consideration.

The authors do not propose too many major changes to the paper as a
result of these comments. They acknowledge that there are some
limitations to their study (e.g. lack of convergence), but make the
case that it is too late to change the experimental design, and argue
that there is enough useful information in the paper to warrant
publication. I am inclined to agree with this assessment. However, I
suggest that the limitations should be spelled out more clearly in the
abstract. I suggest adding some text along the following lines (it
doesn’t have to be exactly this, but I think these caveats need to be
noted):

- Title: I’m still quite concerned about the lack of
convergence. Therefore, I wonder if it’s more accurate to remove
“inversion” from the title. Perhaps something like: “An
investigation of top-down constraints on atmospheric oxidative
capacity using methyl chloroform and a global 3D model”

- At the end of the first paragraph, or start of the second paragraph,
I suggest noting the lack of convergence (e.g. “While our main
solutions did not fully converge, they suggest interannual
variations in the global oxidative capacity…”)

- When talking about the inter-annual variability, I suggest being
explicit that the standard deviation of the derived OH concentration
was small (< 3% per year) over the 20 year period, but that
substantial fluctuations (~±10 %) were derived between certain
years.

- At the start of the third paragraph of the abstract, it needs to be
made clear what the “adjustments” are with respect to
(i.e. “compared to a widely used prior OH distribution…”, or perhaps
even cite the relevant dataset).

- At the start of the fourth paragraph of the abstract, clarify
(perhaps in parentheses) broadly what the added value of the 3D
model is.

- Final sentence of the abstract: Given the lack of convergence, I
think it is too strong to suggest that these particular results be
used in studies of the CH4 budget (which is implied), and therefore
suggest that this be softened to “Therefore, we consider that
variations in OH derived from MCF inversions with 3D models can add
value to studies of the budget of e.g. CH4”


---

In the supplement, line 35: Which section do you mean S2?