General Comments

This manuscript presents an inversion study of five HFCs (different combinations of three at a time) to infer global annual mean hydroxyl radical (OH) concentrations using a 12-box model.  The inferred OH anomalies are compared against other estimates from past MCF and CTM-based analyses.  Finally, the impact of the optimized OH on the growth rate and emissions of CH₄ is derived from the same 12-box inversion relative to a Spivakovsky et al. climatology.  The results suggest that variability in the annual OH anomaly is less than 2% with no trend over the period 2004-2021, that OH abundance in year 2020 was likely low but not significantly lower than in prior years (2018 especially), and that CH₄ emissions using the optimized OH had a smaller increase than is inferred using climatology, though the difference is small.

Overall, this is a compelling study focused on an important topic.  The global oxidizing capacity is a subject of much debate, and further observational constraints to quantify it are always needed.  The methodology used is sound, and I see no shortcomings in what is presented.  I point out in my comments an opportunity for expanded discussion and a couple small clarifications, but otherwise, I think the article is well-presented, is of interest to readers of ACP, and represents a significant advance beyond the use of MCF as the main observational proxy of OH.

Specific Comments

L39: Turner et al., PNAS, 2017 (https://doi.org/10.1073/pnas.1616020114) could be added to the list of MCF studies

L275: It would be informative to expand on the discussion of the “shortest lived species” a bit.  It is stated earlier in the manuscript that the derived OH may be more sensitive to the shorter-lived species in the inversion.  Wouldn’t it also make sense that, in the actual atmosphere, the shorter-lived species would adjust more quickly to either changes in emissions or variations in OH?  This is not explicitly stated in the text, but since the authors separated out inversions that included the shorter-lived species (e.g., in Fig. 3), why not discuss the implications more?

Figure S1: For the lower row of panels in each set, I think it would help to indicate on the y-axis that this represents a difference (something like “HFC-32 Difference, AGAGE – NOAA (ppt)” or similar).

Technical Corrections

Figure S5: Should the caption state “ten inversions” rather than six?

Figure S6: Figure seems low resolution, e.g. when compared to Figure S5.

Comment on "Estimation of the atmospheric hydroxyl radical oxidative capacity using multiple hydrofluorocarbons (HFCs)" by Rona Thompson et al.

GENERAL COMMENTS

The paper describes the setup and result of an inverse modelling study to derive global hyrdoxyl radical (OH) concentrations from hydrofluorcarbon (HFC) observations for the period 2004-2021.  Concentrations of OH are important for the estimation of the global CH4 budget, which is the second most  important greenhouse gas after CO2, and as OH is one of the main uncertainties in there, this study is of great relevance.  The method is based on the idea that emissions of selected HFCs  are more or less known, and that destruction by OH is their only loss-process; observations of these HFCs are then used to decrease the uncertainty in OH concentrations as well as emissions. Similar approaches have been used in combination with methylcloroform (MCF) and to some extend also with HFCs, but this study is unique it its focus on 5 different HFCs and the long time span of 2004-2021. 

A strong element of this study is its completeness. Inversion experiments have been performed for all 10 possible combinations of 3 our of 5 available HFCs observation time series, 2 adjustments of observation scales have been compared, and also sensitivity experiments were done. The resulting range of estimated OH concentrations are compared with results obtained in other studies, and found to be in good agreement. Special attention is also paid the possible impact of the 2021 lockdowns on concentrations. In addition, also the impact of the result on CH4 concentrations is discussed.

The study uses a 12-box model to simulate global atmospheric mixing. Although this is a very crude model, it makes sense here since the amount of observations is limited, and results are discussed at global scale only. Could the authors indicate what is the expected gain from using a full CTM, as suggested  in the discussion at line 330? The current paper does not discuss any results at the level of the 12-boxes, for example by latitude band or by atmospheric layers. Could useful information already be found in here, for example on the difference between tropo- and stratosphere?

For the discussion it might be useful to spent some lines on the long term perspective of OH studies. A major driver for this study was the current amount of MCF remaining in the atmosphere these days is too small to be used as proxy for OH. Due to the Kigali Amendement to the Montreal Protocol, also emissions of HFCs are expected to decrease and eventually disappear. What is the time range for which the authors expect that their method can be used? And do they have suggestions for observations that should be soon setup to ensure that also in following decades estimates of the OH concentrations can still be made?

SPECIFIC COMMENTS

line 101: Could some main characteristics of of H(x), the AGAGE 12-box model, be described here?

  For example, how is the exchange between the boxes parameterized?

line 152-153: The temporal correlation scale length that is described here, is that in combination  with an e-folding shape?

line 159: The "reanalysis data" mentioned here, is that ERA5?

line 159: There are no correlations assumed in the prior OH concentration uncertainties?

line 274: correlation with ENSO Index in Figure 4b: maybe a scatter plot (with connected dots?) would be more clear here to show the (lack of) correlation.

SPELL AND GRAMMAR

It was a pleasure reading the paper, it is very well written and illustrated. Textual comments are therefore very limited.

line 152: "The emission errors in each box were assumed to be uncorrelated *with other boxes*, ...

line 157: ".. the temperature for each month and box was taken *from* the European Centre ..."