This is an excellent paper, using results both from MLS measurements and the WACCM model to obtain trend estimates of ClO and HOCl over the ~15 years for which MLS has been making measurements.  With the exception of requiring a better explanation for a line used in two of the figures, this manuscript is certainly publication worthy in its present form.

However, I do have to express disappointment with this study because almost all of the analysis of this wonderful MLS dataset and the sophisticated modeling study that accompanies is reduced to plots of linear trends.  Beyond some lines drawn through two figures (which are dominated by annual cycles) the reader is left with no sense of how well a linear trend actually fits this data.  The authors use a quite complicated regression fit that includes some quite long-period terms such as ENSO and 11-year solar cycle terms.  Before comparing the trend fits between model and measurement, it would be good to know how well these agree or whether they are significant, since differences in these terms could influence the calculated trends.  If they don’t make a difference please say so.  

A few simple measurement (and possibly model) timeseries plots of annual ClO and HOCl anomalies (without ENSO, F10.7, or QBO fits) or something similar from 50S to 50N at a few of the altitudes shown in Figure 10 would be of great interest.  It would visually help the reader to understand how easy it is to identify a linear trend in this data, would help to answer the question of the importance of the multi-year terms in the fit, and would provide some indication of the importance of endpoints.  At present, only Figures 2 and 8 provide any timeseries information, and these are very cluttered and difficult to read, dominated by the annual cycle, and given only for very specific (for some reason different for the 2 species) locations. 

Figure 2 and Figure 8 – I’m afraid that I don’t understand the meaning of “the model fit to MLS data”.  Is this just WACCM minus MLS?  The authors seem to be using “fit” to refer to the regression.  Line 246 seems to suggests that the pink line is just a difference: “WACCM time series actually fit the MLS data better than the regression fits do”.  I indeed hope that this line is just a model minus MLS difference with the bias removed, since this would seem to be the most useful and basic thing to plot.

Minor points:

Line 138 – “only” is an unnecessary word here.  Also, somewhere in this paragraph it should be mentioned that additional details about the standard HOCl retrieval are included in the MLS data quality document (i.e. Livesey et al. 2020).  I realize that is mentioned in 3.2.

Lines 178-193, or thereabouts- I’m pretty sure that “the model” is always referring to WACCM, but it would be nice to spell that out somewhere in this long paragraph.

Line 211 – Just curious as to why here the fitted component apparently follows a different solar model than that mentioned in 2.2.

Line 216 – “year-long blocks”.  I agree that this is probably a reasonable choice, but I’m just curious if the authors have any particular reason for this choice.  I certainly don’t insist on any change in the manuscript. 

Review of the manuscript ACP-2021-880 by Froidevaux et al. submitted to ACP and entitled “Upper stratospheric ClO and HOCl trends (2005-2020): Aura Microwave Limb Sounder and model results.

This manuscript presents satellite measurements and model simulations over 50°S-50°N for the time period 2005-2020 for two inorganic chlorine reservoirs, namely chlorine monoxide (ClO) and hypochlorous acid (HOCl). These global measurements characterize the upper stratosphere with a vertical resolution of 3-4 km (ClO) and 5-6 km (HOCl), they have been derived using the Optimal Estimation Method from Aura-MLS radiometric observations that have been consistently gathered over 16 years (i.e., without the hardware failures that have affected other MLS products/channels, e.g., HCl, N₂O). Online and offline products are used, and they present a very good sampling, with about 3500 profiles available per product and per day.

This places the authors in a good position for robust trend determinations for difficult targets, especially HOCl, in support of the Montreal Protocol on substances that deplete ozone. To this end, they use a method or approach that accounts for the auto-correlation often present in geophysical data series, and their model further includes proxies for parameters known to affect the abundance of stratospheric tracers (the solar cycle, the QBO, …).

The investigations are supported by model simulations performed by the WACCM model constrained by MERRA-2 meteorological fields. These simulations are available at the observation sampling, and they are analyzed in the same way than the observations to provide ClO and HOCl climatologies and trends. The model is also used to perform a sensitivity study in order to determine the dependence of the simulated HOCl distributions to the assumed kinetics for its formation reaction (ClO + HO₂ -> HOCl + O₂), these parameters being still insufficiently defined, and possibly responsible for significant satellite-model biases.

All these investigations are conducted with great care and the results are well presented. The manuscript is well organized, and all the figures are clear, self-explanatory and useful. The text is sometimes a bit lengthy, for instance in section 4 (discussion), but this is probably more a matter of taste than a real issue. It is important to note that this study is dealing with inorganic chlorine reservoirs for which only few observations are available (and the measurements are challenging!), its publication would therefore bring original and interesting elements to the community, also in the context of the continuous evaluation of the success of the Montreal Protocol. In my opinion, this manuscript is almost ready for publication, and I only have a few suggestions to bring to the attention of the authors.

Primary suggestions or questions.

In the introduction (line 73), the authors remind us that mid-term variability complicates the trend detection in the lower stratosphere. Among others, Strahan et al (2020) have reported about hemispheric asymmetry in stratospheric transport trends and its impact on the distribution of long-lived tracers. This question is absent elsewhere in this manuscript and global trends are consistently reported. Moreover, most of the material/figures presented do not allow the reader to make his/her own opinion about possible upper stratospheric signals that could characterize the observed and modeled ClO and HOCl data sets. The time series are restricted to a Northern/Southern belt for ClO/HOCl (Fig. 2/8). Still, Fig. 3 and 4 exhibit some asymmetric signals, e.g., at 31.6 hPa for the first one, or the “light green bubble” seen for ClO in the Southern hemisphere in the 20-40°S and 2-10 hPa ranges in Figure 4. My bet is that the authors decided to report global trends because it is already challenging enough, especially for HOCl. But still, I would suggest to indicate in the introduction and/or conclusion that there were no signs of asymmetry in these upper stratospheric data sets, or that they were not searched for.

In section 2.2 (line 190), it is indicated that the WACCM6 runs have been augmented to cover the more recent years than the initially available simulations. But since the boundary conditions for the halogenated source gases have been provided by a reference published a few years ago (Meinshausen et al., 2017), I wonder which data have been used in order to describe the post-2016 evolutions of the source gases? A more recent reference is probaly neeeded here.

In section 4 (starting line 505), the authors indicate that changes in upper stratospheric temperatures should have had little effects on chlorine partitioning, with less than a 1K decrease as observed by Steiner et al. (2020) over the period of interest here. Regarding the partitioning and trends derived from WACCM6, I guess that the (small) temperature change is accounted for thanks to the MERRA-2 meteorological fields? Or in other words, is the temperature trend in MERRA-2 in agreement with the ~1K trend of Steiner et al. (2020)?

Figure 3: I would suggest here to swap the axes (latitude for x-axis; trend for y-axis). This way, Figure 3 would report the latitude on the horizontal scale, as is the case for Fig. 1, 4 and 6.

Minor point

In the introduction (line 59), I would also mention the effect of photolysis (in addition to transport and mixing) to explain the conversion from tropospheric chlorine into the reservoirs.