Review of "The interhemispheric gradient of SF6 in the upper troposphere" by Tanja Schuck and colleagues. 

This article analysed long-term measurements of SF6 from many aircraft campaigns, covering a period of 2006-2020. They have calculate the age of air based on the aircraft observations in the upper troposphere, in reference to continuous surface measurements. Further explanation to the observed results are derived from 12-box model of the atmospheric transport. The paper is very well written and organised. My main concern is the use of a very low resolution of the transport model, which I think would be very difficult convince the reader as a state-of-the-art. The limitation of 12-box model are known and I do not need to elaborate (overstretched to compare with measurements). Given the importance of the data set prepared for this analysis and discussion of the possible outcome, I would recommend publication of the article after minor revision. Hopping that the results presented will inspire future research activity. 

Minor comments: 

Line 21 : Aren't SF6 lost also " by electron attachement" ?

Line 24 : I think the age of air concept was laid out first in Bischof et al., 1985

Line 38 : This phenomena was first reported elsewhere, using CO2 in a landmark paper (Nakazawa et al., 1991).

Page 11: TransCom-CH4 experiment (Patra et al. 2011) found dependency of the lag time with emission patterns and rates. Are the differences between the campaigns arising from emission dependency or the interannual viability in transport itself ?

Line 240 : It would have been good to check using the model experiments in TransCom-CH4, to probe the trends in inter-hemispheric exchange rates.

Figure 5 (tropics north; green triangle), and line 9/10: are they consistent ?

Finally, regarding the data availability: will it be possible to create a dataset for all the aircraft campaigns with troposphere flagging for further research. I see that as a useful outcome of this paper.   

References: 

Bischof, W., R. Borchers, P. Fabian, and B. C. Krüeger, 1985: Increased concentration and vertical distribution of carbon dioxide in the stratosphere, Nature, 316, 708-710, doi:10.1038/316708a0.

Nakazawa, T., K. Miyashita, S. Aoki, and M. Tanaka, 1991: Temporal andspatial variations of upper troposphere and lower stratospheric carbon dioxide, Tellus, Ser. B, 43, 106–117. https://doi.org/10.1034/j.1600-0889.1991.t01-1-00005.x

Patra, P. K., Houweling, S., Krol, M., Bousquet, P., et al., 2011: TransCom model simulations of CH4 and related species: linking transport, surface flux and chemical loss with CH4 variability in the troposphere and lower stratosphere, Atmos. Chem. Phys., 11, 12813–12837. https://doi.org/10.5194/acp-11-12813-2011. 

This paper uses in situ trace gas measurements, primarily SF6, from the upper troposphere and surface along with a 12-box model to investigate tropospheric transport and SF6 emissions over a recent 15 year period.  The combination of US and European based aircraft data sets provides a relatively complete latitudinal and temporal representation of the upper troposphere over this time period.  The authors optimize both emissions and transport parameters within the 12-box model to best match the observations resulting in a number of interesting findings.  This is a really nice use of a simplified model to diagnose what the observations can tell us about large scale features of atmospheric transport and trace gas emissions.  The methodology and discussion of results are clearly described.  I recommend publication in ACP with consideration of the minor comments below.

Specific comments:

Section 4 or 5:  It would be nice to include a brief comparison of previous model estimates of the interhemispheric transport time such as from Waugh (2013) and Orbe (2016, 2021) to the original estimates from the 12 box model.  The Waugh (2013) transport time was less than 2 years to the SH but still somewhat longer than the observational transport time.  This comparison would help give the reader an idea of how much CTM transport needs to be adjusted to better match the observations.

Line 165:  Related to the above comment, it might be helpful to briefly state how the initial values of T were obtained for those not familiar with the Rigby 2013 paper.  For instance, were they based on reanalysis output or a best fit to observed mixing ratios of some trace gases?

Lines 248-9:  The difference between the PBL and UT gradient is also much smaller in the model compared to the observations.  That seems worth pointing out here.

Line 261: ‘were performed’

Review of “The interhemispheric gradient of SF6 in the upper troposphere” by Schuck et al.

General comments

This study combined upper tropospheric (UT) SF6 datasets from different measurement projects and attempted to interpret them in terms of emissions and transport. Such efforts to utilize datasets from multiple projects are of particular importance from long-term point of view. I think that a follow-up study using a 3D transport model is needed, but this manuscript has made a good job to motivate such future studies. Before agreeing to publication of this study, I would like to encourage the authors to consider my comments below to enrich scientific discussion and provide better guidance to following studies.

Orbe et al. (2021) presented detailed analyses of the SF6 age (or lag time as called in this study) including the data from the ATom campaign. This study repeats analysis of the ATom data, and extends to CARIBIC, GhOST and HIPPO, therefore number of the data in the UT is extensively increased. However, spatial distributions of the SF6 age presented in this study is Figure 4 only. With the data newly analyzed here, the authors could provide more presentations about spatial variations of the SF6 age like Figures 4 and 5 in Orbe et al. (2021). In particular, using the CARIBIC data, the authors might discuss longitudinal variation of the SF6 age over NH, which could reflect distribution of the underlying sources as well as areas of effective vertical transport. Also, the authors might confirm (or discuss difference of) the vertical gradient of the SF6 age over different latitudinal regions in comparison to Orbe et al. (2021). I think that such more in-depth analysis would highlight value of this study and further motivate follow-up studies using a 3D transport model.

One important question (but I cannot find the answer clear) in this study is whether the SF6 age in the UT of SH has a secular decreasing trend, in contrast to Orbe et al. (2021) who showed negative trends of the SF6 age south of 30°N at the surface. In abstract and conclusion, the authors argue that it exists, but they describe it a bit more carefully in section 3. As the authors writes, data are relatively limited in SH, so the trend line shown in Figure 5 could be unrepresentative (not free from sampling bias). It would be interesting to see the authors’ argument more clearly about whether they regard the trend significant/well represented or still difficult due to limited availability of data. Regarding this, I wonder about possible use of the modeling results. The authors optimized the model (though it was tuned by more number of NH data). Figure 8 indicates almost constant value of data minus model with time in the UT exT-S. Does it mean that the model consistently shows a negative trend? What if the authors add the optimized model trend line to Figure 5? Thinking that the additional constraints from the newly analyzed data seem to be value of this study, I would like to see extended presentation of the outcome (e.g., modified Figure 5 or new figure after Figure 8) to answer the question.

Minor comments

Introduction: Interhemispheric transport with surface emissions mainly in the NH mid latitudes shapes latitudinal gradient of trace gas of interest. Citing previous studies, the authors highlight more effective interhemispheric transport in the UT than near the surface. Many trace gases with major emissions in the NH (e.g., CO2, CH4, CO, SF6 etc.) shows smaller latitudinal gradient in the UT than near the surface, but this cannot be explained by different efficiency of interhemispheric transport in different vertical layers only. I think that discussion on upward air inflow from surface to the UT over the tropics (the major pathway of surface air into the UT), is missing despite that it also mitigates latitudinal gradient in the UT. Japanese groups have made significant contributions to interpretation of latitudinal gradients of CO2 and related gases in the UT and near the surface (Nakazawa et al. 1991; Miyazaki et al. 2008; Sawa et al. 2012; Bisht et al. 2021). The authors also mention to the Asian summer monsoon. Important in the context of this study is probably that Asian summer monsoon effectively uplift South Asian surface air (“polluted” air) to the UTLS, the gateway to interhemispheric transport as the authors cite previous studies like Yan et al. (2021) and Belikov et al. (2022). Such an express pathway from NH low latitude surface to the UT of SH also weakens the latitudinal gradient in the UT. I think that the introduction could be reformulated so as to inform readers about supply of surface air into the UT by various pathways also play an important role in determining the latitudinal gradient in the UT.

P5 L114: I think that the original measurements were not made on the WMO X2014 scale (officially the scale is named without “NOAA”). It would be better to describe about the original measurement scale and the conversion applied retrospectively.

Table 1: The precision of SF6/N2O could be given in ppt/ppb, to be consistent with the following discussions. Use of ppb/ppbV is confusing. I think that the authors distinguish them on purpose, but no explanation is given. For instance, TRIHOP CO is in ppbV while UMAQS CO is in ppb despite the same measurement principle. This should be sorted out or consistent explanation should be given.

P7 L135: “Because of their higher time resolution” high resolution does not mean observation of large variability. High-resolution measurements could capture phenomena that could not be resolved by low-frequency sampling, but underlying nature is that large variability is happening at place of interest. For instance, SF6 measurements at a very remote site (e.g., Antarctica) show small variability even if measured at high frequency.

Figure 2: The reference SF6 time series could be added to visualize the concept of the SF6 lag time.

Section 2.3: Orbe et al. (2021) pointed out that different choice of the reference time series could result in different trend of the SF6 lag time. The authors could discuss how this affects and why the simple marine boundary layer choice is made in this study.

Figure 3: It would be interesting to see latitudinal distributions from surface sites along with those in the UT.

Figure 4: As in my earlier comment, I hope to see more in-depth analyses of the SF6 lag time, not only for latitude but for longitude and altitude.

P11 L195: As I commented to introduction, the smaller latitudinal gradient in the UT than near the surface is not fully attributable to active interhemispheric transport in the UT but is also contributed by intrusion of surface air into the UT.

Figure 5: As in my earlier comment, this could be compared to results of the optimized model output. The trend line for ex-tropics south is hard to see in its color. Labels could be larger in size.

Section 4: The authors mention to possible southward shift of SF6 emissions in the NH as discussed in previous studies (e.g. Orbe et al. 2021). I think that the authors could have made an experiment in which latitudinal emission patterns change with time in their box modeling. Yang et al. (2019) pointed out that vertical transport in the extratropics of NH (T04 in this study) is also important for north-south gradient of the SF6 age. I wonder whether the authors have made sensitivity tests for this parameter. The authors argue that “horizontal transport seems to be more important than vertical transport” (P18 L333), at which I wondered about inconsistency with Yang et al. (2019).

Reference

Nakazawa et al. (1991) https://doi.org/10.1034/j.1600-0889.1991.t01-1-00005.x

Miyazaki et al. (2008) https://doi.org/10.1029/2007JD009557

Sawa et al. (2012) https://doi.org/10.1029/2011JD016933

Bisht et al. (2021) https://doi.org/10.1029/2020JD033541