Review of Vogel et al., Evaluation of vertical transport in the Asian monsoon 2017 from CO2 reconstruction in the ERA5 and ERA-Interim reanalysis

The paper by Vogel et al aims at quantifying vertical transport in the UTLS of the monsoon region. They combine in-situ measurements of CO2 with simulations of the Chemical Lagrangian model of the stratosphere (CLaMS) driven by ERA-Interim, ERA5, and 1x1 regridded ERA5 reanalysis data.

They apply backward trajectory transport analysis extending backward by more than a year with age of air derived from CLaMS for the different driving reanalysis data sets and compare these with long-lived tracers to infer ascent time scales.

They use surface CO2 observations in different regions and combine these with the trajectories and show that the reconstruction using ERA5 gives a good agreement of reconstructed CO2 and measurements up to 410K, Above the reconstruction is affected by mixing with stratospheric air.

The authors conclude, that the results are highly sensitive to the representation of vertical transport in the troposphere in the different reanalysis data sets. According to their methods ERA5 yields the most reliable results compared to the observations. Using quasi-inert tracers (C2F6, HFC-125) they their results indicate a good agreement with ascent rates from ERA5 (also 1x1) with large mean age differences at 470 K between ERA-Interim derived age and ERA5 (1x1) of about one year.

The paper is well written and the methodology is clearly given. The results regarding the different reanalysis data sets are important for the community, since a lot of conclusions on stratospheric transport were based on ERA-Interim before the release of ERA5. The reconstruction with CO2 is impressive and balanced discussed. Therefore the paper clearly merits publications and I have only a few comments, which are minor.

Minor comments:

Since a large number of species have been measured at the STRATOCLIM mission, I wondered, if one could include other shorter-lived species to further support the transport time results above the tropopause.

In general shorter-lived species should fade out (NMHC) or decrease to background  (CO) when being uplifted. I wondered, if the authors thought about including such constituents, which would strengthen their estimates at least above the tropopause.

Fig.2: Could you add the Mauna Loa curve and the classical tropical boundary condition for CO2 at the tropopause as given by e.g. Andrews et al., 1999, which is the mean of American Samoa surface cycle and Mauna Loa?

l.315-335: Ascent rates: Would it be possible to support the ascent rates (20days) with measured vertical gradients of short-lived species, which should show a considerable decrease over 20 days?  

This would complement the stratospheric analysis based on the very long-lived species presented in Fig.10. Was SF6 available for age calculations?

l.392: How reliable is the use of just one location at the surface to derive mean transport time? The authors state in l.400 ff that a detailed CO2 reconstruction using comprehensive data is needed, which makes more sense. I'd recommend to skip l.392-397.

Fig. 10 (and general discussion of mean age of air): How well does CLaMS age of air resembles the observational derived age of air (either by the species in Fig 10, or by CO2 itself or eventually SF6 or N2O)?

Fig.7: Looking at Theta > 430K: Which role plays transport and mixing from the TTL and tropical lower stratosphere for the calculation of fractions and further below the transport time estimates, also for the age of air and the CO2 reconstruction?

The CO2 cycle at the tropical tropopause is probably similar as at the monsoon tropopause, but how does this affect the reconstructed values and times?

References: Andrews et al., Empirical age spectra for the lower tropical stratosphere from in situ observations of CO2: Implications for stratospheric transport, JGR, 1999, doi/epdf/10.1029/1999JD900150

The manuscript describes the transport properties in the Asian monsoon region during the StratoClim campaign in July-August 2017 and compares the results derived from ERA5 and ERA-Interim reanalysis with quantities derived from the observations. At first sight, it looks like an extended appendix of a published paper (Vogel et al., 2023, V2023 herafter) and there are indeed a number of common figures and elements of text but it brings also a number of new useful results beyond the reanalysis comparison. However, these new results are not necessarily well discussed or exploited and there are a number of problems that need to be addressed, each one being relatively minor but resulting in a fairly heavy weight when added together.

General remarks : First, the title does not reflect the content of the paper. One expect a focus on the CO2 reconstruction while it is only used in section 4.6, that is basically one page, among 15 pages of results which are mostly about proportion of boundary layer air in the campaign samples and its age. Besides this, this section is the least conclusive of the results. Therefore the title should be changed to indicate the real focus of the paper. Then, more comparison should be made with previously published results, in particular with Bucci et al. (2000) (B2020 hereafter) who produced results and figures directly comparable to several ones of the manuscript. Third a careful rewriting should be made as a number of sentences are too long, with several subordinates and hard to read, and a number of needed clarifications must be added.

Detailed comments :

1) 3.1 The number of backward trajectories used in this study (11000) is small by present standards and certainly limits the statistics that can be produced. For comparison, B2020 used 1000 more trajectories in their study of the same campaign for a better accound of mixing.

2) 3.1 The choice of a transition at p/psurf = 0.3 does not mean 300 hpa over the Tibetan plateau but rather 170 hPa on the average, that is just in the middle of the TTL. This choice is rather infortunate and might have som impact on the results.

3) 3.1 In any case, as the backward trajectories are run to the boundary layer, a significant part of the path is under the region of the potential temperature levels and the applied method should be described. In particular, it is very useful to know whether convection is represented and how, and whether this differs along the reanalysis.

4) 3.1 It would be useful to know whether the 1° x 1° version of ERA5 is obtained by subsampling or filtering the high resolution version. The fact that the tropopause range (fig. 12) is about the same in the two version suggests the first choice which is not the good one.

5) 3.2, l 167-168 : It is hard to understand what kind of interpolation can be performed if data are available only at 3 locations over the Asian continent. It is also unclear what is actually used in the reconstructions shown in 4.6.

6) 3.2 The choice of the boxes shown in fig.3 is a bit hard to understand regarding the continental boundaries. Why is the box surrounding CL and called Bangladesh extended only to the east to cover Birmania ? Why is the Tibetan Plateau truncated on the north and west. Why is NTL influenced by the Gange and Indus valley agriculture and industry representative of the Tibetan plateau where there is no industry and a delayed vegetation cycle ? As far as I can see from V2023, the reconstruction up to 400K depends mostly to the ground seasonal cycle at NTL and the reconstruction performed above might result from a clever design of the region boundaries. It would be useful to know whether this design was done a priori or a posteriori.

7) 3.3 This description again misses to mention what is done in the tropospheric levels below the 0.3 p/psurf transition and whether convection is parameterized and in the same way as the backward trajectories. It misses overall to mention what is calculated. If the ages are sampled along the StratoClim flight tracks in the 3d ClaMS records, the comparisons made in this paper are sound. If the ages are sampled on each level over a wide area and several decades, the basis of the comparison is much more fragile and their meaning is questionable.

7bis) 3.3 198 : This sentence seems copied and pasted from a description related to BDC calculations. The transit time is here from the BL This sentence also does not suggest that the age is sampled on the StratoClim flight tracks.

7ter) 3.3 It should be mentioned whether the calculations used here are a by-product of the calculations used in Ploeger et al. (2021) or are new and specific.

8) 4. 214 : This sentence is an example tht requires rewritting.

9) 4. The word « released » for parcels reaching the BL backward in time is somewhat confusing. At least a sentence should be added to define exactly what is meant.

10) 4.1 Fig 4a could be compared to Fig.10 of B2020.

11) 4.1 The transition at about 370 K for young trajectories is similar to the crossover defined from forward trajectories by Legras et al. (2000). This is not by chance as a discontinuity should also appear in the backward trajectories that represent convection.

12) 4.1 The age of air is bounded above 400 K for the young trajectories and above 440 K for the old trajectories. This is an effect of the truncature of the age spectrum. In principle, since the age spectrum is calculated and shown in fig.11, it should be possible to perform a tail correction like in 3D ClaMS. Perhaps this is difficult due to the small number of trajectories. As the median is only 250 days for long trajectories extended over more than a year, it means a fairly flat tail in the age spectrum.

13) 4.1 By the way, why using here a median age and not a mean like in 3D ClaMS ? Does it differ strongly from the mean ?

14) 4.1 It is not easy to understand exactly what is shown by the gray area. Are the boundaries of this area the two alternate age curves ?

15) 4.1 An interesting result of Fig. 4a is that long trajectories are needed to reproduce a pattern of the BL fraction that matches the N2O curve shown in V2023. This has some implication for the confinement of the Asian Monsoon Anticyclone (AMA). As there is almost no in-mixing up to 400-420 K (Vogel et al., 2019, Legras et al., 2020). This means that the old tropospheric N2O air was captured when it was formed in 2017 and kept inside.

16) 4.1 The drop of the BL fraction from about 90 % at  370 K to about 0 % at 420K in two months for young trajectories is compatible with a mean ascent rate of about 50/60 =  0.8 K/day inside the AMA. This is smaller that the estimate of Legras et al. (2020) which is 1.1 K/day but larger than the estimate made in 4.5.

17) 4.1 l.250 : Specify your are commenting the old trajectories curve.

18) 4.1 l.260 : The sentence starting with « In pure back-trajectory ... » is distracting and could be removed.

19) 4.1 l.264 : As far as turbulent diffusive mixing is concerned, that is the type of mixing represented by the basic algorithm of ClaMS, it should not influence the dispersion of a group of parcel beyond a few days. After this diffusive time, the dispersion is due to shear and strain and becomes exponential in time. If now the convective mixing parameterized in Konopka et al. (2019) is concerned, longer time effects are expected. This is were an accurate description of what is done regarding convective mixing in section 3 is missing.

20) 4.1 The final discussion and final sentence of this section are somewhat delusive. Why showing the curves from 3D ClaMS in Fig.5 if no conclusion is drawn besides some technical comments ? If the authors believe that the differences could be explained on such technical basis, this should be thoroughly tested as it jeopardizes all the discussion. I believe they are not but more discussion is needed here.

21) 4.2 Fig.6 should have the Tibetan plateau outlined and it should be compared with Fig. 4 of B2020

22) 4.2 Use a readable color scale for Fig. A2 and add it to the figure.

23) 4.2 Fig.7 should be compared with Fig. 10 of B2020

24) 4.2 l.312-314 : « However ... » I do not understand this sentence.

25) 4.3 l.322 Add that iit is the first day after the beginning of the backward trajectory if it is the case.

26) 4.3 l.341 : What do you mean by « an idealized parcel » ? Do you mean an age calculation made on a single parcel without any averaging ? Nobody is doing that and this sentence is basically misleading and useless. Please explain better why the two calculatios can be compared.

27) 4.4 I wonder why this short section and Fig.11 has been dropped here and not merged with 4.1.

27) 4.4 Fig.11 should be drawn with logarithnmic vertical axis.

28) 4.5 This section is not at all about CO2 but purely about transport and ages. Fig. 12 does not display any obvious differences between the three cases and what is new in Fig.13 with respect to Fig. 4. I believe this section is a remain of an early version of this manuscript derived from V2023.  The content should be merged with 4.1

29) 4.5 l.372-374 The right location for generalities on CO2 is the introduction.

30) 4.5 l.381 : Add a reference to Tegtmeier et al., (2020) when discussing the tropopause shift between ERA-Interim and ERA5. This shift is mainly due to the higher vertical resolution of the ERA5 with 3 times more levels in the TTL. The improved hozizontal resolution in ERA5 is also inducing the tropopause to jump up and down in convective regions. It is perhaps not meaningful to define a tropopause from WMO criterion at such high resolution. Calculating the tropopause on horizontally filtered data is more reasonable. In prcatice 1° is a good compromise. However, using a 1° subsampling, as I suspect to be the case here, does not removde the problem.

31) 4.5 Defining the level of maximum CO2 in the TTL is the only usage of CO2 here. The sentence on l.392-394 is badly written but also wrong. The transport properties from the ground have not been uniform over the range of time from the pre-monsoon to the beginning of August and the CO2 profile is a convolution between the source variation and the changing transport. This is exactly what is done in the CO2 reconstruction discussed in 4.6 and therefore this naive interpretation is totally at odd.

32)  4.5 l.400. Is it not true that the reconstruction up to 400 K is due to NTL and so, what it doing CL here.

33) 4.6 It is not clear from the reconstruction shown in the bottom row of Fig. 14 that ERA5 should be preferred to ERA-Interim. Both are good below 400 K but above this level ERA-Interim is better and with less dispersion. The reconstruction below 400 K is based on young trajectories and on the CO2 cycle over north India while the reconstruction above 400 K depends on a much wider distribution of sources, both in time and in space. The result is somewhat infortunate and of course cannot be used to support a general conclusion that ERA5 is better but thre were perhaps here too much problems to solve at once.

34) 4.6 In both rows of Fig.14, there is a sort of jump at 400K. Is it a feature or a technical issue ?

Full references are found in the manuscript