This work uses a novel method to detect tropopause especially the double tropopause and shows the statistics of the tropopause using this new method. The definition of the tropopause is an important question in the upper troposphere lower stratosphere dynamics and chemistry, and the information presented in this paper is very quantitively and clear. However, the discrepancy between this new method and existing method seems to be too large and I suggest the authors show more clarification in section 3, please see major comments for details. I suggest accepting this paper after resolving this issue.

Major comments:

1. Figure 8, and lines 281-290. This part needs more detailed discussion and clarification:

Line 281-282: “of which the largest proportion (11384 profiles) is identified as DT by bi-Gaussian method but ST by LRT”, this result is astonishing. According to the WMO double tropopause definition, ‘If the average lapse rate above this "first tropopause" between any level and all higher levels within 1 km exceeds 3°C/km, then a "second tropopause" is defined by the same criterion as the first’,  I checked the significance test in section 2 and only find that “the slope is not less than 0.5°C/km”, is it a reason for this very different result between the bi-Gaussian method and LRT?

Figure 8a: I suggest also showing ratio (contradictory results/ all observation of this station) in addition to population.

line 287: ‘peak at 0.5… peak at -0.5’: I find it difficult to understand this normalized height. Does this mean that for all first DT, the altitude is -0.5, and for all second DT, the altitude is 0.5? Could you please add more explanation to the text? Or maybe do not define ‘0.5’ and ‘-0.5’, just use text ‘first DT’ and ‘second DT’, because 0.5 maybe misleading, looks like 0.5 km.

Line 288 ‘the bimodality become gradually unobvious poleward’: in figure 8a, it looks like that the DT events happens more frequently at higher latitude, but the bimodality is not very clear? Could this be a result of overestimating DT events over higher latitude? Could you please add more discussion on this result?

1. In addition to the tropopause definition described in this work, in the subtropics, dynamics tropopause based on PV or PV gradient is also widely used to detect the transport between stratosphere and troposphere. This paper also describes PV field in section 5, and shows a good agreement between the PV field and the tropopause. It will be interesting to add discussion regarding the dynamics tropopause.

Minor comments:

Figures 1, 9, 10, 11: please add longitude and latitude to your map

Figures 6, 8, 9, 10: text is blur

Line 9-11: I would say ‘the tropopause is an important transition layer, and can be a diagnostic of .. ’

Line 23-24: ‘in the latitude range [16°N, 50°N] ‘ and ‘at mid-latitudes [30°N, 40°N]’: please use the same format. This problem in also in lines 300-308.

Line 41: ‘, climate model simulations’: replace ‘,’ into ‘and’

Line 58: ‘buoyancy frequency N has been introduced’, not ‘introduced’ because it is already defined earlier. It is just ‘used’.

Line 66: ‘from different monsoon circulation systems, such as the Asian summer monsoon and polar vortex..’, polar vortex is not a monsoon circulation system

Line 101: ‘as described in detail in the literature (Guo et al., 2016)’

Line 102: ‘representing an excellent opportunity’, sounds wired, consider use ‘providing an excellent opportunity.’

Table 1: incorporate the most important information in the text. For example, mention that the vertical resolution is 5-8 m when saying ‘higher vertical resolution than reanalysis’.

Line 110: ‘there are 37 vertical layers from 1000 hPa to 1 hPa’: what is the vertical resolution over the levels of interest in this paper?

Line 119: between the upper (stratosphere) and lower (troposphere) parts: this is ambiguous, consider just say between the stratosphere and troposphere, or stratosphere (upper altitude) and troposphere (lower altitude).

Line 125: ‘in view of’ this sentence is blur. The three sentences in this paragraph are: ‘in view of the fact.. in addition’ are both a description previous works, but ‘ next’ follows by the work in this paper. This logic flow does not make sense.

Section 2.3.1 the title is ‘tropical tropopause layer’, but in figure 2 it is a subtropical station  

Line 326: ‘Bi-Gaussian method prefers to define higher and colder temperature inversion layer as DTH2, which leads to an increase in the occurrence frequency and thickness of DT’: could you explain more about this logic chain?

Line 344: ‘tropopause height has an increasing trend under global warming’ I doubt this value may not be very large, could you please quantify it according to the estimation in your reference?

Line 345: ‘the change trend’ ‘trend’ is misleading with the trend in the last paragraph. It sounds like a trend changing with time.

Line 365: ‘(Buchart, 2022)’ this citation is not included in the reference. I didn’t check the full reference list and the authors should double check it to make sure this mistake does not happen again.

Section 5 : I suggest use another name instead of ‘discussion’, since this is an important part of this paper, not a discussion.

This paper proposes a new method to identify the tropopause using a “bi-Gaussian function” that identifies local minima in temperature within a profile. While exploration of novel methods to improve definition of the tropopause is a worthwhile effort, I unfortunately found the present study to be critically lacking in myriad ways which I elaborate upon below. Some of the critical shortcomings could be resolved by improved narrative and discussion and others require more extensive analysis and demonstration of the proposed method (or an alternative).

General Comments

First, the fundamental requirement of any tropopause definition is that it provides a demonstrated reliable identification of the troposphere-stratosphere transition layer. Existing definitions have been demonstrated to do this in myriad ways, with composition observations being the best utilized for such demonstration. The present study does not demonstrate that the new definition captures well the troposphere-stratosphere transition (or often more simply thought of as a boundary), with any such efforts limited to comparison with the WMO lapse-rate tropopause (LRT) definition. Even so, there is a surprisingly large number of cases where the authors’ application of the LRT or the proposed bi-Gaussian method fail to identify a tropopause. This result alone is surprising and questionable, as the LRT definition virtually never fails to identify a tropopause so long as a sufficiently deep profile of data that encompasses the upper tropopause and lower stratosphere exists. Perhaps the authors did not control for this in their dataset or perhaps their application of the existing, well-demonstrated LRT method is flawed. Regardless, the result that the proposed method fails to identify a tropopause in ~12.5% of profiles is a major shortcoming that is not addressed.

Second, the physical meaning and justification of a definition based on local minima in temperature is not clear and otherwise presumed to be weak. Past literature demonstrates thoroughly that cold-point definitions are not appropriate outside of the tropics and even in the tropics commonly result in identification of a level that is not dynamically or chemically relevant to the purposed use of such a definition – to accurately identify the bound (or transition zone) between troposphere and stratosphere air. Conversely, temperature minima near and above often result from convection and wave activity and can be an important failure mode for some existing definitions. To rely upon an error-prone basis of tropopause definition as local temperature minima provide is therefore highly questionable. Moreover, because the LRT has been comprehensively demonstrated to be reliable most of the time, differences between the LRT and any proposed definition solicit increased scrutiny of an alternative definition. It must be clearly demonstrated why an alternative definition is more reliable than the LRT or a similarly reliable definition (there are recent relevant studies not cited), otherwise the exercise presents simply a difference without explanation or significance.

Third, unless a definition is created to serve a very specific region or purpose, I consider global comparisons of a new tropopause definition with existing ones to be a necessary element of such a study. The narrow focus on China in this study is thus a major shortcoming given the aim of the effort.

Fourth, while I do sincerely appreciate the authors’ attempt to identify multiple tropopauses since only two proven definitions currently do so, the result that double tropopause occurrences increase from ~3% based on the LRT to more than 70% with the new definition is extremely concerning. Namely, as is true and necessary for any tropopause definition, an identified tropopause (primary or otherwise) in a profile should have an important physical or dynamical meaning. Otherwise, you attain nothing but vast identification of arbitrary levels that happen to have a local minimum in temperature. I do not expect the authors to demonstrate physical or dynamical linkages for their multiple tropopause definitions, but such have been well documented for double tropopauses that result from the LRT definition. The fact that you see such a tremendous increase casts serious doubt on the potential utility of such a definition, especially because it has also been demonstrated that multiple tropopauses identified by the LRT do not always have a clear physical or dynamical explanation.

Fifth, an important – though not mandatory – expectation for a tropopause definition is that its application is straightforward and not prone to confusion or misuse by others. The proposed definition is quite complicated, with many conditional steps that are likely to be inconsistently and inappropriately applied by others. Thus, simplification of the procedure should be a priority. Moreover, it is never specified what units are used for the conditional elements of the proposed definition, which are ultimately necessary for others to replicate this work in the future.

Specific Comments

Because of the substantial concerns I have with the design and execution of the study, I will not list myriad technical corrections here, but highlight some additional problematic statements or impressions.

There are multiple recent efforts to develop tropopause definitions that are not acknowledged or cited. There are also many other contextual works that would help greatly in the presentation, framing, justification, and discussion of such work. I encourage the authors to dive deeper into literature review to improve upon these issues, which will help direct future efforts towards accomplishing this study or another iteration.  

Lines 34-36: the tropopause does not perform a role in stratosphere-troposphere exchange (STE), but its definition is required to assess it; dynamic mechanisms are the role for STE.

Line 39: there are many other (and increasingly comprehensive) studies of the tropopause and its relation to climate change that are not cited.

Line 60: should acknowledge here and elsewhere that this “cliff-like decline” is broadly recognized as the “tropopause break” and cite additional work.

Line 77: “subject to controversy” is overstated. It would be better described as “active areas of research”

Lines 94-96: this is not appropriate motivation for the use of radiosonde observations. Radiosonde observations are the traditional and most widely used data for studying the upper troposphere and lower stratosphere and defining the tropopause.

Line 102: this is also a very unusual introductory statement and motivation. There have been multiple well-cited studies that demonstrate why double tropopauses are frequent in the midlatitudes.

Lines 106-117: this is presented suddenly and without explanation of its significance and intended use.

Line 119: this statement is not true in multiple ways and is contradicted throughout the article. Most existing tropopause definitions have been demonstrated to be chemically, physically and/or dynamically meaningful. At least two existing definitions have been demonstrated to be universal – the LRT and the recently-developed potential temperature gradient tropopause (PTGT) definition.

Figure 2: are the lines in panels (b)-(d) averages? This analysis is not well explained or described.

Section 2.3.2: there are several issues here. First, it is presented as though only the Brunt-Väisälä frequency is used for tropopause definition. Second, one curve-fitting method from a single study (Homeyer et al. 2010) is used without explanation that such is the source.

Line 183: TH is not defined and is difficult to follow its meaning here and after.

Line 225: extremely overstated. A high correlation for the fitting process does not demonstrate potential for accurate tropopause definition, but rather that you have success at identifying local temperature minima. 

General Comments:

This paper is an interesting study and methodology of the definition and computation of single and multiple tropopauses (TPs). Although the authors have made intensive investigation with the excellent radiosonde measuring net in China, however, the results of the bi-Gaussian fitting method (BGF) are not convincing for me and with respect to the publication requirements of ACP.

I have listed below various items regarding the analysis and especially the form of the presentation of the study, which needs further improvements. If the authors consider most of the comments in a revised version, the article may be acceptable for publication in ACP.

In general, I have concerns about the quality of the BGF method. The study misses a validation of the tropopause results with respect to frequency (double TP events) and especially height of the tropopause (TPH) with independent measurements (e.g. GPS occultation) and methods. Although, this is partly done in Fig. 6, I was a bit puzzled that later, differences of >1km between the tropical TPH of lapse rate TP and BGF are described with ‘small’. It is already obvious from former studies that the LRT is usually placed below the cold point in the tropics. So, why do you compare apples and oranges? Consequently, I was a bit surprised that Fig.6 shows no clear indication for a positive bias (are most of the profiles not really in the tropics?), but many TPHs are quite high (>17 km), which looks very tropics-like. However, a closer look seems to show such a ‘positive’ bias in Fig. 6 for STH/DTH1 compared to LRTH1. This fact is not discussed properly in the manuscript with respect to different definitions of both TP methods.

Detailed Comments:

L11: ‘physiochemical’ unusual wording, please change.

L15: ‘in mathematical statistics’ not clear to me why this term is necessary.

L37: ‘stratosphere vis this “gate”’ Is vis really the correct wording here?

L40: delete ’in’

L46: ‘concept of the dynamical tropopause’

L55: lapse rate minimum tropopause  (LRM)

 L57: gauge -> estimate

L59: what do you mean with ‘ideal models’? please clarify.

L74: ‘key stratification’ sounds misleading to me.

L128: Is there a lower boundary of the tropopause? Please, clarify.

L129: ‘four’ I count only three TP definitions (LRT, CPT, and N^2). Please, clarify.

L143: delete ‘And’: The cold point ...  

L166: what is different? Please be more specific with your statements.

L167:  please correct, ‘close to the CPTH’

L171: ‘highly effective’ for what? Do you mean the methods?

L172:  … in the extratropics …

L176: DT, you may have to introduce DT not only in the Abstract but also in main text.

L200-215: How do you handle triple structures of the TP? Is the method robust, does it detect the upper or lower 2^nd TP?

L220: The parameter of the formula of Table 3 are frequently used in the manuscript. Consequently, they must be introduced in text and not in the table, as well as a more detailed description is necessary.  

L230ff: I cannot follow the arguments on R^2 and why this number should give me confidence that the TP is detected correctly. It’s just the quality of the fit. It is necessary to check the quality of the fit.

L254: ‘darkest patches’ ? Red is not dark compared to blue. ‘The majority of the events are located on the …’

L274: Please reword the sentence. It is not clear to me what you loke to say. Why is a threshold critical for an accurate result of the TP? It’s part of the definition.

L282: by the bi-Gaussian method, but only ST by LRT .  

L280ff: I cannot really follow the description and conclusions of Fig. 8. I would suggest writing the whole section and caption new. More details on the methods are necessary. Why are both TPHs constant on +/- 0.5 units? The normalization is not really described in detail and difficult to follow. The arguments with R^2 are again very confusing.  

L300: new section and subsection, please introduce ‘the occurrence frequency’ of what kind of parameter?

L307: Please rewrite this sentence ‘The thickness …’. I can’t get a handle on the terms ‘latitudinal plain’ and ‘giant topography’.

L333-344: The discussion is misleading. It is always clear that CPT and LRT will not deliver the same tropopause height due to the definition of both parameters. In the tropics there should be an offset, and this becomes obvious in your Fig. 6. Of course you can show these comparisons, but it is no proof about your TP determination, because the comparison works with ‘apple and oranges’. You may quit this part.

L352: Not the TP is a source of gravity waves but processes in the TP region trigger GW formation.  

L364: Here went something wrong ‘atmospheric dynamic processes …’, please reword.

L382: ‘… and high static stability of the air masses creates …’

L437: delete ‘which is more than half …’

L432: Is TT1 introduced before?

L435: and increases DTT2.

L437: ‘… intensifies the atmospheric mixing’ may be better.   

L442: I have doubts that the argumentation with R^2 makes sense, especially in the conclusion section (see above and concerns by other reviewers).

L449: Again, I cannot follow the argument ‘… ambiguity of LRT constrained by thresholds.’. The bi-Gaussian method is not constrained by thresholds but by the bi-Gaussian fit approach and the quality of the fit, which is also very likely a threshold criterium.

L470: formatted -> formed

Technical issues:

Most of figures show a lack in resolution, which makes it difficult to read numbers and figure legends. For publication this needs definitely a substantial improvement (Fig 1, 5, 6, 7 -10)

Fig 3: please enlarge the figure and especially the font size. What do mean with ‘Modal’? This is not used in the text, please change this term.

Fig 3b: What is the red sub-plot in (b), Temp versus Altitude. This is not explained in the Figure capture and makes no sense to me at all.  If possible, just delete it.

Fig. 5: Fonts are far too small!

Fig. 7: please, enlarge the text fonts (e.g. dT/dz). In addition, there seems something wrong in the wording ‘Case A indicates that presents …’. Could it be better: ‘Case A indicates the presence of a higher …’  

Fig.10: For me it would be better to use identical TP height ranges for all three TPHs. The color code is misleading, e.g. why should STH be higher than DTH2, but it is just the color code?  Again, the resolution of the figure is not good enough. It is not possible to read all the letters and numbers properly.

Fig 12a: Is this PV plot presented for a specific theta level? If so, please add this important information.