This study is a nice exploration of the utility of various tropopause definitions for characterizing the extratropical upper troposphere and lower stratosphere (ExUTLS). Both airborne field campaign and ozonesonde observations are leveraged and help to provide a comprehensive assessment of ExUTLS composition change relative to the tropopause. The authors find interesting differences in composition change and variability estimates in the ExUTLS for each tropopause definition explored and provide recommendations on their use for future studies. The analysis approach and diagnostics explored are largely robust, though I have some general comments on choices made and argumentation given that the authors should consider for revision. The quality of the writing and presentation of the results are excellent and I have identified very few required technical edits.

General Comments

• On utilizing ozone variability as an assessment tool for reliable discrimination of troposphere and stratosphere observations (i.e., Section 4.2): I'm not convinced this approach is a reliable assessment of tropopause definition performance. One particularly problematic condition to consider (which is not discussed at length in the present draft) is stratosphere-troposphere exchange (STE). Recent STE will undoubtedly result in air that chemically appears to be stratospheric/tropospheric but is located in the troposphere/stratosphere (respectively). Thus, using variability as an indicator of performance/appropriateness would seem to reward definitions that "remove" or otherwise minimize the inclusion of STE events. This complexity should at the very least be acknowledged and considered further in the interpretation of the meaning/significance of these results. 
  
  
  
  
• On the cross-tropopause gradient (i.e., Figure 7): using the magnitude of the ozone gradient across the tropopause as an assessment tool also seems problematic to me. One could achieve the largest gradient by simply introducing a high bias in tropopause altitude such that the ubiquitous sharp stratospheric increase in ozone contributes entirely to the diagnosed cross-tropopause gradient. I think the alternative approach of comparing (or rather, differencing) the gradients/slopes in ozone vs. altitude BELOW and ABOVE tropopause would be a more meaningful and reliable assessment of the appropriateness of each definition. That is nearly accomplished here, but this slight adjustment in approach would resolve an otherwise misleading means of assessing performance. 
  
  
  
  
• On the case study analyses in Section 4.4: this analysis was very brief and I found to be minimally convincing. In particular, where are the observations contextually, especially those in Figure 8? Is the high ozone portion of the flight flagged as tropospheric in panels (e) and (f) of Figure 8 the result of an STE event? For example, could this be a tropopause fold? To convincingly demonstrate appropriateness of these definitions for a case study, more information should be given. Though tracer-tracer analysis was mentioned early, I thought its exclusion from this study was a missed opportunity. 

Technical Edits

Line 9: "larger variability near the tropopause" would be better stated as "larger composition variability near the tropopause"

Line 107: "substances" should be "substance"

Line 151: "across the tropopause" would be better stated as "from troposphere to stratosphere in the extratropics"

Line 153: "the vast majority of" could be stated simply as "most"

Line 179: suggest revising "last decades" to "last several decades"

Line 304: I find the statement "a large number of" to be somewhat misleading. What is considered large? Could you express this as a fraction of all aberrations? From my interpretation of the analysis, it appears to have a minimal impact on the statistics, so that gives the impression that these observations are still few though certainly more numerous than remaining definitions. 

Line 313: "stratosphere" should be "stratospheric"

Lines 322-323: this sentence also appears to be an overstatement. The variability in ozone is not largest for the thermal tropopause in all seasons, though the range appears to be. In particular, the mode is largest for the N2O definition in autumn and winter. 

Line 338: delete floating paren after "Fig. 6b"