|The revised manuscript has addressed many of my minor concerns, but my principal criticism has not been addressed adequately yet. The manuscript provides valuable observational evidence for an inverse relationship between aerosol and mixing height, but this is not sufficient to demonstrate the presence of a feedback without further supporting evidence.|
In their response, the authors reassert the existence of the dynamical feedback. The theoretical basis for this is sound, but the potential strength of this study is in providing observational evidence for it. The authors are too quick to demonstrate that the observations are consistent with the theory, and neglect the much more valuable goal of testing the theory based on the observations. This is evident in their response to my request for justification for fitting an exponential curve rather than a reciprocal. The authors present an interesting analysis in the response to reviewers, but have not provided a justification or made any changes to the manuscript. What relationship fits the observations best, and what implications does this have for the theory about a feedback?
The conclusions of the paper remain weak, as noted in my original review, but have not been altered in the revised version.
As I commented in my original review, the final sentence of the abstract needs revision: most good air quality models have included this feedback for many years (albeit without strong observational support, which this study could provide) so it is too late to "suggest that the feedback mechanism should be considered". The study just reconfirms that it should be considered. The changes made to the final paragraph of the introduction (l.80-85) have improved it and now more accurately summarise the findings of the study, but the abstract does not reflect this yet.
L.183: As noted in my point above, the fitting of an exponential to Figure 3 still isn't explained. The new text at L.195 is not adequate to explain it. No reasoning or support is provided for ascribing the observed behaviour to a positive feedback (L.200-201).
The English language is generally reasonable, but needs further polishing before the paper is suitable for publication.
For an English Language journal it would be appropriate to replace the "D" and "G" in Figure S1 with "L" and "H" (the figure should be in English, not Chinese).
A number of the references remain incomplete, e.g., missing journal for Wang et al. 2018, missing details/date for Petaja et al. and Ding et al., and formatting errors for several other references.
The English in the title has been corrected, but in practice the title does not accurately reflect the topic or findings of the paper.