Both referees consider the paper to have been improved by revision but both continue to be concerned by aspects of the paper and neither is yet willing to recommend publication. To me the most straightforward issue is raised by the Referee 1, that the K_yy (isentropic eddy diffusivity) fields that you shown in what is now Figure 1 differ significantly from the expected structure (and indeed the structure that you describe).

You say: "Figure 1 shows the seasonal mean isentropic diffusion coefficient Kyy (l=1 day) (hereafter referred to as Kyy). Large Kyy values reveal strong isentropic mixing in the stratospheric surf zone in both hemispheres, but is stronger in the NH than in the SH. The hemispheric asymmetry can be attributed to the differences in planetary wave activity (e.g., Shepherd et al.,
2000). The small Kyy equatorward of about 30° is indicative of a barrier to horizontal transport between the tropics and mid-latitudes in both hemispheres (Plumb, 1996). At SH high latitudes, Kyy becomes large after the break-up of the Antarctic polar vortex (figure not shown). Strong cross-tropopause eddy mixing is present in the subtropics and mid-latitudes,
poleward of the subtropical jet stream, whereas the eddy mixing is strongly suppressed near the core of the subtropical jet stream."

But the fact is that very little of what you describe matches the Figure -- K_yy is largest in the jet core, not in the 'surf zone' and the structure in the UTLS region does not correspond to K_yy large poleward and equatorward of the subtropical jet and small in the jet core.

Referee 1 demonstrates that a relatively simple calculation of K_yy gives much of the structure that you describe (to be contrasted with the structure that can actually be keen in your Figure).

Please consider in your revision how to resolve the above problems and provide a suitable response and make appropriate changes.

Referee 2 on the other hand is concerned with the presentation in Section 2.2. Please consider Referee 2's comments and respond appropriately. I wonder (but am not certain) if part of the issue is that Referee 2 has not realised that you are using a non-standard formalism (the MIM formalism) -- which, of course you and others have constructed and applied in several previous papers. You need to make that clear and provide enough detail to allow the reader to understand how the calculations have been done. On the other hand I do not feel that it is appropriate here to go into too much detail on the pros and cons of the MIM formalism -- that can be dealt with by referring the reader to previous papers.

One referee now recommends publish as is, the other (who was the referee who pointed out the significant problem with your K_yy fields and who has taken significant trouble over refereeing this paper) recommends publication after minor revision and provides a long and detailed list of recommended corrections. Please can you provide a revised version of the paper in which these corrections are made (or justification is given when there is not). I hope then to be able to accept the paper without further consultation with the referee.

I am pleased to accept this paper for publication in ACP. But in checking some of your corrections in response to the list Referee 2's last report I see that not all on that list have been implemented.

Can I draw your attention in particular to:

Figure 6 -- the referee suggested that you added the latitudinal range to the title of each panel, as was done for Figure 2. You haven't done this. I think it would be a helpful clarification -- can I ask you to consider again?

Table 2 -- you say 'The linear trend with a confidence level greater than 95% is shown in bold.' -- but as far as I can tell NONE of the numbers in Figure 2 are in bold.

There may be other small improvements that could be made. Please check the paper one more time before providing the final version.