Dear authors,

please find enclosed the report from referee 2. Unfortunately, the referee was not satisfied with the revisions of the manuscript. The major criticism from the referee is that the simulated HNO3 distribution depends critically on the other parameters which are still not properly addressed in the your work and that this forces you to "speculate" about underlying causes of the discrepancy between observations and model. The referee also requests, as already done in his previous report, more investigations, e.g model simulations. Further, referee 2 states that “as the work stands almost nothing else has been taken into consideration in the simulations and therefore that only one "consistent explanation" (for which there no actual physical evidence) has been explored”.

In my opinion the criticism by the referee is justifiable. However, I do not think that you necessarily need to perform further model simulations, you rather should discuss more the possible influence of the other factors and state more clearly that you have one theory here that is tested. Some more detailed comments will follow below:

Abstract:
How does the reduction of the settling velocity agree with reality? Later you explain it, but in the abstract the explanation is missing and would be worth to be included here as well. Other questions that came to my mind when reading the abstract was: Are the calculated settling velocities wrong so that one needs to reduce them? Isn't what you are doing just another way to “tune” the model to the desired results? Doesn't the discrepancy just mean that the size distribution of the the NAT particles derived from the model is incorrect due to the processes that have not been considered as stated by referee 2?

P4, L21-L25:
Do I get it right that you use the reduced settling velocities to overcome the problem that in the model spherical particles are assumed though the particles in reality can be aspherical?

P7, L23:
In the model one of the latest parameterizations for NAT nucleation is considered. However, in the model no ice formation is considered and thus you cannot consider formation of NAT on ice or vice versa NAT on ice which also could lead to the large particles measured by the FSSP. I do not expect you to include ice formation in the model to calculate other possible formation mechanisms, but you at least should discuss these processes and the possible influence they could have on the results of your study. Further, other studies discussing PSC formation during that winter (Khosrawi et al., 2011; Pitts et al., 2011) show that ice was frequently observed during mid of January and that a warming occurred around 24 January, so that the probability for ice on 25 January may have been low.

P8, L27:
I am quite skeptical that the difference in the settling velocity of spherical and aspherical particles can cause such a large difference between measurements and observations.

P11, L21:
The denitrified air masses above have not been measured by MIPAS-STR and this should be clearly stated. That denitrification/renitrification occurred during that time period is shown e.g. in Khosrawi et al. (2011) in Figure 4 using measurements from MLS. It would be useful to refer to this figure to confirm the measurements by MIPAS-STR.


P17, L19:
What are the growth rates of spherical/aspherical particles? Which of these particles grow faster. Can NAT particles really grow to such large sizes as measured by the FSSP? These have been larger than the ones by Fahey et al and such sizes have not been measured before. How do you know that these particles are real and that these are really NAT and not ice particles?

Figure 1 and model/observation comparison:
What does the FSSP measure NAT particles or all type of PSC particles? How can you be sure that the large particles measured by the FSSP are NAT particles? What are the uncertainties of the FSSP? There is a large discrepancy between the size distributions and this discrepancy cannot be solely explained by the differences in the settling velocities of spherical/aspherical particles. Other processes must play a role.

Figure 7.
This figure should also be included for the ClaMS simulations with reduced settling velocity.

Figure 8:
From this figure I would say everything except a settling velocity of 30% looks like a good agreement between measurements and model simulations. However, you conclusion is the opposite.

Best regards, Farahnaz Khosrawi

Dear authors,

thanks a lot for the effort you have put into the revision of the manuscript. Both, the referee and I, agree that you have considered our comments and suggestions for revision satisfactory. There are only some technical corrections left as given by the referee which should be considered before publication in ACP.

Best wishes, Farah