The manuscript by Beer et al., 2022 presents simulated ice nucleating particles and their effects on cirrus cloud properties in a global chemistry-climate model. As a difference from other similar studies, they include two ice nucleating particle types that are rarely included in model simulations: crystalline ammonium sulfate and glassy organic particles. They find that ammonium sulfate plays a large role in nucleating ice at cirrus conditions, of larger or at least comparable magnitude compared to dust aerosols.

This is a robust scientific study describing the results of careful extensions to the model freezing scheme. However, after reading it through I think the paper reads too much as a report on the (otherwise excellent) work, missing a sharper line of thought including more discussion on the main novel outcomes of this work and sharper conclusions. Because of this, I was initially not certain what is the key novel result of the manuscript.

I would therefore suggest bringing the large importance of crystalline ammonium sulfate for cirrus freezing (which is in my opinion the main novel result of this work) more to the forefront of the manuscript.  Indeed, this may be a somewhat controversial conclusion, that demands more discussion and (if possible/available) more evidence beyond your modeling work.

Secondly, the manuscript falls short of pointing at the radiative importance of the separate ice nucleating particle species. I think the manuscript would significantly benefit if the authors could add the analysis of the radiative relevance of implemented INP species for the top-of-the-atmosphere radiative budget.

The manuscript will be suitable for publication in final form, once the above-mentioned and other comments are addressed.

Additional general comments

1. Is there a way to find observational evidence of ammonium sulfate playing such a large role (outside of lab studies)? This would go against the generally accepted notion of dust as the key ice nucleating particle at cirrus levels (e.g. Cziczo et al., 2013, Froyd et al., 2022, and many more).
2. What is the climatic role of different ice nucleating modes? Is the addition of ammonium sulfate as separate ice nucleating species as important also from the radiative perspective?
3. Could you comment on the uncertainties in ammonium sulfate ice nucleating properties, e.g. in its onset ice nucleation temperature, critical supersaturation, etc? As an example, a recent study by Bertozzi et al., 2021 found the onset of ammonium sulfate freezing at -54°C at a critical supersaturation of 1.3. Such assumptions would probably significantly reduce the role of ammonium sulfate as ice nucleating particles compared to the used assumptions.
4. Detrainment of ice is one of the key sources of upper tropospheric clouds and likely accounts for most of the upper tropospheric clouds in the tropics below about 14 km and a large fraction of the extratropical summertime high clouds. How does detrainment affect in-situ cirrus and their ice nucleation? How are the ice number and mass sources of detrainment treated, and does detrainment interact with in-situ ice nucleation?

Specific comments

Page 1, line 2: the word climate modifications alludes to artificial geoengineering-type of modifications. You could consider using a synonym.

Page 1: The last sentence of the abstract describes (in my opinion) the main result of the manuscript. The reader should ideally be informed of the key result earlier than at the end of a relatively long abstract.

Page 7: It seems like you have put a lot of effort into numerically representing and due to the need for 3 additional tracers also computing ammonium sulfate aerosols. Is there a simpler way of simulating crystalline ammonium sulfate that the other modeling groups may be more likely to implement in their models?

Figure 1, panel (a): Surface or 300 hPa level?

Figure 5: panel (c ) should use the same colorbar limits as all other panels, not to artificially overemphasize the role of aircraft soot.

Page 19, lines 414-117: The two sentences seem to be contradicting each other (the first one claiming the occurrence frequencies of heterogeneous freezing are largely preventing homogeneous freezing, the second one saying that heterogeneous INPs do not completely suppress homogeneous freezing)

Page 19, lines 429 and further: I agree INP measurements at cirrus conditions are practically non-existent, but you may still want to comment on the surprising result that ammonium sulfate is found to be the key INP species at cirrus conditions, contrary to the mainstream cirrus literature.

Page 21, lines 477-484: You don’t comment at all about the surprising result of the large importance of ammonium sulfate! From this paragraph, it sounds like dust is clearly the main INP, which is different from your model results.

Page 22, lines 497-500: I would suggest modifying for clarity the italicized part of the sentence: “….this study demonstrates the importance of including additional ice nucleating particle types together with…. “   to ammonium sulfate

And then just conclude with “Glassy organic particles probably have only a minor influence…”

References

Bertozzi et al., 2021, doi: 10.5194/acp-21-10779-2021

Cziczo et al., 2013, doi: 10.1126/science.1234145

Froyd et al., 2022, doi: 10.1038/s41561-022-00901-w

In this manuscript the distribution of ice nucleating particles (INPs) originating from different sources is investigated using a sophisticated aerosol module for a GCM (EMAC-MADE3). In addition to the usual candidates of INPs, glassy organic aerosols and crystalline ammonium sulfate are also taken into account. The authors find that ammonium sulfate surprisingly plays a major role in terms of providing the most INPs. The INPs are also coupled to a cirrus cloud module in the GCM for investigating the competition of heterogeneous ice nucleation and homogeneous freezing of solution droplets in the upper troposphere.

Generally, this is a very solid and well written manuscript, the implementation of the different species into the model is explained in a comprehensive way and the results are well described. Therefore, this is a suitable contribution to ACP, and I recommend to accept the manuscript subject to minor revisions (see below).

(1) The two major outcomes of the study are not really prominently stated. It was claimed for a long time that glassy organic aerosols are an important class of INPs and they must have an impact. It is quite clear from this study that the impact is rather weak, if not negligible. The dominant role of ammonium sulfate is also new. I would suggest to emphasize these interesting results more prominently.

(2) For aerosols, the large scale transport is the most important pathway, thus the distribution of aerosols can be simulated well with GCMs, even in this coarse resolution of T63. However, for treating (ice) clouds in GCMs, the variability of thermodynamic variables plays an important role. The ice cloud model relies on former work for the EMAC model, which only marginally treats subgrid scale variability, e.g., using TKE or gravity wave drags for determining the variability of vertical velocity; however, the horizontal/vertical variability is only taken into account by crude cloud cover schemes. From observations (satellite, surface observations, and many others) we know quite well that also ice clouds have internal structures, leading to heterogeneous cloud layers, which provide additional spatial and temporal time scales. The use of microphysics schemes for the whole grid box, driven by a large-scale motion, without these scales in between might lead to an overestimation of the impact of nucleation pathways. A similar issue might occur for the removing of INPs (as included in ice crystals) by sedimentation of ice crystals, since sedimentation of ice particles in quite coarse vertical resolutions is highly tuned. It is quite clear that the authors cannot change the coupled ice cloud scheme. However, I would suggest to add some comments on these issues, since the competition of heterogeneous and homogeneous nucleation might be affected by small/meso scale motions.

(3) How sensitive are these results if the nucleation thresholds for the different INPs are changed (within their uncertainties)? Have you checked this in sensitivity analyses?