See the attached file.

General comment:

This paper addressed the important problem about the modeling capability of ice cloud radiative forcing in climate simulations. An irregularly shaped Voronoi ice cloud particle model which was proven to be effective and efficient in satellite remote sensing retrieval purposes has been implemented in the RRTMG RTM and CAM5 climate model. Comparisons of modeling results with the Voronoi model along with the other four previously proposed ice cloud models are carried out. Further comparison between model results and the CERES SYN1deg radiative fluxes indicates that the Voronoi model provides the closest cloud radiative forcing to observation. This study could be a good supplement to understand the influence of ice cloud optical properties on simulated cloud radiative effects. The topic of this paper is within the scope of the journal of Atmospheric Chemistry and Physics. But unfortunately, the paper is not acceptable in the present form due to various issues.

Major questions:

1. Overall, the quality of this paper does not meet the standard of ACP. The problem is all-round, from simple wording and sentence expressions to the quality of figures and tables, data analysis, conclusions, and so on. There are so many things to be improved to make the paper better (please see details below).
2. The authors should be cautious about the definition and the use of abbreviations. Several abbreviations are defined again and again. While some other abbreviations are defined and never used again. Abbreviations like AGCM are used without definition. This may be just subtle issue but it could be an indicator of how the paper is carelessly prepared …
3. The authors need to pay more attention to the way to cite papers. Some of the names of the authors are wrong! For example, Line 47, “Hulst” should be “van de Hulst”; Line 68, “Labonnote” should be “C.-Labonnote”. Incorrect citation formats also exist, for example, at Line 61, 63, 83, 86, etc. This is another indicator that the paper undergoes insufficient examination before submission.
4. It looks like the “Baum-Yang” scheme in this paper is different from the “Baum-Yang” scheme in Wang et al. (2018). It may be better to rename the schemes to avoid confusion when the readers are comparing the two studies.
5. Among the various schemes, Fu scheme actually has different definitions of effective diameter (see Fu et al., 1997). So the question is how can the Fu scheme be compared directly with the other schemes?
6. Line 89-92: What’s the point of mentioning CIESM at this point? Since CIESM is no different from CESM regarding the ice cloud scattering properties, there seems no need to mention it at all. After all, the authors are actually using the original CAM5, isn’t it?
7. I don’t like the way the authors organized the figure panels. It’s strange to me to use panel a1, a2, … and b1, b2, … in a same figure. Please consider following the conventional panel naming habit of (a), (b), (c), …
8. I don’t like the organization of section 3 either. Particularly, Line 140-159 is a mess. It may not be a good idea to briefly referring to something you will mentioned in detail later. It makes no sense and just add to the confusion of the reader.
9. More details about the particle size distributions should be given. The authors may add a figure to show how the PSD looks like.
10. Why do the authors choose to use CERES_SYN1deg_Ed4A products for comparison with GCM modeling results? What is the temporal range of data used? Usually the CERES EBAF dataset is a better choice for this purpose. Since the authors’ choice will apparently affect the evaluation results, it is quite necessary for the authors to elaborate the reasons more convincingly.
11. The authors should do a better job of relating the optical properties of different ice cloud models with the simulated SWCF and LWCF. It is still confusing to me that why the Voronoi model possesses the lowest asymmetry factor in the SW but however exhibits the lowest SWCF compared with the other models? In short, why the Voronoi model could be the better choice?
12. Many grammar mistakes and sentence errors could be found in the manuscript. The authors should pay more attention to polish the English language. Several captions of figures and tables also need to be rephrased. For example, Table 3 and Figure 9 all miss the units. While caption of Figure 5 is too complicated to understand.

Minor questions:

1. Line 17-18: “While abundant irregularly shaped ice particle habits present a challenge for modelling ice clouds.” – please be clear about what challenge is the authors referring to.
2. Line 24: There may be no need to express the names of the other four schemes since they are not used again in the abstract, and the readers still could not understand what the names are referring to.
3. Line 25, 27, 30: RRTMG, CERES, SW and LW are never used again in the abstract. May not need to define the abbreviations.
4. Line 91: “CAM5 in CIESM was modified with several new schemes”- what are the changes which relate to this study?
5. Line 101-105: Very complicated sentence which contains error.
6. Line 109: “has proven” should be “has been proven”
7. Line 138: It’s very odd to see equation (1) here without any explanation.
8. Line 148: What is the temperature used in the Planck function?
9. Line 157: “to validate the cloud radiative properties” – do you mean “cloud radiative forcing”?
10. Line 166-167: Sentence error.
11. Line 197-198: What is “standard tropics”? How are the 60 vertical levels distributed? Please give a reference.
12. Line 201: “the same with” should be “the same as”
13. Line 242: should be Zhao et al. (2018)?
14. Line 264-266: Please change a way to express the range of values since the present form easily cause confusion.
15. What is the version of RRTMG used?
16. Line 320: The authors may need to specify the contribution of all authors.
17. Line 513: “Ice particle effective size” – please be specific. Is it diameter or radius?

General comments

This study focuses on the comparison of Voronoi model with four other ice cloud models. For the validation purpose, authors used CERES data. Authors conclude that Voronoi model based results are closer to CERES data than results obtained from other cloud schemes. The overall goal of the study looks interesting; however, the paper is poorly organized with several mistakes in English writing, literature reference, equation citation, and so on. The discussion part is also poor. 

Specific comments

1. Figure 1 shows single scattering properties of only Voronoi model, though Figure 3 shows band averaged values for all cloud models. Why not to show the single scattering properties for all models in Figure 1? It can make easy to understand Figures 3 and 4 as well as other results.
2. It may be better to show the difference in terms of percentage (relative values) in Figure 4.
3. There is an unclear description about particle size distribution (PSD) of ice clouds. Authors state that they utilize 14408 groups of microphysical data. Do authors use a single or multiple PSD function in this paper? For clarity, it is important to describe how PSD function is derived from 14408 groups of data to use in this study. If possible, PSD is suggested to be shown. If not possible, authors may tabulate the parameters of PSD function(s) used in this study.
4. Figure 2 is not clear. It may be removed or improved. The methodology is well understood even without Figure 2.
5. Equations are described in the text very randomly. For example, in section 3, Eq. 7 is described after Eq. 2. Equations and Figures are needed to appear in the text ascending order.
6. What is the necessity to integrate over wavelength in Eq. (7) ?
7. Eqs. 11-13: Equations corresponding to long wavelength bands need to be rewritten or a symbol to represent S and J may be used and stated below those equations.
8. Authors state that the wavelength range is from 0.2 micron to 15 micron for Voronoi database and they assumed unchanged properties for wavelength larger than 15 micron. What about database for other cloud models? Do they also have such assumption? If such assumption is only for Voronoi database, what are the effects in results shown in Figure 3 and onward?
9. Authors discuss about cloud forcing in Eq. (14). Can authors also discuss about the comparison of downwelling and/or upwelling fluxes for cloudy scenario between CERES and each cloud scheme? I guess comparison of fluxes rather than cloud forcing may help to better understand the performance of each model.
10. How water cloud is treated here is not clear. Authors may describe a more about water cloud properties and how they are merged with ice clouds in the simulation. Authors may add information (height, properties etc) related water cloud in Table 2.
11. What is Downward flux in Figure 5. Is it Direct+Diffuse flux? please clarify. If it is Direct+diffuse flux, why is it largely different than Difdown flux for a cloudy condition? Are clouds optically very thin for such difference?

The paper by Li et al. presents an analysis of a proposed broadband ice cloud scheme based on the Voronoi ice cloud particle model. The comparisons of model simulations using RRTMG and CAM5 between Voronoi and other four ice cloud schemes were carried out, indicating that the Voronoi scheme is superior to the other conventional schemes and should be sufficient for ice cloud modeling. I believe this study can be valuable to the relevant community, and it helps to better understand the ice cloud optical properties and their impact on cloud radiative effects modeling.

Overall, the study established a straightforward objective and was done in a comprehensive way. The employed scheme seemed valid and the extensive comparison was performed and discussed properly. The drawn conclusions are in line with the experimental results. From my point of view, the paper is suitable for Atmospheric Chemistry and Physics, although I do have some concerns that need to be responded.

To enhance the potential of the proposed scheme, I would encourage the authors to submit a revised manuscript by addressing my specific comments below:

• As pointed out by the other reviewers, the English language of the current manuscript requires a substantial improvement. There are a number of grammatic and wording errors (not described here as most of them have been noted by the other reviewers) in the article. A careful proofreading throughout the manuscript would be necessary.
• Please check Equation (1) at line 138 since the current layout seems weird.
• Please consider to revise Figure 2 as the flowchart does not look very helpful to me. If possible, please also include a short overview of Figure 2 in the begining of Section 3 or reorganize this section, particularly the first paragraph. Here, you do not have to provide equation indices since you will detail them in the following subsections anyway.
• Line 151: In Section 1, you actually only introduce the four conventional ice cloud schemes without sufficient (mathematical/technical) details. Readers would expect more details from Section 3. So this could be another point to reorganize Section 3.
• Section 3.1: I am okay with the contents. However, I would like to see a clearer structure. Each equation should normally follow the corresponding text.
• Line 269: It sounds unclear to me based on what quality criteria the authors ranked the five models.
• Lines 293-296: Please explain Figure 9 in detail, more explicitly, why the Voronoi model performed the best. So far I am not convinced by the statement in the current manuscript "...differences box of Voronoi scheme are most concentrated on the zero ...".
• An additional appendix including all acronyms and abbreviations used in the manuscript would be useful to readers.