Overview

This manuscript describes the methodology of finding the optimal solution for a combination of parameters used in an analytical model of downburst outflows. This methodology is then applied to a particular case of downburst that occurred in Rumania in the summer of 2021. It is clear to me that the authors are knowledgeable in this field and the description of the method and results is very comprehensive. Perhaps even too comprehensive in some instances. While I have a number of comments below, I consider most of them to be minor. I believe the manuscript is well suited for this journal and it is most certainly a topic of importance in downburst research and wind engineering.

Specific comments:

1. Begin instead of begins.
2. Is the scale of less than a kilometer related to a downburst or thunderstorm? It should be related to a downburst, but it is not clear based on how the sentence is structured.
3. Xhelah et al. is not the proper reference. I suggest the authors write “Xhelaj et al. (2020) presented…” The reader is automatically informed that the research was published in 2020.
4. L46 and elsewhere. I believe that “et” should not have a dot.
5. L22–L72. This paragraph is well written, but it is too long. Please split this large paragraph into 2–3 smaller paragraphs to increase readability of this section. For example, L56 (goals of this research) can be the beginning of a separate paragraph.
6. I suggest rewriting this sentence as follows: …was produced during the passage of an intense mesoscale convective system in the form of a bow echo over the town of Sânnicolau Mare.
7. Some parts of the manuscript can be shortened. For example, the definition of thunderstorms in L100 is probably not needed. Even if needed, that general discussion should be in the introduction. Similarly, L103 and the difference between downburst and atmospheric boundary layer winds is probably not needed as well.
8. Figure 1. Labels in panel (b) should be of higher resolution and the red dot that indicates the tower location should be larger.
9. Figure 2. In principle, I have no problems with this figure but is it really that the researchers don’t have their own photograph of the tower and have to use Google Street View? I suggest replacing this figure with their own photograph.
10. Figure 3b. It is very interesting to note that the squall line seems to be stratiform parallel, which is one of the rarest types of squall lines. See Markowski and Richardson (2010) and references therein on squall lines. The authors, of course, do not need to pursue this comment further, it’s just an interesting observation from this reviewer.
11. Figure 5a. Tmin = 14.5 degC should not be over the line. There is plenty of space to move it elsewhere.
12. L171–174. Downburst is a wind event and hail is hailstones falling from the cloud. If hailstones cause the damage, that is not the same as caused by wind (i.e., downburst). One might rephrase this to state that the downburst was also associated with hail that caused substantial damage. Then the reader knows that the damage was not wind-driven but hail-driven.
13. Figure 6. Indicate the North direction in this figure.
14. Table 1. The units should not be italicized.
15. This sentence is the same statement as the previous sentence, which is that Figure 7 shows the convergence pattern of the objective function.
16. Figure 7. The mean convergence curve does not converge at about 70 iterations. Indeed the envelop curves seem to converge at about that value, but not the mean and standard deviation curves.
17. Remove the comma after 1024.
18. Symmetry rather than simmetry.
19. (2) The word and should not be italicized.
20. Probably an incorrect reference format. Please double-check.
21. Rewrite to “The hierarchical tree in Figure 9 (i.e., dendrogram) is constructed following the Wards’ method (Ward, 1963).” and then delete the following sentence because it contains the same information.
22. Figure 9. Indicate in the figure caption that the three colors serve to visualize three identified clusters.
23. Discussion about Figure 10. When this method is applied to other problems in engineering and/or atmospheric sciences, is clustering that represents ~60% of the total variance an acceptable value? In other words, this problem is related to finding the optimal combination of downburst parameters, and Figures 9 and 10 show that the clustering of solutions in the present way explains ~60% of variance among all solutions. If one looks at other (similar) problems in meteorology, engineering, earth sciences, etc., does one observe a similar level of model confidence? A few references on this subject might help.
24. Section 5.1 and other sections. Please separate your sections into multiple paragraphs. Having one paragraph that covers more almost one whole page reduces the readability of your manuscript.
25. Related to my previous comment, this manuscript should be shorter. I think that the level of English is satisfactory, but certain parts of the manuscript can be shortened.
26. How does the k-means algorithm work to improve the partitioning? What is the mechanism by which k-means algorithm moves the clusters from being overlapped to disjoint (Figure 11)?
27. Table 4. Define p1, p2, and Vk in the caption of this table.
28. Table 4 and the associated text. Explain what is the weight of a variable in the context of your analysis and the quality of representation (projection)?
29. Rewrite the first sentence in this line for better English.
30. Figure 17 and associated discussion. This is very nice. Is it possible to constrain your space of solutions by fixing some of the parameters using the observations (e.g., direction of damage, translation speed and direction of the cloud, etc.). In L545–546 you conclude that one needs to conduct many simulations, but wouldn’t make more sense to constrain simulations with known values of parameters rather than letting all parameters take arbitrary values?
31. I think the title should more highlight the main topic of this paper and that’s the application of the Teaching Learning Optimization Algorithm to your analytical model. You are using an objective method to find the optimal solution from the space of all solutions and the particular downburst is just a case study that you used to validate your method. A title such as “The Application of Teaching Learning Optimization Algorithm to Analytical Model of Downburst Outflows” might better capture the main topic of this paper, but I leave it up to the authors to decide.

Please see review_egusphere_2023_1683.pdf.