I commend the authors for overhauling their article as substantially as they did, including adding a couple of case studies to analyze. Their numerous, thorough revisions greatly improved the manuscript. While I do have a handful of minor comments that should be able to be quickly addressed by the authors, my chief concern at this point is that it is not really novel information that different combinations of WRF physics parameterizations perform well or struggle in certain situations. Many, many studies have shown this. However, I do think there still is value in demonstrating this for extreme weather events that impacted a wind farm in the Belgian North Sea. I think that there was little consistency across these three case studies for which parameterizations tended to perform better or worse is itself still interesting, and points to the need for more robust statistics with (much) larger sample sizes to make more general conclusions with confidence about a “best configuration” for accurately modeling the wind speed and wind direction at offshore wind farms in that region, or whether the “best configuration” will change based on the weather regime. I would like to see the authors state that caveat more clearly, especially in the Conclusion.
1. Uncapitalize terms that are abbreviated if they are common nouns (e.g., extreme weather events, numerical weather prediction, planetary boundary layer, lateral boundary conditions, mean absolute error, etc.), but retain capitalization only for proper nouns (e.g., normalized Euclidean distance).
2. When defining an abbreviation, the term that is abbreviated should be singular. For instance, EWE should stand for extreme weather event. If there are multiple extreme weather events, then use EWEs. The same principle applies to LBC: LBC should mean lateral boundary condition, while LBCs should mean lateral boundary conditions. Once this change is made, ensure that all subsequent mentions of the abbreviation correctly refer to the singular (EWE, LBC) or plural (EWEs, LBCs) usage.
3. Generally speaking, single-digit numbers should be spelled out. For instance, “3 PBL parameterizations” should be “three PBL parameterizations” (line 59), “5 classes” should be “five classes” (line 94), and “1-way nested” should be “one-way nested” (line 177). There are exceptions to this general rule, of course (especially when pertaining to things like measured values, time, etc.), but there are quite a few more cases like the three I mentioned that should be corrected.
4. Lines 42, 221: Delete the “see” before “e.g.”. The “see” is already implied.
5. Lines 48, 49, 53: Add “the” before “PBL”.
6. Lines 73–74: “by translating it into compensating subsidence, a combination of vertical advection, moisture, and temperature.” — This is awkwardly worded and needs to be revised.
7. Line 89: Change “vertical” to “the 3D”.
8. Line 138: Change “therein” to “herein”.
9. Line 142: Change “Storm Ciara is” to “Storm Ciara was”.
10. Line 150: Change “British isles” to “British Isles” (capitalize Isles, as the term as a whole is a proper noun).
11. Figs. 1 and 19: These maps are much improved in this revision. That said, I suggest (but do not require) additional revisions to these plots, specifically the contour intervals. Specifically, I think the upper limit of the range should be something more appropriate to the data (even 10 mm/h barely appears on the filled contour maps, so why is 100 mm/h the upper limit?). I would suggest something like [0.01, 0.10, 0.50, 1.00, 5.00, 10.00, 20.00, 30.00), or whatever set of contour intervals would be helpful to capture the meaningful variations in the data better. This is not a critical fix, but is a principle that can be applied to all sorts of figures; you do not have to stick with uniformly spaced contour intervals, whether in linear or exponential space, if that is not the most meaningful way to display the field. Also, is the star filled white, and thus blocking any view of precipitation behind it, or is it transparent/not filled? Ensuring the star is transparent/not filled white would be a small but important fix.
12. Line 157: Note that the low in Fig 2b is centered over the English Channel and Normandy, not over the North Sea.
13. Fig. 3: Because you made the right axis red to match the line color, you should do the same for the left axis and data series (either make both blue or both black). Also, the green dotted lines are tough to see. Maybe shade that strip green instead? Or make the linewidths thicker.
14. Lines 178–179: Change “terrain following pressure levels” to “terrain-following model levels”.
15. Fig. 4: Delete “and WRF Post-processing System (WPS)”, as it is unnecessary (it is also the WRF Pre-processing System, anyway). You can either crop the title out of the figure to avoid needing to define WPS here, or you can modify the title in the plotgrids NCL script to something like, “WRF Domain Configuration”.
16. Table 1. In the row for horizontal grid spacing, use commas instead of x symbols to separate the values of the different domains, as you already did for the time steps a few lines below.
17. Line 194: Singular-plural agreement: It should either be “a combination…is considered” or “combinations…are considered”.
18. Line 195: Change “variations of update interval of LBC, PBL, cumulus, and microphysics schemes” to “variations of the update interval of the LBCs, and the PBL, cumulus, and microphysics schemes”.
19. Lines 212–213: “NED is defined as the resultant of…” — Resultant has a reserved meaning in mathematics, and the right-hand-side of the definition of NED is not a resultant. It is the square root of the sum of squares—maybe there is a fancy mathematical term for that?
20. Sec. 4.3: It would also be interesting to note that Run 2, while the best member for Case 2, was the worst member for Case 3. Different weather situations lead to different skill for various schemes. And here MYNN PBL was not that bad (or at least not as consistently poor as in other cases).
21. Line 272: Change “Errorbars” to “Error bars”.
22. Line 276 and elsewhere: Change “observations” to “results” (or similar). You frequently use “observations” or “is observed” in this manuscript to describe what you noticed in a figure or table, but that is confusing when “observations” elsewhere refers to measurements used as truth for validation. You should generally find other words than “observations” or “is observed” when describing the results or conclusions you saw or determined from figures, tables, etc.
23. Line 360: Change “schemes” to “scheme”.
24. Line 367: Change “highlighting” to “highlight”. I will also add that the other thing that this finding highlights is the case-to-case variability in skill.
25. Line 372: Change “better in comparison” to “better by comparison”.
26. Line 380: Change “3 unique case studies” to “three EWE case studies”.
27. Line 396: Hyphenate “lower-order”.
28. Line 405: Add a comma after “(Newman et al., 2022)”.
29. Line 406: Change “latter raw data” to “raw radar observation data”.
30. Line 420: “In addition, the simulated precipitation is qualitatively compared to radar data from RMI-B.” — Precious little qualitative analysis in precipitation rate was performed here. You simply included Figs. 1 and 19, and stated that precipitation in WRF varied by physics configuration, but you did not describe any qualitative analysis. Either some text should be added in the Discussion (or elsewhere) that better describes your qualitative analysis, or you should remove this part.
31. Line 421: Change “impact of update interval of LBC” to “impact of the update interval of the LBCs”. Also change “used for PBL” to “used for the PBL”.
32. Lines 433–434: Change “combine scale-aware” to “combine the scale-aware”. Change “with scale-aware” to “with the scale-aware”. Change “and 5-class single moment” to “and the five-class single-moment”.