Summary for editor:

The main message of this study, emphasized both in the conclusion and the abstract, is that there is a higher risk of erosion-damage on planned Dutch off-shore wind farms in the North Eastern waters than in the South, because both rain amount and wind speeds are higher in the North East. As a scientific contribution, I find this message to be somewhat trivial, and one that could have been reached with much less effort than is presented in the paper. In general, the work reads rather like a project report, explaining what the authors have done rather than a scientific study that others can learn from. In addition, several of their statements are simply not correct. Despite some impressive material, I therefore suggest that the paper is reconsidered after major revisions or rejected, such that the authors have sufficient time to rethink their study.   

Specific comments:

• In the introduction, the authors mention that “The literature review reveals a significant gap in knowledge regarding the large-scale mapping of rain erosion risks for wind turbine blades, particularly in the Dutch North Sea.” A large part of the Dutch North Sea is however covered in the erosion atlas based on the NORA3 simulations from Hannisdottir et al 2024b, which is cited in the previous paragraph. The authors also critique the Hannisdottir study based on the course resolution of reanalysis data, but in the end they do not utilize the high-resolution data that they themselves generate. There seems to be no need for this study, because the NORA3 reanalysis covers the Dutch North Sea, and such data could easily have underpinned the claimed main conclusion of the study. Further, the authors should very clearly state the literature regarding all rain erosion atlases to help the reader understand their contribution. 
  
  
  
  
• Although not explicitly mentioned as an objective, I believe that the authors wish to formulate an alternative way of constructing a rain erosion atlas than what has been done before. By not acknowledging this objective, and by not comparing their detailed methods with already published methods, a reader cannot understand what the new methodology brings to the field. The authors should focus on the complementary aspects of a new methodology and what the difference between their results and already published data can teach the community regarding direction forwards for the development of erosion risk atlases. (They could also focus it on the pros and cons of the LES based approach in comparison with re-analysis or meso-scale simulations, which is currently not covered at sufficient depth in the study.)
  
  
  
  
• In the abstract, the authors claim to have validated their model runs using in-situ wind and rain observations. This would normally mean that the estimates of the models can be trusted within a certain error margin. However, the use of the word “validation” is here very questionable, because the estimates of the models are as much as several 100% different from the observations (see table 2). In the text, the authors claim that the models capture trends of the erosion related parameters effectively, but this is not generally correct (for example, regarding the accumulated rain amount, which show different relative behaviors for the three sites in observations and simulations). In summary, the authors do not validate the models. In the introduction, the authors mention that the LES simulations are performed to verify the meso-scale simulations. However, the meso-scale simulations show a better performance concerning wind modeling and the precipitation fields of the models are not compared in any meaningful way to justify the impression that the authors communicate regarding the LES model’s superiority. No side-by side rain and wind fields are shown to document the LES model’s strengths or weaknesses.
  
  
  
  
• Interestingly, the authors state that the LES based approach is better at capturing extreme events, without showing this important aspect in the paper! Instead, they only show results regarding rain-intensity  in the high-resolution simulations (Figure 2), but here they omit to show the results from the meso-scale simulations. This omission is problematic, because it is important to show the comparison for the simulation that the atlas is based on. In general, neither approach is well evaluated, and it is hard to judge the pros and cons of their methods based on the presented results.

• For erosion studies on wind turbines, a key parameter is the incubation period (IP), which is the life-time of the leading-edge-protection system. I find the low numbers in the last column in Table 2 here (and elsewhere) highly problematic, mainly because they are not discussed. The measurement data indicates that the IP at the offshore LEG site is little more than two years. The overall result of this study indicates that it would be significantly shorter in the planned wind farms in the North, which, in turn, would indicate that these wind farms’ O&M costs would be enormous. It is important to discuss what these low numbers mean in terms of cost-efficient operation of wind farms. Do they put a question mark on all the North Sea wind exploitation plans? Or is this low number simply a reflection of  the methods behind the IP estimation are incorrect?

• The results in the Figures 3-5 are not very easy to understand, partly based on the absolute numbers used (with very many decimals). The authors should normalize the numbers such that deviations can be seen in per cent. Also, the results from both models should be shown.
  
  
  
  
  
  
• Another reason for recommending the rejection of this paper is that the authors fail to report a conflict of interest, although they are presenting results from a commercial tool in a core area of the company that produces this tool. One of the authors is hired by this company and the other seems to have strong links to it via project funding. Hence, there is an obvious conflict of interest, and by not acknowledging it, the presented research cannot be judged in a transparent way by readers. The results regarding the LES simulations are of high scientific interest, but we can only take these results seriously if we trust the presented work. To acknowledge their obvious conflict of interest is a first step towards creating such trust. Many authors mistake the existence of conflict of interests with scientific misconduct. However, it is only misconduct if the situation is not acknowledged (in a European context, this is elaborated here H2020 INTEGRITY - Conflict of interest in research: what is it and why it matters?).

Minor comments:

Lines 150-153: The authors write However, it is essential to note that DSD measurements obtained from current sensors, such as disdrometers, remain quite uncertain (Letson and Pryor, 2023; Caboni et al., 2024; Asta Hannesdottir et al., 2024a). This is due to the fact that these sensors and their algorithms are typically optimized to accurately detect total precipitation amounts rather than the DSD itself. This is not correct. The disdrometers are made to detect droplets;  not optimized for rain rates, which is a derived parameter from the instrument (see Johannsen et al 2020).  Precipitation measurement from different types of disdrometers also vary (Angulo-Martinez et al 2018) .

Johannsen LL, Zambon N, Strauss P, Dostal T, Neumann M, Zumr D, Cochrane TA, Blöschl G, Klik A. Comparison of three types of laser optical disdrometers under natural rainfall conditions. Hydrol Sci J. 2020 Jan 21;65(4):524-535. doi: 10.1080/02626667.2019.1709641.

Angulo-Martínez, M., Beguería, S., Latorre, B., and Fernández-Raga, M.: Comparison of precipitation measurements by OTT Parsivel² and Thies LPM optical disdrometers, Hydrol. Earth Syst. Sci., 22, 2811–2837, https://doi.org/10.5194/hess-22-2811-2018, 2018.

Figure 7: The legends should not block the data shown.

Summary for editor:

The study addresses the critical issue of rain-induced leading edge erosion (LEE) on offshore wind turbine blades. As offshore wind energy expands in the Netherlands, understanding and mitigating LEE becomes essential for maintaining turbine performance and reducing maintenance costs. This research aims to fill the knowledge gap by developing a long-term erosion risk atlas for the Dutch North Sea using advanced weather and rainfall modelling simulations and integrating a previously developed fatigue-based erosion damage model. The work deserves publication after subject proposed revisions are addressed.

Specific comments and proposed revisions:

-The weather modeling used captures large-scale meteorological trends and spatial patterns in wind and precipitation, especially the northeast-southwest gradient in the Dutch North Sea. The paper clearly states that both LES and mesoscale simulations compared to rainfall experimental measurements underestimate accumulated LEE damage related to a referenced wind turbine, particularly due to underrepresentation of extreme events and large raindrops. The summary rightly identifies the Marshall–Palmer distribution’s inability to represent large droplets (>3 mm), which is a key limitation discussed in the paper.

-In one hand, authors employed Whiffle’s ASPIRE model, a GPU-accelerated large-eddy simulation (LES) and mesoscale weather modeling platform, to simulate weather conditions over two timeframes: a high-resolution one-year LES (2022–2023) for validation purposes and a lower-resolution ten-year mesoscale simulation (2014–2023) for long-term trend analysis. These simulations were compared against real experimental measurements of wind and rainfall collected at offshore, coastal, and onshore sites in the Netherlands. While the LES simulations more accurately captured extreme weather events and aligned better with observational data, the mesoscale simulations were deemed sufficient for long-term trend analysis and atlas development. However, an important result is that both simulation types underestimated accumulated damage compared to measurements. This discrepancy is primarily due to two factors: the underrepresentation of large raindrops in the Marshall-Palmer drop size distribution used in the model, and the lower frequency of extreme events in the simulations compared to observations.

---The authors should clearly identify the novelty of this result and explain how LES and meso-scale simulations serve as complementary tools in atmospheric research, each appropriate for different types of studies and objectives.

---However, despite effectively reproducing these trends, a significant limitation of the study that should be clarified  is its systematic underestimation of absolute leading-edge erosion (LEE) damage.

-In the other hand, the erosion model used in the study estimates the incubation period—the time before visible erosion begins—based on ASTM regression equations and assumes a polyurethane leading edge protection (LEP) system on a 15 MW reference wind turbine. The results reveal a clear spatial variation in erosion risk across the Dutch North Sea. The estimated incubation period ranges from 8–9 years in the southwest to 6–7 years in the northeast. This variation is attributed to higher average wind speeds and greater rainfall in the northeastern regions. The study concludes that although there are uncertainties in absolute damage estimation, the rainfall simulations effectively capture spatial trends in erosion risk due to weather conditions.

---The authors should clearly outline the aspects of their work that have been validated: whether it is the comparison of rainfall observations to simulations, or the progression of erosion damage, which is not currently depicted in the work.

-Moreover, the use of normalized incubation resistance (NOR) derived from rotating-arm rain erosion tests (Slot et al., 2025) may not fully reflect the complex real-world conditions encountered offshore, potentially contributing to the observed discrepancies. The paper emphasizes that while the rainfall model captures trends, the absolute values of damage remain uncertain without direct validation from experimental observations in operational wind turbines. Moreover, authors state future work will focus on improving the representation of drop size distributions and fall velocities in the ASPIRE model and exploring real-time erosion forecasting to enable adaptive turbine operation during extreme weather events.

---The authors should clarify first how rainfall simulations relate to damage progression uncertainty analysis for future work, as this is not detailed or validated in the paper.

-Additionally, the simplified linear damage accumulation approach adopted by the incubation period (IP) model (Caboni et al., 2024), based on the Palmgren–Miner rule, may not accurately capture non-linear effects associated with severe meteorological events. As a result, without direct validation through observed erosion damage on operational wind turbine blades, these predictions related to material damage evolution remain theoretical. The reliability and practical applicability of these estimations are consequently limited, underscoring the critical need for direct comparison with actual erosion observations from operational offshore wind farms under comparable environmental conditions to validate and refine the predictive capabilities of the proposed model.

---The authors should consider these limitations and clearly state the scope of the proposed study in the paper.