Please find comments in the attached pdf

The paper deals with the issue of use of reanalysis data for wind-energy purposes. In particular, it is known that power spectrum of wind speed resulting from numerical methods suffer of energy loss at high frequencies, leading to an underestimation of extreme wind speeds needed for the structural design of wind turbine. In the manuscript, the authors evaluate the possibility of use Convenction-Permitting Models to better reproduce the theoretical -5/3 slope in wind power spectrum as alternative to the most used datasets like, e.g., ERA5. The study is very interesting and contributes significantly to the important open debate on the reliability of reanalysis data for structural design purposes. The brief communication is suggested to be published after minor revision.

Comment #1 - Page 3, Lines 81-83

 “Originally recorded at 10-minute intervals, these data were subsequently aggregated to hourly values by arithmetic averaging to facilitate direct comparison with the simulations of the CPM models.”

Please explain whether the hourly sampled CPM data are representative of 10-min average wind speed or 1-hour average wind speed. Indeed, e.g., in ERA5 dataset the provided parameters are available hourly, and they are defined as either instantaneous value referring to a specific point-in-time (thus not averaged) or mean rate value averaged over a given time period. If the case is the first, then the “arithmetic averaging” of field measurements seems to be wrong.

Comment #2 - Page 45, lines 116-117

“the hourly time series of CPMs, ERA5, and NEWA, were first detrended by subtracting their mean value, thus removing the constant component.”

The detrending operation is usually made to remove low-frequencies oscillations that can introduce an unwanted mean component to short records. Please specify better what you mean by “detrend” in this case: is it perhaps to make time series zero-mean for the purpose of deriving power spectra?

Comment #3 - Figure 1.

It is shown that CPM simulations provide enhanced spectral contribution at larger frequencies. Is there a possible physical explanation for this phenomenon?

Comment #5

In addition to the KIT measurement site, the PSD of randomly-selected 10 locations are discussed. Since they can be relevant, please describe both roughness and orography conditions at KIT measurement site as well as at the 10 locations of Figure 2.

Comment #6

The study is limited to comparison of Power Spectrum. Evaluate the possibility of compare recorded and simulated yearly maxima at KIT measurement site, e.g. by showing their empirical distribution function or, at least, the right tail of the parent distribution.

Dear authors, thanks for a short, interesting and well-written manuscript!



See my comments in the pdf attached.



I'd like you to review existing, recent works already published which adress the same topic and use CPM models. This is not to question to novelty of your work, but instead to bring you closer to the small community of CPM modellers with an interest for Wind Engineering applications (not only Wind Energy, but also Wind Hazards in general). 



All the best

Rémi Gandoin, C2Wind, Denmark.