The authors investigated the possible effects of measurement data gaps on long-term offshore wind estimates for three offshore sites in Europe with 2-year concurrent measurement. This work is an interesting work for the wind resource assessment, and the manuscript is well organized. Therefore, I recommend minor revision including the implementation of the points and comments below. 

• The metrics used to evaluate the MCP method performance for wind redirection data may bring some deviation in the Northerly direction sector, such as 359°and 1°. The authors are advised to address this issue more clearly.
• It is recommended that the different curves in Figure 5 be distinguished by colors instead of line types.

The manuscript aims to show the impact of gaps in site measurement used for wind resource assessment, considering long-term extrapolation to a reference period, using reanalysis or NWP data. The study is valuable and interesting but leaves some questions unanswered (see list below). The text is generally well-written and presents the background, hypothesis, methods, and results in a clear way. My recommendation is to accept with minor revisions, mostly expanding on the questions below. 

Specific comments and questions

• P1L10-11: You say it's lower than expected, but you don't say what you expected and why you expect that. You should help the reader set what reasonable expectations are.
• P1L11-12: Throughout the paper, you focus mostly on the mean outcomes, but is that the most important measure to consider? what happens to the results if you consider the uncertainty? the P75, P90 results? I would expect many readers to want to know about the worst outcomes.
• P1L12-13: Why would I use a slightly worse method to do long-term correction, than the one I use for gap-filling?
• P2L38-39: How many months is "various months"?
• P3L78-79: I would rephrase the "exact" part here. It sounds as if mast measurements have no error or uncertainty.
• P5L113-114: Did you resample to 1H by averaging, or simply taking the 10 Min values every hour?
• P7L186: How is the circular nature of wind direction treated in the KNN model? e.g. is a wind direction of 359.9° a close neighbor of a wind direction of 0.1°?
• P10-11 Gap generation: As you allude, the worst-case gaps are probably large contiguous gaps near the extremes of the annual cycle, i.e. summer and winter in your cases. Did any of the gap-generation methods result in large gaps in summer or winter in both years? having measurements from one summer can probably alleviate the problems from large gaps in the other year, but if it's missing from both it would be a problem. Perhaps this is out of the scope of your study? It would be good if you mentioned that in the paper in that case. Even so, it would be interesting to know what the results are for the case where both summers (JJA) are removed or both winters (DJF), which would probably represent worst-case scenarios for a 180-day gap. In general, I am missing some more comments/discussions about the annual cycle and the effect of generating gaps in the extremes of the cycle. Total availability (e.g. 75\%) is not as interesting as the distribution of availability.

Technical corrections

• P6L148: Missing number. What Type?