• Major comments
• Overall, the authors provide a good review of turbine control and an interesting concept that uses a combination of a wind speed estimate (WSE) and a lidar measurement to inversely solve for the ideal blade pitch and generator torque controls of the wind turbine.
• Many components of the controller (and the ROSCO baseline) are described in great detail but are not novel contributions in this article.  Could those sections be streamlined for this audience?
• The controller relies heavily on the Cp surface.  It's an interesting result that using the same Cp surface in the wind speed estimator can relieve biases in the Cp surface.  Are there constraints (smoothness, monotonicity) that are required to get a unique pitch and torque at each time step?
• It is interesting that the authors don't use any time delay or buffering in their lidar measurements.  This can be a source of difficulty when using lidar measurements.  I'm guessing that the controller performance then changes with wind speed because the lead time of the lidar changes, while the lag time of the WSE should be relatively constant.
• I wonder if you are reporting better performance at high wind speeds because the lead time of the lidar is lower and similar to the lag of the WSE.
• The reporting of results in timeseries form could be improved.  Right now, the reader must scroll between figures and pages to evaluate long timeseries with subtle differences.  Is there a short, illustrative section of time that demonstrates the difference between your NOR controller and ROSCO.  One that shows the impact of lidar measurements?  Another way to demonstrate differences is by binning the results in time and plotting against wind speed. 
• How exactly are you limiting the maximum thrust and fatigue loads to the levels in Section 2.5?  What part of the controller (Algorithm 1) contributes to fatigue load reduction?  The reduced variation in generator torque?
• A control engineer in practice will want to adapt these loads, and see the downstream effect on power.  Are you able to do this with your control scheme?  This would be an interesting result of the control concept.
• Minor comments
• Line 175: there is, in most cases, a power sacrifice near rated power when using peak shaving control.
• Section 2.5: I don't understand your units on the fatigue loads.  Are these DELs?  They should have the same units as the load, N-m.  What are the nominal loads of the baseline control?  How do these compare?
• Results: it may make more sense to only compare RMS rotor speed error above rated wind speeds
• L498: Should \hat{M}_a have \hat{v}_x in it?
• Editorial comments
• openFAST should be OpenFAST
• Figs 4 and 5 could be combined; they look very similar to the ones in the ROSCO paper.
• The capitalization and font size in figures should match the text.  Also there are labels like "windspeed" that should be "Wind speed (m/s)," for example.  Fig 5 has a legend with a variable, and it's not defined in the caption.
• When substituting equations, it could help the reader to add a few descriptive words referencing the equations from earlier.

The paper presents a wind turbine control design using the nonlinear output regulation (NOR) method to reduce fatigue loads. The topic is interesting and relevant, but the motivation could be more clearly articulated. Given that the authors have published similar work in 2021, it would be helpful to clarify the specific motivation for employing the NOR method in this study.

The proposed controller maximises power output by tracking the rotor speed set-point, which is computed based on a static Cp surface and the optimal tip-speed ratio. This approach closely resembles the traditional K-Omega squared law. However, unlike the K-Omega squared method, the proposed approach relies on estimated wind speed, introducing additional sources of uncertainty. From an industrial perspective, this could be a potential drawback compared to the traditional method.

Additionally, the title suggests that the paper focuses on reducing fatigue loads, yet Algorithm 1 does not explicitly account for fatigue mitigation. Is this achieved through peak shaving or another mechanism? Providing further details on this aspect would strengthen the paper’s contribution.

The paper also states that averaging the estimated wind speed with LiDAR measurements improves the estimation of low-variation real-time wind speed. It would be helpful to elaborate on why simple averaging was chosen over a weighted sum. What was the motivation behind this decision?

Finally, while the results show that the proposed method outperforms ROSCO, this is perhaps expected given that it incorporates a DAC approach and LiDAR. A more informative comparison might be against other LiDAR-assisted control methods to better assess the advantages of the proposed approach.

Given these points, I believe the paper would benefit from further clarification and refinement. I encourage the authors to address these concerns, as doing so would strengthen the manuscript significantly.