This study focuses on the application of UAV attitude dynamics in wind speed and direction estimation. Through a complete research chain—comprising "wind tunnel calibration, meteorological tower validation, and coastal field application"—the study presents a wind measurement solution that is "sensor-independent, low-cost, and offers high spatiotemporal resolution." This approach effectively addresses the conflicts traditionally seen in wind measurement technologies, particularly in vertical wind measurement, flexible deployment, and cost control. The study is well-designed, with substantial data support, and the results are both scientifically innovative and valuable for engineering applications. The research holds significant reference value for the interdisciplinary fields of low-altitude economy and environmental science. However, there is still considerable room for optimization in terms of the method's generalizability, experimental details, depth of data analysis, and refinement of application scenarios. The technical character of the manuscript makes it suitable to be submitted as a Technical Note rather than a research article in Atmospheric Chemistry and Physics.

1. This paper develops an attitude-wind speed power-law model based on the DJI M300 RTK; however, it does not explicitly address the model’s applicability to other rotorcraft UAVs (e.g., small consumer drones or multi-rotor agricultural UAVs). It is recommended to include the corresponding differences in attitude for other UAV platforms, such as the impact of body weight and the number of rotors on tilt angle.
2. The wind speed measurement in the paper relies on the UAV's attitude parameters. If the wind field does not affect the UAV's attitude, how is the wind speed measured in such cases? For instance, if the wind is purely vertical. In real-world atmospheric conditions, although purely vertical winds are rare, significant vertical wind components are common, such as in thermal convection, terrain-induced uplift, frontal systems, or urban heat island circulations.
3. The paper reports that the drag coefficient (Cd) is measured in wind tunnel experiments and significantly changes with wind speed (tilt angle), ranging from 2.75 to 0.20. Cd stabilizes at around 0.20 for wind speeds greater than 5 m/s. What is the physical explanation for this change? Is it related to the variation in the effective projected area of the UAV at different tilt angles, or the transition of the flow field from a separated to an attached flow state? The paper mentions that the projected area changes between 880-1120 cm². Is this range sufficient to explain the drastic change in Cd?
4. The sensor’s non-zero readings under zero wind speed (0.1-1.5 m/s) are an important source of system error. Does the proposed calibration framework include a correction for this "zero bias"? In the linear fitting shown in Figure 4, is the intercept significant, and does it represent this zero bias?
5. The random forest model yields excellent results (R² > 0.9), but its input features are limited to attitude angles (pitch, roll, yaw). Was there any consideration of incorporating other easily accessible flight data as features, such as altitude, GPS speed, or even motor speed, to enhance the model’s generalization ability under complex flight conditions (e.g., acceleration, deceleration, turning)? If motor speed is included as a parameter, could it potentially measure the previously mentioned pure vertical wind field?
6. The model was trained on hover data from a coastal site and validated against meteorological tower data. How well does it generalize to wind field estimation in different terrains (e.g., urban canyons, forest canopy) or under varying atmospheric stability conditions?
7. The paper mentions that “the forward top load (Mo+f) reduces vibration and sensor error,” but does not specify the basis for selecting the load weight (1.5 kg), such as whether it is representative of the UAV's typical payload. It is recommended to include an analysis of the correlation between load weight and error (e.g., a comparison of 1.0 kg, 2.0 kg payloads).
8. During vertical profile measurements, does the UAV's own upward/downward motion or rotor-induced disturbance (especially at speeds of 2 m/s) significantly disrupt the natural wind field and temperature-humidity stratification at the measurement height, thereby affecting the representativeness of the results? Specifically, in a stable boundary layer, could the UAV's downwash induce local mixing, smoothing the measured concentration gradients?
9. When calculating the average wind profile, how long is the data collection period at each height point? For rapidly changing wind fields, is brief hovering sufficient to obtain statistically representative average values?
10. Line 30: ‘For example, wind play a crucial role in the generation of sandstorms.’ Error in using ‘wind’ and ‘play’ together.
11. ‘Figure 1 UAV flights conducted in a wind wall laboratory (A), at a meteorological observation tower (B), and at a coastal site (C). Schematic diagrams of UAV payload configuration (D) and UAV flights under different relative wind directions.’ Lack of description for subgraph E.

Please see attached document.