the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Multi-Rotor UAV Observations of the Atmospheric Boundary Layer over Complex Terrain: Accuracy Validation and Dynamic Mechanisms of Flight Disturbances
Abstract. Unmanned aerial vehicle (UAV)-based sounding has become a flexible and cost-effective approach for probing the atmospheric boundary layer (ABL), yet its measurement credibility over complex terrain and the dynamic response of UAV flight attitude to boundary-layer dynamics remain insufficiently quantified. Here, we implement multi-rotor UAV field campaigns equipped with well-calibrated meteorological payloads, including co-located intercomparison observations against a 100-m gradient meteorological tower at the Xingtai Field Scientific Experiment Base, and vertical-profile surveys across the lower ABL over the eastern Taihang piedmont. The comparative observations demonstrate excellent agreement between UAV and tower measurements: vertical profiles of wind speed, wind direction, air temperature, and relative humidity exhibit strong correlations, negligible systematic biases, and coherent vertical structures. Stability parameters derived from high-frequency hovering observations, including potential temperature and gradient Richardson number (Ri), also closely match tower references, verifying the platform’s capability to reliably resolve ABL thermal stratification and dynamic stability. Three successive evening soundings capturing the transition from a convective to a stable boundary layer reveal that UAV attitude disturbances are strongly suppressed within stably stratified layers, but markedly amplified in near-neutral to weakly unstable layers where intense vertical wind shear coincides with weak thermal suppression, with the most severe perturbations occurring near the cloud base. The joint distribution of Ri and vertical wind shear further indicates that turbulence-induced flight bumpiness generally intensifies with increasing shear and peaks under unstable stratification, while a localized anomaly suggests possible resonance-like coupling between turbulent eddy scales and airframe dynamics. These findings establish a quantitative observational basis for reliable UAV-based ABL sensing and low-altitude flight-safety assessment over complex terrain.
- Preprint
(6676 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2026-4218', Zhenxin Liu, 12 Aug 2026
-
AC1: 'Reply on RC1', Zhen Zhang, 02 Nov 2026
Response to reviewer#1’s comments
Referee #1’s comment:
Scientific comments:
This study employs a DJI M350 RTK multi-rotor UAV to investigate the reliability of atmospheric boundary layer (ABL) observations and flight safety over complex terrain. The authors conducted comparative observations against a 100-m gradient meteorological tower and vertical-profile surveys over the eastern Taihang piedmont. They first demonstrated that the UAV platform can accurately measure both basic meteorological variables (temperature, wind, humidity) and derived stability parameters (potential temperature and gradient Richardson number). Furthermore, the study quantitatively reveals the combined effects of vertical wind shear and stratification stability on UAV attitude disturbances, and proposes a possible vortex-airframe resonance-like mechanism, which is a novel contribution to UAV-based meteorological applications. Notably, the paper employs a sample-pairwise finite-difference method for Ri calculation, which preserves the distribution information of parameters between stratified layers and better captures turbulence details than conventional layer-averaging approaches.
The topic is well within the scope of AMT. The experimental design is rigorous, the dataset is substantial, and the analytical framework is logically structured. The findings provide valuable references for improving UAV-based ABL observations and low-altitude flight safety over complex terrain. The manuscript is well written and the figures are of high quality. Before formal acceptance, I recommend that the authors address the following specific issues.
Response: We sincerely thank the reviewer for the careful reading of our manuscript and for the positive and encouraging evaluation. We are grateful that the reviewer recognized the value of combining UAV–tower comparisons with vertical-profile observations, the quantitative analysis of how vertical wind shear and stratification stability jointly affect UAV attitude disturbances, the possible resonance-like mechanism, and the sample-pairwise finite-difference method used to calculate Ri. We have carefully considered each of the specific comments, and our point-by-point responses are provided below. For clarity purpose, here we have listed the reviewer's comments in bold font, followed by our response in blue plain font, and the modifications to the manuscript are in blue italics.
- The logical chain from observational results to conclusions needs to be strengthened
The authors describe the time-height evolution of the large-scale wind shear field based on Figure 5c, and further analyze the shear characteristics during three flight periods (18:24-18:43, 18:55-19:22, and 19:43-19:55 BJT) in Section 3.2 (starting from page 12, line 328; page 13, line 365; and page 14, line 400, respectively). However, in the detailed discussion of these three cases, the authors do not explicitly link the temporal evolution of the large-scale shear field revealed in Figure 5c with the vertical distribution characteristics of UAV attitude responses during each period.
For example, during the first flight (18:24-18:43), which positive/negative large-scale shear zones did the ascending and descending segments (black solid and dashed lines in Figure 5c) traverse? Could the difference in this large-scale background shear be a significant cause of the asymmetry in flight disturbances between the ascending and descending profiles (Figure 7)? I suggest that the authors actively reference Figure 5c when discussing each case, and superimpose the flight trajectories onto the large-scale shear background for joint analysis. This would significantly strengthen the logical chain connecting the macro-scale background to the local disturbance responses.
Response: We thank the reviewer for this insightful comment. We have revised Section 3.2 to explicitly connect the large-scale wind shear evolution in Figure 5c with UAV attitude responses for all three flight periods. For each case, we describe the large-scale shear zones traversed by both ascending and descending trajectories and discuss how background shear differences shape the asymmetry of flight disturbances. The UAV flight paths are superimposed on the wind shear field in updated Figure 5c, and Figure 5c is cited throughout the case analysis to jointly interpret macro shear environments and corresponding UAV perturbations, reinforcing the physical linkage between large-scale background conditions and local attitude responses.
- The evidence supporting the "resonance-like amplification" hypothesis needs to be reinforced
On pages 16-17, lines 465-470, the authors propose a "resonance-like amplification" conjecture based on Figure 12. This is an interesting speculation, but it is currently supported only by anomalous data points in a specific regime (vertical wind shear approximately 0.007-0.010 s⁻¹ combined with near-zero to negative Ri). The evidence appears somewhat thin. I recommend that the authors: (1) adopt a more cautious tone when presenting this viewpoint, clearly stating that this is a preliminary hypothesis requiring further verification; and (2) if possible, provide additional supportive analysis—for instance, by roughly estimating the characteristic scale of turbulent eddies under this specific shear range and discussing whether this scale is comparable in magnitude to the airframe dimensions or the natural frequencies of the MATRICE 350 RTK in pitch/roll directions. If such analysis cannot be provided, I suggest relocating this discussion to the end of the Discussion section as an outlook for future research directions.
Response: We are grateful for the insightful comments made by the reviewers. We acknowledge that the proposed resonant amplification hypothesis is based solely on limited observational results with a vertical wind shear of approximately 0.007–0.010 s⁻¹ and Ri close to zero or negative, and the related evidence is relatively weak. We have explained this in the text in Section 3.2, using a more cautious tone and clearly stating that this hypothesis is only a preliminary speculation and requires further verification. We added cautious restrictive explanations to clarify that the resonant amplification concept is only a preliminary assumption and that the current data set has limitations that prevent the estimation of the vortex scale.
- Inconsistency in the description of cloud-layer vertical extent
On page 15, line 405, the authors state "in the cloud layer (approximately 900-1200 m) and outside the cloud layer." However, on page 16, line 437, they mention "In the stable cloud-top layer above 1000 m." There is ambiguity and inconsistency between the description of the "cloud layer" (900-1200 m) and the "cloud-top layer" (above 1000 m). I suggest that the authors clarify and unify the vertical distribution and extent of the cloud layer, as this directly affects the accurate interpretation of the conclusion that "stable stratification suppresses disturbances."
Response: We thank the reviewer for pointing out the inconsistency in our description of the cloud’s vertical extent. We have revised the relevant text. In the 19:43–19:55 flight case, clouds were detected between 800 and 1400 m. The cloud top was likely above 1400 m, which was the maximum altitude reached by the UAV. The term “cloud-top” was therefore incorrect, as the observed altitude range corresponded to an in-cloud region rather than the cloud-top layer. Accordingly, we have revised the relevant wording to “In the stable cloud layer above 1000 m”.
Minor Comments:
1.The terms "flight bumpiness" (e.g., page 2, line 49; page 14, line 391) and "flight disturbance" (e.g., page 2, line 33; page 16, line 461) are used interchangeably throughout the manuscript. I recommend standardizing to "flight disturbance" or the more precise term "attitude disturbance" to maintain terminological consistency.
Response: We appreciate the reviewer’s suggestion on terminology standardization. We have systematically replaced all occurrences of “flight bumpiness” with “attitude disturbance” throughout the manuscript to unify terminology, as “attitude disturbance” more precisely describes the UAV attitude perturbations analyzed in this work.
- Figure formatting details:
1)In Figures 4, 6, 8, and 10, the variable names on the x-axes (e.g., theta, Ri ,etc) should be by mathematical variable formatting and italicized to comply with conventions, like: θ and Ri.
Response: We thank the reviewer for this careful check of figure formatting. We have revised the x-axis labels in Figures 4, 6, 8, and 10, applying proper mathematical italic formatting for variables such as θ and Ri to follow standard typesetting conventions.
2)In the captions of Figures 7, 9, and 11, the repeated use of "profile" could be simplified; for example, change to "Same as Figure 6, but for the vertical response of UAV attitude parameters during ...".
Response: We appreciate the reviewer’s helpful suggestion for simplifying figure captions. We have streamlined the captions of Figures 7, 9, and 11 following the recommended format, adopting concise phrasing starting with “Same as Figure 6, but for the vertical response of UAV attitude parameters during …” to avoid redundant repetition of the term “profile”.
-
AC1: 'Reply on RC1', Zhen Zhang, 02 Nov 2026
-
CC1: 'Comment on egusphere-2026-4218', Xuelin Zhang, 17 Aug 2026
The manuscript presents a valuable field investigation of UAV-based atmospheric boundary-layer observations and UAV flight disturbances over complex terrain. The combination of UAV–tower intercomparison and vertical sounding observations is interesting and provides useful observational evidence. The manuscript is generally well structured. I have several comments that may help clarify the methodology and interpretation.
1.The authors define the attitude turbulence index(T) as and state that it can serve as an indirect indicator of atmospheric turbulence intensity. This is a useful indicator of UAV attitude disturbance. However, the attitude response data obtained may also depend on the characteristics of the UAV platform and its control response. It is recommended that the authors briefly explain the physical meaning and scope of application of this metric. If there are relevant references, listing them would further aid in understanding.
- The manuscript identifies a localized region of enhanced attitude turbulence index under moderate vertical wind shear and near-zero to negative Richardson number, and proposes that this may be related to a possible resonance-like coupling between turbulent eddy scales and the dynamic response of the UAV.This is an interesting interpretation and potentially provides an additional perspective on the observed UAV response. At the same time, the current evidence appears to be mainly based on the statistical relationship between atmospheric parameters and UAV attitude response, without direct analysis of the characteristic frequency or spatial/temporal scales of the atmospheric turbulence and UAV dynamics. The authors have already appropriately noted that this interpretation is tentative. Nevertheless, it would be beneficial to explore this perspective further and, where possible, provide relevant references to support this interpretation.
- The calculations of Ri and vertical wind shear are based on discrete UAV vertical profile data. Although the calculation process has been described in great detail, it would be valuable to examine whether the derived gradients are sensitive to the observed vertical resolution. Therefore, we recommend that the authors briefly discuss the feasibility of the current vertical resolution and cite relevant studies that have used discrete radiosonde or UAV profile data.
- The manuscript uses three successive evening sounding cases to examine the transition from convective/weakly unstable to stable boundary-layer conditions. The results show clear differences in UAV attitude response among the different stability conditions. Since several of the conclusions are based on these three representative cases, I suggest that the authors briefly explain the criteria for selecting these cases and discuss their representativeness and the applicability of the conclusions to other atmospheric conditions.
- 5.The UAV was operated approximately 50 m upwind of the gradient tower to minimize mutual interference. A brief explanation or reference supporting the choice of this distance would improve the description of the experimental design.
- The authors are encouraged to further discuss how atmospheric stability modulates the relationship between vertical wind shear and UAV flight disturbances, particularly in the interpretation of Figure 12.
Citation: https://doi.org/10.5194/egusphere-2026-4218-CC1 -
AC3: 'Reply on CC1', Zhen Zhang, 02 Nov 2026
Response to Community reviewer’s comments
Community reviewer’s comment:
The manuscript presents a valuable field investigation of UAV-based atmospheric boundary-layer observations and UAV flight disturbances over complex terrain. The combination of UAV–tower intercomparison and vertical sounding observations is interesting and provides useful observational evidence. The manuscript is generally well structured. I have several comments that may help clarify the methodology and interpretation.
Response: We sincerely thank the reviewer for the time and effort spent reviewing our manuscript and for the positive evaluation of our work. We are pleased that the reviewer found the combination of UAV–tower intercomparison and vertical sounding observations interesting, and that the study provides useful observational evidence on boundary-layer structure and UAV flight disturbances over complex terrain. The reviewer’s constructive comments have been very helpful in clarifying the methodology and strengthening the interpretation of the results. For clarity purpose, here we have listed the reviewer's comments in bold font, followed by our response in blue plain font, and the modifications to the manuscript are in blue italics.
1.The authors define the attitude turbulence index(T) as and state that it can serve as an indirect indicator of atmospheric turbulence intensity. This is a useful indicator of UAV attitude disturbance. However, the attitude response data obtained may also depend on the characteristics of the UAV platform and its control response. It is recommended that the authors briefly explain the physical meaning and scope of application of this metric. If there are relevant references, listing them would further aid in understanding.
Response: We sincerely thank the commenter for the positive assessment of our work and for this thoughtful comment. We agree that T reflects the closed-loop attitude response of the UAV. It therefore depends not only on the atmospheric forcing but also on the characteristics of the airframe and its flight-control system.Physically, T is the magnitude of the pitch–roll angular-velocity vector. For a hovering multirotor, it approximates how quickly the rotor thrust axis is being tilted. The mean wind is balanced by a nearly steady tilt, which contributes little to the angular rates. T therefore mainly reflects the attitude response to gusts and turbulent eddies, including the corrective tilting applied by the flight controller.As stated in Section 2.3, T is used as a relative index to compare layers and soundings flown with the same aircraft (DJI Matrice 350 RTK), payload and control settings, and its transfer function is platform-specific. It is not intended as an absolute measure of turbulence intensity. Applying T to other aircraft, or converting it into an absolute quantity such as turbulent kinetic energy or the eddy dissipation rate (EDR), would require calibrating the platform response, for example against sonic-anemometer measurements. Several studies infer wind and turbulence from multirotor attitude and show that this relationship depends on the platform (Palomaki et al., 2017; González-Rocha et al., 2020; Wetz et al., 2021; Shelekhov et al., 2022). Aviation turbulence reporting makes a similar distinction: response measures depend on the aircraft, whereas EDR does not (Sharman et al., 2014).
2. The manuscript identifies a localized region of enhanced attitude turbulence index under moderate vertical wind shear and near-zero to negative Richardson number, and proposes that this may be related to a possible resonance-like coupling between turbulent eddy scales and the dynamic response of the UAV. This is an interesting interpretation and potentially provides an additional perspective on the observed UAV response. At the same time, the current evidence appears to be mainly based on the statistical relationship between atmospheric parameters and UAV attitude response, without direct analysis of the characteristic frequency or spatial/temporal scales of the atmospheric turbulence and UAV dynamics. The authors have already appropriately noted that this interpretation is tentative. Nevertheless, it would be beneficial to explore this perspective further and, where possible, provide relevant references to support this interpretation.
Response: We sincerely thank the commenter for the interest in this interpretation and for the constructive suggestion. We agree that the current evidence is statistical and this interpretation is not yet supported by a direct analysis of the characteristic frequencies and scales of the turbulence and the UAV dynamics. We identified the anomaly from the joint distribution of Ri, vertical wind shear and T (Figure 12). We did not directly analyse the characteristic frequencies or the spatial and temporal scales of the turbulence or of the UAV dynamics, which is why we described the interpretation as tentative. Our interpretation is based on how an aircraft’s response to turbulence depends on frequency. The attitude response can be viewed as the turbulence spectrum filtered by the vehicle’s own response function. It is therefore strongest where turbulent energy falls within the vehicle’s sensitive frequency band (Cornman et al., 1995), and for multirotors this band is limited (Wildmann and Wetz, 2022). Boundary-layer eddies can be comparable in size to small UAVs (Davoudi et al., 2020). Their scale, and hence the frequency at which a hovering UAV encounters them, varies with stability and wind speed (Kaimal et al., 1972). A particular combination of shear and stability could therefore concentrate turbulent energy within the airframe’s sensitive band and amplify the attitude response.
3. The calculations of Ri and vertical wind shear are based on discrete UAV vertical profile data. Although the calculation process has been described in great detail, it would be valuable to examine whether the derived gradients are sensitive to the observed vertical resolution. Therefore, we recommend that the authors briefly discuss the feasibility of the current vertical resolution and cite relevant studies that have used discrete radiosonde or UAV profile data.
Response: In this study, Ri and vertical wind shear were computed by finite differences between adjacent hovering levels. These levels were 8–16 m apart in the tower comparison and 200 m apart in the attitude-disturbance soundings, so the results represent layer-averaged gradients. At the 8–16 m spacing, the UAV-derived Ri agrees well with the tower values at the same levels (Figure 4b). Previous studies have shown that both shear and Ri are sensitive to vertical resolution (Balsley et al., 2008; Houchi et al., 2010; Shao et al., 2023). Richardson numbers derived from discrete profiles are nonetheless widely used, for example to estimate boundary-layer height from radiosondes (Guo et al., 2016) and turbulence from multicopter profiles (Greene et al., 2022).
We fully agree with the commenter that the derived gradients may be sensitive to vertical resolution. We acknowledge that this sensitivity was not evaluated in the present work, and that the 200 m spacing of the attitude-disturbance soundings is a limitation. Ri and shear in these layers should be regarded as bulk-layer values, and the Ri thresholds should be interpreted with caution.
4.The manuscript uses three successive evening sounding cases to examine the transition from convective/weakly unstable to stable boundary-layer conditions. The results show clear differences in UAV attitude response among the different stability conditions. Since several of the conclusions are based on these three representative cases, I suggest that the authors briefly explain the criteria for selecting these cases and discuss their representativeness and the applicability of the conclusions to other atmospheric conditions.
Response: We sincerely thank this constructive suggestion. The cases were selected primarily from observations during the transition from a convective to a stable boundary layer. We acknowledge the following limitations of the present work:
All three cases come from a single summer evening at one site in the eastern Taihang piedmont, under cloudy conditions with weak synoptic forcing. Other conditions are not covered, such as a fully developed daytime convective boundary layer, strong-wind or frontal situations, nocturnal low-level jets, other seasons and other terrain types. The specific thresholds and magnitudes, as well as case-specific features such as the ascent–descent asymmetry and the cloud-base disturbances, should therefore not be generalized directly.
5.The UAV was operated approximately 50 m upwind of the gradient tower to minimize mutual interference. A brief explanation or reference supporting the choice of this distance would improve the description of the experimental design.
Response: The 50 m distance was chosen mainly to avoid mutual interference. Placing the UAV upwind gives it free inflow that is not disturbed by the tower structure. At this distance, the rotor-induced flow of the UAV is also expected to have decayed before reaching the tower sensors. The distance also provides a safety margin for operating the aircraft near the tower. A similar upwind arrangement was used by Wetz et al. (2021), who hovered quadrotors about 20 m upwind of a 99 m tower to ensure undisturbed inflow. Our larger separation is more conservative with respect to interference and flight safety for the larger M350 RTK.
6.The authors are encouraged to further discuss how atmospheric stability modulates the relationship between vertical wind shear and UAV flight disturbances, particularly in the interpretation of Figure 12.
Response: Atmospheric stability does not act on UAV flight disturbance directly. Instead, it modulates the relationship between vertical wind shear and flight disturbance through the turbulent kinetic energy budget. Shear produces turbulence, while buoyancy enhances it under unstable stratification and suppresses it under stable stratification. However, how stability modulates the relationship between shear and turbulence is closely linked to local weather and thermal conditions, such as surface heating, cloud cover and the evolution of the boundary layer. The present study is considerably limited in this respect, because it is based on a small number of soundings at a single site under one weather situation. This question is therefore beyond what the present study can resolve.
References
Balsley, B. B., Svensson, G., & Tjernström, M. (2008). On the scale-dependence of the gradient Richardson number in the residual layer. Boundary-Layer Meteorology, 127(1), 57–72. https://doi.org/10.1007/s10546-007-9251-0
Cornman, L. B., Morse, C. S., & Cunning, G. (1995). Real-time estimation of atmospheric turbulence severity from in-situ aircraft measurements. Journal of Aircraft, 32(1), 171–177. https://doi.org/10.2514/3.46697
Davoudi, B., Taheri, E., Duraisamy, K., Jayaraman, B., & Kolmanovsky, I. (2020). Quad-rotor flight simulation in realistic atmospheric conditions. AIAA Journal, 58(5), 1992–2004. https://doi.org/10.2514/1.J058327
González-Rocha, J., De Wekker, S. F. J., Ross, S. D., & Woolsey, C. A. (2020). Wind profiling in the lower atmosphere from wind-induced perturbations to multirotor UAS. Sensors, 20(5), 1341. https://doi.org/10.3390/s20051341
Greene, B. R., Kral, S. T., Chilson, P. B., & Reuder, J. (2022). Gradient-based turbulence estimates from multicopter profiles in the Arctic stable boundary layer. Boundary-Layer Meteorology, 183, 321–353. https://doi.org/10.1007/s10546-022-00693-x
Guo, J., Miao, Y., Zhang, Y., Liu, H., Li, Z., Zhang, W., He, J., Lou, M., Yan, Y., Bian, L., & Zhai, P. (2016). The climatology of planetary boundary layer height in China derived from radiosonde and reanalysis data. Atmospheric Chemistry and Physics, 16, 13309–13319. https://doi.org/10.5194/acp-16-13309-2016
Houchi, K., Stoffelen, A., Marseille, G. J., & de Kloe, J. (2010). Comparison of wind and wind shear climatologies derived from high-resolution radiosondes and the ECMWF model. Journal of Geophysical Research, 115, D22123. https://doi.org/10.1029/2009JD013196
Kaimal, J. C., Wyngaard, J. C., Izumi, Y., & Coté, O. R. (1972). Spectral characteristics of surface-layer turbulence. Quarterly Journal of the Royal Meteorological Society, 98(417), 563–589. https://doi.org/10.1002/qj.49709841707
Palomaki, R. T., Rose, N. T., van den Bossche, M., Sherman, T. J., & De Wekker, S. F. J. (2017). Wind estimation in the lower atmosphere using multirotor aircraft. Journal of Atmospheric and Oceanic Technology, 34(5), 1183–1191. https://doi.org/10.1175/JTECH-D-16-0177.1
Shao, J., Zhang, J., Wang, W., Zhang, S., Yu, T., & Dong, W. (2023). Occurrence frequency of subcritical Richardson numbers assessed by global high-resolution radiosonde and ERA5 reanalysis. Atmospheric Chemistry and Physics, 23, 12589–12607. https://doi.org/10.5194/acp-23-12589-2023
Sharman, R. D., Cornman, L. B., Meymaris, G., Pearson, J., & Farrar, T. (2014). Description and derived climatologies of automated in situ eddy-dissipation-rate reports of atmospheric turbulence. Journal of Applied Meteorology and Climatology, 53(6), 1416–1432. https://doi.org/10.1175/JAMC-D-13-0329.1
Shelekhov, A., Afanasiev, A., Shelekhova, E., Kobzev, A., Tel’minov, A., Molchunov, A., & Poplevina, O. (2022). Low-altitude sensing of urban atmospheric turbulence with UAV. Drones, 6(3), 61. https://doi.org/10.3390/drones6030061
Wetz, T., Wildmann, N., & Beyrich, F. (2021). Distributed wind measurements with multiple quadrotor unmanned aerial vehicles in the atmospheric boundary layer. Atmospheric Measurement Techniques, 14(5), 3795–3814. https://doi.org/10.5194/amt-14-3795-2021
Wildmann, N., & Wetz, T. (2022). Towards vertical wind and turbulent flux estimation with multicopter uncrewed aircraft systems. Atmospheric Measurement Techniques, 15(18), 5465–5477.
-
RC2: 'Comment on egusphere-2026-4218', Anonymous Referee #2, 02 Oct 2026
This manuscript evaluates multi-rotor UAV measurements against a meteorological tower and examines how boundary-layer stratification and wind shear influence flight disturbances over the eastern Taihang piedmont. The study addresses an important observational need and provides valuable findings for atmospheric boundary-layer research and low-altitude meteorological applications. The combination of measurement validation and aircraft-response analysis is a strength, offering useful insights into environmental conditions associated with flight perturbations. The manuscript is generally well organized and relevant to the atmospheric observation community. Some clarification of measurement uncertainties, the Richardson-number calculation, and the interpretation of the turbulence index would further strengthen the study. Several mechanistic interpretations would also benefit from additional discussion or more cautious wording. Minor inconsistencies in dataset descriptions, figure annotations, and language could be corrected to improve clarity. Overall, I consider the manuscript suitable for publication after minor revisions.
Specific comments:
- The validation below 100 m is useful, although its applicability to the full 1500-m sounding range merits discussion. Please clarify sensor uncertainties, response delays, and residual rotor-flow effects, and assess their influence on gradients and Ri. Height-dependent validation statistics would help establish confidence in measurements at higher altitudes and under cloudy conditions.
- The all-pairs Ri calculation combines sequential observations and produces dependent samples. The authors are encouraged to examine its sensitivity to temporal variability, vertical spacing, and small shear denominators through comparison with layer-mean estimates. Please also discuss effective sample size and the mathematical dependence between Ri and shear in Figure 12.
- The turbulence index represents aircraft response and may also reflect flight-control behavior. Please clarify the consistency between Equation (5) and the magnitudes in Figures 7, 9, and 11, and describe filtering and averaging. Comparison with independent turbulence measurements, where available, would strengthen its interpretation.
- Ascent–descent differences may also reflect temporal evolution, horizontal displacement, and control effects. A discussion of these influences would strengthen the interpretation of atmospheric heterogeneity. Statements concerning cloud-base entrainment, moisture effects, resonance, and operational thresholds could be supported further or expressed more cautiously as hypotheses or case-specific findings.
- Section 2.2 reports 39 hovering datasets and 46 vertical profiles, whereas Figure 3 identifies 46 comparison profiles. Please reconcile these numbers and specify the datasets used in each analysis.
- Figure 5c describes signed shear as a “magnitude.” Please distinguish the signed wind-speed gradient from vector shear magnitude and ensure consistency among the equations, text, figure labels, and captions.
- “top and right marginal distributions ……” in Lines 450 to 451 is inconsistent with Figure 12.
- Correct “yileds” (line 244), “July 31th” (line 222), and duplicated punctuation (lines 364 and 368).
Citation: https://doi.org/10.5194/egusphere-2026-4218-RC2 -
AC2: 'Reply on RC2', Zhen Zhang, 02 Nov 2026
Response to reviewer#2’s comments
Referee #2’s comment:
This manuscript evaluates multi-rotor UAV measurements against a meteorological tower and examines how boundary-layer stratification and wind shear influence flight disturbances over the eastern Taihang piedmont. The study addresses an important observational need and provides valuable findings for atmospheric boundary-layer research and low-altitude meteorological applications. The combination of measurement validation and aircraft-response analysis is a strength, offering useful insights into environmental conditions associated with flight perturbations. The manuscript is generally well organized and relevant to the atmospheric observation community. Some clarification of measurement uncertainties, the Richardson-number calculation, and the interpretation of the turbulence index would further strengthen the study. Several mechanistic interpretations would also benefit from additional discussion or more cautious wording. Minor inconsistencies in dataset descriptions, figure annotations, and language could be corrected to improve clarity. Overall, I consider the manuscript suitable for publication after minor revisions.
Response: We sincerely thank the reviewer for the careful evaluation of our manuscript and for the positive assessment of its relevance to atmospheric boundary-layer research and low-altitude meteorological applications. We are glad that the reviewer found the combination of measurement validation and aircraft-response analysis to be a strength of the study. Following these suggestions, we have revised the relevant descriptions and corrected the inconsistencies in the dataset descriptions, figure annotations and language. For clarity purpose, here we have listed the reviewer's comments in bold font, followed by our response in blue plain font, and the modifications to the manuscript are in blue italics.
Specific comments:
1.The validation below 100 m is useful, although its applicability to the full 1500-m sounding range merits discussion. Please clarify sensor uncertainties, response delays, and residual rotor-flow effects, and assess their influence on gradients and Ri. Height-dependent validation statistics would help establish confidence in measurements at higher altitudes and under cloudy conditions.
Response: We appreciate the reviewer’s distinction between near-surface validation and the higher-altitude application. The benchmark comparison is limited to the tower levels (8–96 m AGL), whereas the attitude-response soundings extended to approximately 1500 m. We have revised Sections 2.1 and 3.1 to state explicitly that the reported biases, RMSEs and correlations establish performance only over the sampled tower range; they are not a validation at 1500 m or under cloud exposure. The sensors were laboratory-calibrated before deployment, and the meteorological package was located 0.6 m above the fuselage to reduce direct rotor-flow exposure. These precautions do not quantify residual downwash, ventilation dependence or the response delays of temperature and humidity sensors. We now explain that these effects may shift or smooth vertical gradients, and that the uncertainty can be amplified when those gradients are combined to form Ri. Since tower measurements end at 96 m, no independent height-dependent verification is available higher up.
2. The all-pairs Ri calculation combines sequential observations and produces dependent samples. The authors are encouraged to examine its sensitivity to temporal variability, vertical spacing, and small shear denominators through comparison with layer-mean estimates. Please also discuss effective sample size and the mathematical dependence between Ri and shear in Figure 12.
Response: Thank you for identifying an important interpretative limitation of the all-pairs approach. The computation remains exactly as described in Eqs. (1)–(4): observations at adjacent hovering levels are paired to retain a distribution of Ri values. In the revised Section 2.3, we now distinguish the number of pairwise combinations from the number of independent observations. Samples are reused across pairs and may be temporally autocorrelated; consequently, Ni × Ni+1 must not be treated as the effective independent sample size. We also explain that sequential hovering can confound height and time, finite layer spacing leaves unresolved vertical structure, and a small squared vector-shear denominator can amplify measurement and differencing errors. We attempted to calculate Ri using layer-mean values, but this approach substantially reduced the sample size. Given the small number of resulting estimates, we could not confidently assess the reliability of the results. Therefore, we did not conduct a comparison with layer-mean Ri in this revision; instead, we explicitly discuss the aforementioned limitations in Section 2.3. Finally, we make explicit that Ri and vector wind shear share the same gradients in Eq. (4), so their joint distribution in Figure 12 has a partly mathematical dependence and cannot, by itself, establish a physical coupling or causal threshold.
3.The turbulence index represents aircraft response and may also reflect flight-control behavior. Please clarify the consistency between Equation (5) and the magnitudes in Figures 7, 9, and 11, and describe filtering and averaging. Comparison with independent turbulence measurements, where available, would strengthen its interpretation.
Response: We thank the reviewer for this important comment and agree that the turbulence index represents the aircraft’s attitude response rather than a direct measurement of atmospheric turbulence. In the revised Section 2.3, T is defined as the instantaneous Euclidean magnitude of the recorded pitch and roll angular velocities and is described as a platform-specific attitude-response proxy, to which the aircraft dynamics, stabilization commands and flight phase can also contribute. It therefore has the same units as these angular velocities, namely °/s. The pitch and roll angles were sampled at 1 Hz, and the corresponding angular velocities were calculated using a central difference based on the angle measurements taken 1 s before and 1 s after each time point. calibrating T against such measurements would be required for a physical interpretation, and we will address this in future experiments.
4. Ascent–descent differences may also reflect temporal evolution, horizontal displacement, and control effects. A discussion of these influences would strengthen the interpretation of atmospheric heterogeneity. Statements concerning cloud-base entrainment, moisture effects, resonance, and operational thresholds could be supported further or expressed more cautiously as hypotheses or case-specific findings.
Response: We thank the reviewer for these helpful suggestions. The three case discussions place the ascent and descent tracks in the changing signed-shear background of Figure 5c. We now also state explicitly in Section 3.2 that the ascent and descent passes were separated by several minutes and need not have sampled identical air. Temporal evolution, possible horizontal displacement and differences in autopilot correction may all contribute to the asymmetric response, and the available observations do not separate their effects; the statements linking background shear to the asymmetry have been revised accordingly. The lower turbulence index during UAV ascent compared with descent is likely attributable to the control actions of the flight-control software. The cloud-base entrainment explanation has been changed from a demonstrated process to a plausible interpretation, and the humidity-related statements in Section 3.2 and the Concluding remarks no longer exclude moisture or cloud effects. For Figure 12, the localized bin of elevated index is presented as an observation specific to this dataset rather than as evidence of eddy–airframe resonance, for which spectral or airframe-frequency evidence is not available. References to quantitative operational thresholds in the research objectives and the Concluding remarks have been replaced with case-specific conditions associated with increased aircraft response.
5.Section 2.2 reports 39 hovering datasets and 46 vertical profiles, whereas Figure 3 identifies 46 comparison profiles. Please reconcile these numbers and specify the datasets used in each analysis.
Response: We thank the reviewer for pointing out this inconsistency. The campaign comprises 46 vertical profiles in total: 39 obtained from the tower–UAV comparison experiments at the Xingtai base (July 26–30) and 7 obtained from the attitude disturbance experiments in Qianxian Township (July 31). Each dataset corresponds to one vertical profile, i.e. one ascent or descent together with the hovering measurements at each level; the original text described the same datasets as both “hovering datasets” and “vertical profile datasets”, which caused the confusion. The label “Profile number = 46” in the originally submitted Figure 3 was an error, because Figure 3 is based only on the 39 comparison profiles; it has been corrected to 39, and the statistics in the figure are unchanged. Section 2.2 now uses one term throughout, states the composition of the 46 profiles, and specifies that the tower–UAV comparisons in Figures 3 and 4 use the 39 Xingtai profiles and that the three evening cases analysed in Section 3.2 were selected from the 7 Qianxian profiles.
6.Figure 5c describes signed shear as a “magnitude.” Please distinguish the signed wind-speed gradient from vector shear magnitude and ensure consistency among the equations, text, figure labels, and captions.
Response: We agree. Figure 5c contains positive and negative values and therefore cannot be described as a shear magnitude. Its shaded field is now explicitly called the signed vertical gradient of horizontal wind speed (signed shear): positive where wind speed increases with height and negative where it decreases. This differs from the non-negative magnitude of the vertical gradient of the horizontal wind vector, whose square forms the denominator of Eq. (4) and which is used in the shear–Ri analysis of Figure 12. We have revised the Figure 5 and Figure 12 captions, the description of Figure 5c in Section 3.2 and the discussion of Figure 12 in Section 3.3 accordingly. In Section 2.3, the text introducing Eqs. (1)–(3) now refers to the vertical gradients of the two horizontal wind components rather than to wind shear. The Figure 5 caption also states that the colour bar labelled “Wind shear” shows this signed shear. The unit printed on that colour bar (m·s⁻¹) is a typographical error; the correct unit is s⁻¹, as given in the caption.
7.“top and right marginal distributions ……” in Lines 450 to 451 is inconsistent with Figure 12.
Response: Thank you for catching this description error. In Figure 12, the marginal strip showing the mean TI for each Ri bin is at the bottom, and the strip showing the mean TI for each wind-shear bin is on the left. The sentence in Section 3.3 has been corrected to refer to the “bottom and left” marginal distributions, and it now states explicitly which quantity each strip shows. The later reference to the “top” marginal distribution has been removed, together with the specific numerical TI values (see our response to Comment 3).
8.Correct “yileds” (line 244), “July 31th” (line 222), and duplicated punctuation (lines 364 and 368).
Response: Thank you. All the typographical errors and duplicated punctuation have been corrected as suggested.
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 361 | 216 | 44 | 621 | 38 | 33 |
- HTML: 361
- PDF: 216
- XML: 44
- Total: 621
- BibTeX: 38
- EndNote: 33
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
General Evaluation
This study employs a DJI M350 RTK multi-rotor UAV to investigate the reliability of atmospheric boundary layer (ABL) observations and flight safety over complex terrain. The authors conducted comparative observations against a 100-m gradient meteorological tower and vertical-profile surveys over the eastern Taihang piedmont. They first demonstrated that the UAV platform can accurately measure both basic meteorological variables (temperature, wind, humidity) and derived stability parameters (potential temperature and gradient Richardson number). Furthermore, the study quantitatively reveals the combined effects of vertical wind shear and stratification stability on UAV attitude disturbances, and proposes a possible vortex-airframe resonance-like mechanism, which is a novel contribution to UAV-based meteorological applications. Notably, the paper employs a sample-pairwise finite-difference method for Ri calculation, which preserves the distribution information of parameters between stratified layers and better captures turbulence details than conventional layer-averaging approaches.
The topic is well within the scope of AMT. The experimental design is rigorous, the dataset is substantial, and the analytical framework is logically structured. The findings provide valuable references for improving UAV-based ABL observations and low-altitude flight safety over complex terrain. The manuscript is well written and the figures are of high quality. Before formal acceptance, I recommend that the authors address the following specific issues.
Scientific comments
1. The logical chain from observational results to conclusions needs to be strengthened
The authors describe the time-height evolution of the large-scale wind shear field based on Figure 5c, and further analyze the shear characteristics during three flight periods (18:24-18:43, 18:55-19:22, and 19:43-19:55 BJT) in Section 3.2 (starting from page 12, line 328; page 13, line 365; and page 14, line 400, respectively). However, in the detailed discussion of these three cases, the authors do not explicitly link the temporal evolution of the large-scale shear field revealed in Figure 5c with the vertical distribution characteristics of UAV attitude responses during each period.
For example, during the first flight (18:24-18:43), which positive/negative large-scale shear zones did the ascending and descending segments (black solid and dashed lines in Figure 5c) traverse? Could the difference in this large-scale background shear be a significant cause of the asymmetry in flight disturbances between the ascending and descending profiles (Figure 7)? I suggest that the authors actively reference Figure 5c when discussing each case, and superimpose the flight trajectories onto the large-scale shear background for joint analysis. This would significantly strengthen the logical chain connecting the macro-scale background to the local disturbance responses.
2. The evidence supporting the "resonance-like amplification" hypothesis needs to be reinforced
On pages 16-17, lines 465-470, the authors propose a "resonance-like amplification" conjecture based on Figure 12. This is an interesting speculation, but it is currently supported only by anomalous data points in a specific regime (vertical wind shear approximately 0.007-0.010 s⁻¹ combined with near-zero to negative Ri). The evidence appears somewhat thin. I recommend that the authors: (1) adopt a more cautious tone when presenting this viewpoint, clearly stating that this is a preliminary hypothesis requiring further verification; and (2) if possible, provide additional supportive analysis—for instance, by roughly estimating the characteristic scale of turbulent eddies under this specific shear range and discussing whether this scale is comparable in magnitude to the airframe dimensions or the natural frequencies of the MATRICE 350 RTK in pitch/roll directions. If such analysis cannot be provided, I suggest relocating this discussion to the end of the Discussion section as an outlook for future research directions.
3. Inconsistency in the description of cloud-layer vertical extent
On page 15, line 405, the authors state "in the cloud layer (approximately 900-1200 m) and outside the cloud layer." However, on page 16, line 437, they mention "In the stable cloud-top layer above 1000 m." There is ambiguity and inconsistency between the description of the "cloud layer" (900-1200 m) and the "cloud-top layer" (above 1000 m). I suggest that the authors clarify and unify the vertical distribution and extent of the cloud layer, as this directly affects the accurate interpretation of the conclusion that "stable stratification suppresses disturbances."
Minor Comments:
1. The terms "flight bumpiness" (e.g., page 2, line 49; page 14, line 391) and "flight disturbance" (e.g., page 2, line 33; page 16, line 461) are used interchangeably throughout the manuscript. I recommend standardizing to "flight disturbance" or the more precise term "attitude disturbance" to maintain terminological consistency.
2. Figure formatting details:
1)In Figures 4, 6, 8, and 10, the variable names on the x-axes (e.g., theta, Ri , etc ) should be by mathematical variable formatting and italicized to comply with conventions, like: θ and Ri.
2)In the captions of Figures 7, 9, and 11, the repeated use of "profile" could be simplified; for example, change to "Same as Figure 6, but for the vertical response of UAV attitude parameters during ...".