the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 27 Sep 2023
Experimental evaluation of the motion-induced effects on turbulent fluctuations measurement on floating lidar systems
Abstract. This study investigates the impact of motion on the line-of-sight (LOS) turbulent velocity fluctuations derived from lidar profiler measurements. Onshore tests were conducted using a WindCube v2.1 lidar, referred to as the mobile lidar, mounted on a hexapod to simulate buoy motion, with a fixed lidar used as a reference. To assess the motion-induced effects on turbulent velocity fluctuations measured by floating lidar systems, the root-mean-square error (RMSE) of LOS velocity fluctuations obtained from the fixed and mobile lidars was calculated. A comprehensive wind dataset spanning 45 hours was analyzed, with a focus on regular motions involving single-axis rotations and combinations of rotations around multiple axes. The investigation of single-axis rotations revealed that the primary influencing factor on the results was the alignment between wind direction and the axis of rotation. The highest RMSE values occurred when winds propagated perpendicular to the rotation axis, resulting in pitch motion, whereas the lowest RMSE values were observed when wind propagated along the rotation axis, leading to roll motion. Furthermore, yaw motion was found to increase the RMSE compared to scenarios without yaw motion. Moreover, the addition of motion around extra axes of rotation was found to increase RMSE. High wind speed emerged as a significant driver of RMSE, with higher velocities leading to higher RMSE values. The study also indicated that the role of wind shear in influencing RMSE of LOS velocity fluctuations requires further investigation. Additionally, the study explored the impact of motion period, revealing that motion frequencies affect the LOS velocity spectra within the expected inertial sub-range. However, the impact on RMSE was found to be limited in comparison to the amplitude, wind direction, and wind speed.
This preprint has been withdrawn.
-
Withdrawal notice
This preprint has been withdrawn.
-
Preprint
(5580 KB)
Interactive discussion
Status: closed
-
RC1: 'Comment on wes-2023-126', Anonymous Referee #1, 08 Oct 2023
General comments:
The submitted manuscript "Experimental evaluation of the motion-induced effects on turbulent fluctuations measurement on floating lidar systems" by Thebault et al. describes a study where two WindCube profiling wind lidar units were located next to each other. One of the two lidars was mounted on a hexapod platform so that it was subject to controlled motion while wind measurements were taken. The radial wind speed data from both units were compared and differences are attributed to the effects of motion.The study provides some novelty in so far that controlled-motion experiments of wind lidar are not often described in literature. And the study design is successful in achieving statistically relevant results because each motion sequence is performed in a variety of wind conditions, so that the effects of motion cases can be analysed independently from the prevailing wind conditions.
Unfortunately, the depth of analysis in the current version of the manuscript is too low. Many effects are described without substantial interpretation and several aspects lack a conclusion supported by the findings of the study. The presented experiment is valuable for the FLS sector but its potential is not well used by Thebault et al. I am missing any kind of simulation, model or at least some theoretical assumptions that predict the results. Such predictions could be validated by the experiment and in a future study be used to extend the findings to more motion cases. I will detail this in the specific comments.
I recommend to reconsider the manuscript for publication in WES only after a complete revision. A revised manuscript must demonstrate a deeper understanding of the results based on theoretical considerations of how a profiling wind lidar samples radial velocities under the influence of motion. Without this, the scientific quality is insufficient for publication in WES.
Specific comments:
Abstract:
The abstract should clarify that the study is about "wind lidar measurements" to make it easier for the reader to find out if the study is relevant to them. Especially because the title of the manuscript does not give this information.Introduction:
The introduction should differentiate better between FLS measurements of mean wind statistics and turbulence characteristics (e.g. l. 22) and research findings should be reported more precisely (e.g. l. 31, cross-contamination can lead to both over- or underestimation of turbulence).
Controversial or unclear statements that do not built up on a reference should be avoided in the introduction (e.g., l. 38, "high-frequency (in the range of the wave frequencies)", ll. 44-45 "...turbulence intensity is commonly assessed by calculating the variance of the three (!) velocity components...".)
The authors should describe the most important findings of the cited literature and how they influence their own work instead of simply listing references (l. 40, l.43).
A large fraction of the introduction describes studies that have a focus on mean wind statistics (ll. 62-77) without elaborating how these studies relate to the present work which presents no mean data.
The introduction is missing a paragraph that guides the reader through the paper while demonstrating its structure of sections.Data and method:
l. 102: More detail must be given on the sampling pattern of the "prototype configuration". For how long does each beam sample in each direction within its 1Hz scanning cycle? What is the zenith angle of the beams? What are the range gates?
l. 111: More information needs to be given on the scenario with "coupled motion". Are pitch and roll motion in phase or with a phase shift or are the motions performed consecutively? This is crucial for the interpretation of the results.
l. 117: Is the availability based on 10 minute averages? It is hard to believe that none of the LOS measurements were invalid. What CNR and packet count thresholds have been used to reach 100% availability at all heights? It would have been interesting to investigate the results in challenging atmospheric conditions with bad CNR values.
Table 2: "Corresponding Scenario" descriptions are inconsistent (e.g., S2 T=4s "Typical large buoy" but also S4 T=6s "Large (or spar) buoy" or S3 and S5 have different periods but the same description and what does "weak response above" mean?). Typical commercial FLS have tilt response periods of around 3s. Periods above 4 seconds are rather found for larger platforms.
l. 127: Commercial FLS usually use simple single point moorings. Thus, no valid evidence is provided for the important statement that the motion characteristics of the MONABIOP buoy are representative for the motion of FLS.
l. 135: Which "specific conditions"? This description is too general and does not add any value.
l. 144: RMSE is used as the key parameter in this study. Its definition must be given by a well-described formula.
l. 145: Which method was used for getting the "mean-detrended signal"?Results:
l. 155: It would be much easier for the reader to see how well the low frequency component (!) of the time series align if they were given in the same plot instead of in two separate plots.
Fig. 4: Instead of presenting and comparing data from two separate plots the data from both lidars should be presented in one plot (maybe just one or two beams). A second plot could then be used for a zoomed-in section that shows also the motion period for comparison.
l. 163: The authors note that the standard deviation of LOS velocity fluctuations is 70% higher for the mobile lidar than for the fixed lidar. But as a reader I am missing an interpretation of this value. Is this what is expected from the beam rotation with the chosen amplitudes? In a completely homogeneous wind field of a fixed wind speed, the effect of pitch rotation on a single beam can be estimated. It would be of outmost interest to compare the experimental fluctuations with these theoretical fluctuations.
l. 167: "nearly three times higher". I think in l. 174, the authors write about the same numbers that they are "more than four times higher".
l. 170: The analysis of the impact of motion amplitude is missing a comparison of RMSE in the absence of motion. The beams of both lidar units will not measure identical LOS speeds even when they are both standing still (not synchronized, not the same air volumes, not exactly the same angles, random measurement error...). This analysis could also be added to a previous section.
l. 174: Without a simulation model, the interpretation of the results is superficial. The authors state a "linear increase" of RMSE with wind speed. A FLS model will probably show that this statement is only true in the absence of translatory motion of the lidar telescope which is introduced here by its rigid body rotation around a non-zero lever. So, the linear relationship between RMSE and wind speed is only an approximation that should be assessed critically.
l. 185: What could be a reason for the higher spectral energy at low frequencies measured by the fixed lidar? This is a crucial finding that must be investigated further because it has an impact on the RMSE values. It is insufficient to conclude with "distinct characteristics in the spectral energy profiles".
Fig. 8: The spectra should contain a vertical line at the frequencies that correspond to the motion periods. Otherwise it is difficult to analyse the spectral peaks.
3.5: The authors present some good hypothesis for why measurements at high elevations show higher RMSE values. Also here, a model framework and an in-depth analysis of example time series would help to quantify the influence of the different sources of added RMSE. Without it the findings are inconclusive.
3.6: This section lacks a conclusion. How does wind shear influence the RMSE of the mobile lidar?
l. 223: Why does this observation hold only in the range from 12 to 14 m/s? The mean values of all wind speeds shows a different order (dashed horizontal lines in Fig. 11, Ry > Ry+Rz). Without further explanation, it is not convincing that "the rotation around the vertical z-axis is not negligible in terms of RMSE".Discussion:
The weaknesses of the manuscript mentioned above result in a discussion that is superficial for the most part (e.g., l. 265 "wind speed is one of the main driver[s] of the RMSE". This is obvious from theoretical considerations even without any experiment, l. 269 "no clear evidence of the role of wind shear could have been demonstrated."). Other statements are not covered by the evidence presented in the previous sections (e.g. ll. 257-264, findings regarding the frequency dependency of effects of translational motion are only valid for reconstructed wind vectors, not for RMSE of LOS beams.).
Other sentences are confusing or wrong (e.g. ll. 282-285: small buoys, small amplitudes and large buoys, large amplitudes). The recommendation to use small buoys as FLS platform to reduce motion induced errors is likewise misleading.Conclusions:
l. 325: It remains unclear why the conclusions of the study are more significant for France than for other countries with similar offshore wind potential.Technical corrections:
The manuscript contains several minor errors and has room for stylistic improvement. A revised version should be proofread carefully before submission.Citation: https://doi.org/10.5194/wes-2023-126-RC1 -
AC1: 'Reply on RC1', Maxime Thiébaut, 17 Nov 2023
The comment was uploaded in the form of a supplement: https://wes.copernicus.org/preprints/wes-2023-126/wes-2023-126-AC1-supplement.pdf
-
AC1: 'Reply on RC1', Maxime Thiébaut, 17 Nov 2023
-
RC2: 'Comment on wes-2023-126', Anonymous Referee #2, 14 Oct 2023
Dear authors,
Thanks for the manuscript. I think in general the manuscript is well written, coherent and the contents relate to a very interesting topic, which is that of the floating lidar turbulence measurements. However, although the manuscript presents a quite interesting dataset that can be used to analyze the impact of motion-induced effects on lidar turbulence, I think that the manuscript reads more as a technical report than a journal paper. Below I will provide some general and specific comments with respect to this and different aspects of the study.
General comments
- As I mentioned, right now we are reading an interesting technical report but not a research paper. The reader is not gaining anything new from the paper as the data and some analyses are presented without further investigation. The authors mentioned that they are gonna propose a motion-compensation algorithm based on this dataset. I think that that is what this report needs to have potential for a paper, so I encourage the authors to start the paper by the planned algorithm to compensate for motion and investigate its goodness using this dataset.
- Line 31: it is nice you are aware that there are two main errors for lidar turbulence measures (most people do not know this) but it is also strange that you think that the cross-contamination effect always results in an overestimation. This is not the case always. If by some reason your compensation algorithm assumes always an overestimation due to cross-contamination, you need to review it deeply.
- Table 3: I am not sure of the value of this table. The “degree” of deviation of the mobile lidar turbulence compared to the fixed lidar turbulence should be both turbulence and scanning-configuration dependent. Here you seem to average across all cycles which I guess have different turbulence characteristics so you are kind of averaging apples with oranges.
- One important question: did you by chance make a cycle without motion at all? That would be interesting to have as part of the analysis to know whether there is an inherent bias between the units
- Section 3.3/Figure 8: there should not be that much difference between the spectra of the mobile and the fixed lidar (particularly at the large scales) apart from the area around the peak at the specific period. Why is it different (see my previous comment)? Maybe some error bands could show that these differences are not significant as they seem to be?
- Lines 309—314: these lines cannot be part of the conclusions. You have not described the motion-compensation algorithm and you are here giving us hints of what it can do. As mentioned in my first comment, I suggest you start this manuscript by proposing/explaining the algorithm and then you evaluate it by comparison with this nice dataset.
Specific comments
- Line 25: Delete an extra “)”.
- Line 33: Delete “in most cases,… However,”.
- Line 33: Replace “do align” by “compensate each other”.
- Introduction: Peña et al. (2022) presented another way to assess the impact of motion on floating lidars that you would like to check.
- Lines 163—166 are also “strange” ways to measure these impacts (see general comment 3).
- Line 175: “latter metric”… it is difficult to understand what do you mean here
References:
Peña A., Mann J., Angelou N., and Jacobsen A. (2022) A motion-correction method for turbulence estimates from floating lidars. Remote Sensing, 14, 6065
Citation: https://doi.org/10.5194/wes-2023-126-RC2 -
AC2: 'Reply on RC2', Maxime Thiébaut, 17 Nov 2023
The comment was uploaded in the form of a supplement: https://wes.copernicus.org/preprints/wes-2023-126/wes-2023-126-AC2-supplement.pdf
Interactive discussion
Status: closed
-
RC1: 'Comment on wes-2023-126', Anonymous Referee #1, 08 Oct 2023
General comments:
The submitted manuscript "Experimental evaluation of the motion-induced effects on turbulent fluctuations measurement on floating lidar systems" by Thebault et al. describes a study where two WindCube profiling wind lidar units were located next to each other. One of the two lidars was mounted on a hexapod platform so that it was subject to controlled motion while wind measurements were taken. The radial wind speed data from both units were compared and differences are attributed to the effects of motion.The study provides some novelty in so far that controlled-motion experiments of wind lidar are not often described in literature. And the study design is successful in achieving statistically relevant results because each motion sequence is performed in a variety of wind conditions, so that the effects of motion cases can be analysed independently from the prevailing wind conditions.
Unfortunately, the depth of analysis in the current version of the manuscript is too low. Many effects are described without substantial interpretation and several aspects lack a conclusion supported by the findings of the study. The presented experiment is valuable for the FLS sector but its potential is not well used by Thebault et al. I am missing any kind of simulation, model or at least some theoretical assumptions that predict the results. Such predictions could be validated by the experiment and in a future study be used to extend the findings to more motion cases. I will detail this in the specific comments.
I recommend to reconsider the manuscript for publication in WES only after a complete revision. A revised manuscript must demonstrate a deeper understanding of the results based on theoretical considerations of how a profiling wind lidar samples radial velocities under the influence of motion. Without this, the scientific quality is insufficient for publication in WES.
Specific comments:
Abstract:
The abstract should clarify that the study is about "wind lidar measurements" to make it easier for the reader to find out if the study is relevant to them. Especially because the title of the manuscript does not give this information.Introduction:
The introduction should differentiate better between FLS measurements of mean wind statistics and turbulence characteristics (e.g. l. 22) and research findings should be reported more precisely (e.g. l. 31, cross-contamination can lead to both over- or underestimation of turbulence).
Controversial or unclear statements that do not built up on a reference should be avoided in the introduction (e.g., l. 38, "high-frequency (in the range of the wave frequencies)", ll. 44-45 "...turbulence intensity is commonly assessed by calculating the variance of the three (!) velocity components...".)
The authors should describe the most important findings of the cited literature and how they influence their own work instead of simply listing references (l. 40, l.43).
A large fraction of the introduction describes studies that have a focus on mean wind statistics (ll. 62-77) without elaborating how these studies relate to the present work which presents no mean data.
The introduction is missing a paragraph that guides the reader through the paper while demonstrating its structure of sections.Data and method:
l. 102: More detail must be given on the sampling pattern of the "prototype configuration". For how long does each beam sample in each direction within its 1Hz scanning cycle? What is the zenith angle of the beams? What are the range gates?
l. 111: More information needs to be given on the scenario with "coupled motion". Are pitch and roll motion in phase or with a phase shift or are the motions performed consecutively? This is crucial for the interpretation of the results.
l. 117: Is the availability based on 10 minute averages? It is hard to believe that none of the LOS measurements were invalid. What CNR and packet count thresholds have been used to reach 100% availability at all heights? It would have been interesting to investigate the results in challenging atmospheric conditions with bad CNR values.
Table 2: "Corresponding Scenario" descriptions are inconsistent (e.g., S2 T=4s "Typical large buoy" but also S4 T=6s "Large (or spar) buoy" or S3 and S5 have different periods but the same description and what does "weak response above" mean?). Typical commercial FLS have tilt response periods of around 3s. Periods above 4 seconds are rather found for larger platforms.
l. 127: Commercial FLS usually use simple single point moorings. Thus, no valid evidence is provided for the important statement that the motion characteristics of the MONABIOP buoy are representative for the motion of FLS.
l. 135: Which "specific conditions"? This description is too general and does not add any value.
l. 144: RMSE is used as the key parameter in this study. Its definition must be given by a well-described formula.
l. 145: Which method was used for getting the "mean-detrended signal"?Results:
l. 155: It would be much easier for the reader to see how well the low frequency component (!) of the time series align if they were given in the same plot instead of in two separate plots.
Fig. 4: Instead of presenting and comparing data from two separate plots the data from both lidars should be presented in one plot (maybe just one or two beams). A second plot could then be used for a zoomed-in section that shows also the motion period for comparison.
l. 163: The authors note that the standard deviation of LOS velocity fluctuations is 70% higher for the mobile lidar than for the fixed lidar. But as a reader I am missing an interpretation of this value. Is this what is expected from the beam rotation with the chosen amplitudes? In a completely homogeneous wind field of a fixed wind speed, the effect of pitch rotation on a single beam can be estimated. It would be of outmost interest to compare the experimental fluctuations with these theoretical fluctuations.
l. 167: "nearly three times higher". I think in l. 174, the authors write about the same numbers that they are "more than four times higher".
l. 170: The analysis of the impact of motion amplitude is missing a comparison of RMSE in the absence of motion. The beams of both lidar units will not measure identical LOS speeds even when they are both standing still (not synchronized, not the same air volumes, not exactly the same angles, random measurement error...). This analysis could also be added to a previous section.
l. 174: Without a simulation model, the interpretation of the results is superficial. The authors state a "linear increase" of RMSE with wind speed. A FLS model will probably show that this statement is only true in the absence of translatory motion of the lidar telescope which is introduced here by its rigid body rotation around a non-zero lever. So, the linear relationship between RMSE and wind speed is only an approximation that should be assessed critically.
l. 185: What could be a reason for the higher spectral energy at low frequencies measured by the fixed lidar? This is a crucial finding that must be investigated further because it has an impact on the RMSE values. It is insufficient to conclude with "distinct characteristics in the spectral energy profiles".
Fig. 8: The spectra should contain a vertical line at the frequencies that correspond to the motion periods. Otherwise it is difficult to analyse the spectral peaks.
3.5: The authors present some good hypothesis for why measurements at high elevations show higher RMSE values. Also here, a model framework and an in-depth analysis of example time series would help to quantify the influence of the different sources of added RMSE. Without it the findings are inconclusive.
3.6: This section lacks a conclusion. How does wind shear influence the RMSE of the mobile lidar?
l. 223: Why does this observation hold only in the range from 12 to 14 m/s? The mean values of all wind speeds shows a different order (dashed horizontal lines in Fig. 11, Ry > Ry+Rz). Without further explanation, it is not convincing that "the rotation around the vertical z-axis is not negligible in terms of RMSE".Discussion:
The weaknesses of the manuscript mentioned above result in a discussion that is superficial for the most part (e.g., l. 265 "wind speed is one of the main driver[s] of the RMSE". This is obvious from theoretical considerations even without any experiment, l. 269 "no clear evidence of the role of wind shear could have been demonstrated."). Other statements are not covered by the evidence presented in the previous sections (e.g. ll. 257-264, findings regarding the frequency dependency of effects of translational motion are only valid for reconstructed wind vectors, not for RMSE of LOS beams.).
Other sentences are confusing or wrong (e.g. ll. 282-285: small buoys, small amplitudes and large buoys, large amplitudes). The recommendation to use small buoys as FLS platform to reduce motion induced errors is likewise misleading.Conclusions:
l. 325: It remains unclear why the conclusions of the study are more significant for France than for other countries with similar offshore wind potential.Technical corrections:
The manuscript contains several minor errors and has room for stylistic improvement. A revised version should be proofread carefully before submission.Citation: https://doi.org/10.5194/wes-2023-126-RC1 -
AC1: 'Reply on RC1', Maxime Thiébaut, 17 Nov 2023
The comment was uploaded in the form of a supplement: https://wes.copernicus.org/preprints/wes-2023-126/wes-2023-126-AC1-supplement.pdf
-
AC1: 'Reply on RC1', Maxime Thiébaut, 17 Nov 2023
-
RC2: 'Comment on wes-2023-126', Anonymous Referee #2, 14 Oct 2023
Dear authors,
Thanks for the manuscript. I think in general the manuscript is well written, coherent and the contents relate to a very interesting topic, which is that of the floating lidar turbulence measurements. However, although the manuscript presents a quite interesting dataset that can be used to analyze the impact of motion-induced effects on lidar turbulence, I think that the manuscript reads more as a technical report than a journal paper. Below I will provide some general and specific comments with respect to this and different aspects of the study.
General comments
- As I mentioned, right now we are reading an interesting technical report but not a research paper. The reader is not gaining anything new from the paper as the data and some analyses are presented without further investigation. The authors mentioned that they are gonna propose a motion-compensation algorithm based on this dataset. I think that that is what this report needs to have potential for a paper, so I encourage the authors to start the paper by the planned algorithm to compensate for motion and investigate its goodness using this dataset.
- Line 31: it is nice you are aware that there are two main errors for lidar turbulence measures (most people do not know this) but it is also strange that you think that the cross-contamination effect always results in an overestimation. This is not the case always. If by some reason your compensation algorithm assumes always an overestimation due to cross-contamination, you need to review it deeply.
- Table 3: I am not sure of the value of this table. The “degree” of deviation of the mobile lidar turbulence compared to the fixed lidar turbulence should be both turbulence and scanning-configuration dependent. Here you seem to average across all cycles which I guess have different turbulence characteristics so you are kind of averaging apples with oranges.
- One important question: did you by chance make a cycle without motion at all? That would be interesting to have as part of the analysis to know whether there is an inherent bias between the units
- Section 3.3/Figure 8: there should not be that much difference between the spectra of the mobile and the fixed lidar (particularly at the large scales) apart from the area around the peak at the specific period. Why is it different (see my previous comment)? Maybe some error bands could show that these differences are not significant as they seem to be?
- Lines 309—314: these lines cannot be part of the conclusions. You have not described the motion-compensation algorithm and you are here giving us hints of what it can do. As mentioned in my first comment, I suggest you start this manuscript by proposing/explaining the algorithm and then you evaluate it by comparison with this nice dataset.
Specific comments
- Line 25: Delete an extra “)”.
- Line 33: Delete “in most cases,… However,”.
- Line 33: Replace “do align” by “compensate each other”.
- Introduction: Peña et al. (2022) presented another way to assess the impact of motion on floating lidars that you would like to check.
- Lines 163—166 are also “strange” ways to measure these impacts (see general comment 3).
- Line 175: “latter metric”… it is difficult to understand what do you mean here
References:
Peña A., Mann J., Angelou N., and Jacobsen A. (2022) A motion-correction method for turbulence estimates from floating lidars. Remote Sensing, 14, 6065
Citation: https://doi.org/10.5194/wes-2023-126-RC2 -
AC2: 'Reply on RC2', Maxime Thiébaut, 17 Nov 2023
The comment was uploaded in the form of a supplement: https://wes.copernicus.org/preprints/wes-2023-126/wes-2023-126-AC2-supplement.pdf
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 320 | 92 | 17 | 429 | 12 | 8 |
- HTML: 320
- PDF: 92
- XML: 17
- Total: 429
- BibTeX: 12
- EndNote: 8
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1