Acoustic Tomography of the Atmosphere: A Large-Eddy Simulation Sensitivity Study

Round 1

Reviewer 1 Report (Previous Reviewer 2)

Comments and Suggestions for Authors

This paper has undergone extensive revisions from its initial draft, significantly enhancing the content quality. However, many opinion-based discussions in the article require additional references to both support the author's arguments and facilitate readers' extended research. For instance, while Doppler lidar is mentioned in Lines 635-637, no relevant technical details are provided. 

The authors may instead consider exploring prior research on:

 

• Turbulent energy dissipation rate estimation methods using Doppler lidar
• Impacts of scanning strategies on lidar-based turbulence measurements
• Eddy dissipation characteristics in the atmospheric boundary layer

 

Overall, this revised manuscript has addressed all my concerns satisfactorily. I recommend acceptance after implementing the minor revisions suggested above.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report (Previous Reviewer 3)

Comments and Suggestions for Authors

I appreciate the authors' addressing the reviewer comments. Following the changes, the manuscript improved significantly. After addressing the minor comments listed below, it is acceptable for publication.

 

Comment 1: Missing reference to Figures 3 and 14 in the text. 

Comment 2: In Figure 14, what does the red line represent? I suggest adding the information about the colors in the figure caption. They are present in the text, but the conventional way is to include such details in the caption.  

Comment 3: Could the authors expand the timeseries in Figure 3 to show the ramp-up of turbulence and it reaching the quasi-steady state? It would be helpful to see how quickly the turbulence ramps for each of the stability conditions.

Comment 4: The y-axis for the neutral temperature standard deviations in Figure 10 seems incorrectly labeled as occurrences.

Comment 5: The Authors mentioned the operational limits of the algorithm in the abstract and the conclusions, but did not provide any specific limits in the text. 

 

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 3 Report (Previous Reviewer 1)

Comments and Suggestions for Authors

I value the scientific contribution of this research work. By employing LES simulation, the capability of AT and the theoretical computation of turbulent parameters are validated. I'm glad to see that the revised manuscript is improved quite significantly. The paper contains numerous equations that are beyond my ability for sanity check. I have the following minor comments for your consideration.

1. The abstract should include major scientific findings or achievements of the study.
2. There are excessive equations presented but with little physical explanation.
3.  Line 229: "a constant 300-K layer above ground level (AGL)". I guess this is for potential temperature. Please add the information.
4. Line 322: change "squares" to "square".
5. Figure 5 caption: change "neutral ABL and convective ABL case" to "neutral ABL (a) and convective ABL (b) case.
      

Author Response

Please see the attachment.

Author Response File: Author Response.docx

This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Measuring ABL turbulences are very important to understand the quantitative properties of the turbulences and improve forecasting of TIs for wind energy and aviation activities. Nevertheless, it is extremely difficult to detect these micro-scale eddies directly. This work, by reviving an AT facility, conducting carefully configured  super-high resolution LES modeling, and refining the time-dependent stochastic in- version (TDSI), renovates the great potential of the AT technology for extending theoretical research and wind energy applications. Overall, this is a piece of very interesting and valuable work.  The presentation is clear and the conclusions are with good supporting evidences and discussions. However, I found several writing issues that need to be clarified before accepting for publication.

1) In several places though the manuscript, the authors simply refer the acoustic observations as to "data" (in many places). I guess the "data" that the authors intended is the time for the sound waves traveling from a tower to another. I think it would be better just to name the data directly to it to avoid the confusion. 

2)  It is not clear how the observation, i.e., the "data",  of the 8 towers are used even though the whole Section 2 is devoted to describe the theoretical formulation.     

3) Several symbols were used without definitions and some of them are critical for gaining good understanding. For example, r_i  in Eq (1); the subscript i?  Especially, in Line 90? Line 98: what "models"? Eq (9): both tao and N were not defined. Please fix these and the other similar cases.

4) Figure 2 looks unclear and should be improved. 

5) In refining the LES simulation grids, the authors conducted three stages of modeling. Data sample for correlation computation is 1800s (Line 207). Is it between 0 - 1800s in Stage 3?  Line 237: t = 150 s. 150 s into the Stage 3? Line 176: the Stage 3 is run for 21800s. Many run times are not useful?

6) Line 56, "two LESs of NREL’s AT array"? Please double check?

7) Line 233 "It represents a least squares estimation between the LES fields and the reconstructed fields."? Please check the meaning.

8) Line 335, "with the dashed black line representing the LES benchmark PSD". Please double check this sentence for validity.

9) Line 344: "approximately 0.15 Hz, corresponding to a spatial resolution limit of about 1.4 m". Here, I could not understand - with the LES model resolution of 1.25m (Line 171), how the LES results extend far to the right of the vertical black dashed line? 

 

        

Comments on the Quality of English Language

The English presentation is overall high quality. 

Reviewer 2 Report

Comments and Suggestions for Authors

General comments

 

This study uses large eddy simulation to evaluate the performance of a time-dependent stochastic inversion algorithm in acoustic tomography, showing strong results under convective conditions but reduced temperature fluctuation accuracy in neutral conditions. A sensitivity analysis reveals a spatial resolution limit of about 1.4 m and a tolerance for travel-time errors below 0.002 s across both conditions. The article states, “The algorithm demonstrated strong performance under convective conditions, whereas neutral conditions produced accurate velocity reconstructions but resulted in reduced accuracy in resolving temperature fluctuations.” The reason for better performance under convective conditions is that turbulence makes the convective boundary layer more uniform, as evidenced by the distribution of δT in Figure 9. However, this should not be considered a highlight of the paper. The authors should more precisely delineate the key contributions and novel aspects of their study:

 

(1) The logic and written expression of the manuscript

1. For all equations in the manuscript, if they are not original, it is recommended to provide relevant references.
2. In lines 163-167 of the text, it is described that “In the second stage, the computations were restarted from the initial solutions and continued until 20,000 s, incorporating mesh refinement and a reduced time step of 0.1 s to satisfy the Courant–Friedrichs–Lewy (CFL) stability condition, ensuring the CFL number remained below 1.” This requirement of CFL less than 1 is well understood by researchers familiar with LES. However, for those unfamiliar with LES, you should provide relevant references.
3. In atmospheric sciences, distinguishing between neutral and convective conditions primarily relies on stability parameters (e.g., 𝜁 or 𝐿). To facilitate reader understanding, the authors are requested to provide the specific stability parameter values for the neutral and convective conditions mentioned in this study.

 

(2) Figures and tables

1. The color bar in Figure 2 is red and blue, while the indicated line box is also blue. It is recommended to change the line box color to green or purple to avoid confusion. Additionally, the color bar appears to lack units. It is also suggested to improve the image quality, as it is not clearly visible.

 

2. Please provide the unit for the color bar in Figure 4.

 

3. The authors are requested to plot the −5/3 slope in both the neutral (left) and convective (right) cases in Figure 13 to indicate whether the turbulence satisfies the inertial subrange. This will facilitate readers in comparing and verifying the accuracy of the authors' work.

 

 

 

(3) Scientific of the manuscript

1. Lines 284–285 state, “This result means that δu is normally distributed, whereas δv and δT are close to a normal distribution, with ks values of approximately 0.03.” However, the distinction between "normally distributed" and "normal distribution" is unclear. The authors should further analyze these results rather than merely listing them. For instance, the differences in δT between convective and neutral conditions in Figure 9 require further discussion.
2. Lines 485–488 state, “Perhaps most importantly, atmospheric AT research to date has largely relied on the assumption of spatial statistical homogeneity in the mean flow—an assumption that is invalid in wind turbine wakes. NREL’s physical AT array provides an unprecedented opportunity to address some of these outstanding questions.” However, the height of wind turbines is significantly higher than the AT mentioned in this study. Have the authors planned a technical approach to compare AT with Doppler lidar for a more comprehensive analysis?
3. The article notes that "the algorithm exhibited robust performance under convective conditions, whereas under neutral conditions, it achieved accurate velocity reconstructions but showed diminished accuracy in resolving temperature fluctuations." This superior performance in convective conditions can be attributed to the turbulence-induced uniformity of the convective boundary layer, as illustrated by the δT distribution in Figure 9. However, this finding should not be regarded as a central highlight of the paper. The authors should more precisely delineate the key contributions and novel aspects of their study.

 

Reviewer 3 Report

Comments and Suggestions for Authors

Review of “Acoustic Tomography of the Atmosphere: A Large Eddy Simulation Sensitivity Study” by Emina Maric, Bumseok Lee, Regis Thedin, Eliot Quon, and Nicholas Hamilton.

The study tests the performance of the time-dependent stochastic inversion algorithm used in atmospheric tomography for reconstructing the turbulent fields by comparing against the large eddy simulation data.  

Overall, the manuscript is well presented. The theory provides a good introduction to AT for readers. However, the resolution of the figures could be improved. The manuscript started well but became confusing towards the end, especially Section 5. Therefore, I recommend publication after the following comments are addressed. 

 

Comments:

Comment #1: The study only tests the algorithm performance in neutral and convective conditions. I suggest rephrasing the abstract wording “...under varying atmospheric stability conditions” to reflect that.

Comment #2: Figure 1 is good but adds no useful information to the text. Instead, I suggest that the authors add a map of the region showing the terrain variations and overlay the towers. This would provide some context to the upwind terrain complexity as the LES results are used to evaluate the performance of the TDSI algorithm. 

Comment #3: Authors should comment on whether the convective atmosphere with 3 m/s wind speed is driven by buoyant or mechanical turbulence, as this could affect the turbulent length scales and the standard deviations. This can be done by estimating the convective velocity, friction velocity and the stability parameter (for example, see Dosio et al. 2006). Also, the choice of 3 m/s is to be justified. Is it the dominant wind at the NREL site?

Comment #4: To justify Lines 179-180, authors could add a figure showing the average turbulent kinetic energy within the boundary layer reaching a quasi-steady state with time. 

Comment #5: Why 1800 sec choice for analysis? What is the eddy turnover time in the convective atmosphere simulated? For capturing turbulent motions at various scales, it is a common practice to include 2 to 3 eddy turnovers when estimating the turbulent fields. 

Comment #6: Figure 5: Is the distance mentioned on the x-axis along-wind direction? 

Comment #7: How come the integral length scales for stream-wise and cross-stream velocity components are the same? 

Comment #8: Figure 8: There is no red line in the figure. 

Comment #9: Instead of or in addition to Figures 11 and 12, it would be easy for the reader to digest the text in Lines 322-329 if the mean error or box plot for that duration is plotted against N, showing that it doesn’t improve after N=4. As it is now, Figures 11 and 12 are very hard to follow. 

Comment #10: Similar difficulty in visualizing Figure 13. A key portion of the plot is at mid- and lower frequencies, which is very hard to see. PSD trends for frequencies less than 0.1 Hz seem independent of N, and calling N=4 as optimal size is hard to accept.

Comment #11: Line 350: SigmaT for neutral case is different than in Table 4. 

Comment #12: Section 5.2 needs more justification. Why would you compare the LES results with fields that were reconstructed with different standard deviations? Changing standard deviations would change the turbulent intensity, which makes it different from the LES reference state.   

References:

Dosio, A., Vilà-Guerau de Arellano, J., Holtslag, A.A. and Builtjes, P.J., 2003. Dispersion of a passive tracer in buoyancy-and shear-driven boundary layers. Journal of Applied Meteorology, 42(8), pp.1116-1130.

Reviewer 4 Report

Comments and Suggestions for Authors

The study presents a well-structured and relevant investigation into the application of time-dependent stochastic inversion (TDSI) for acoustic tomography (AT) in atmospheric studies. The manuscript is well-written, and the methodology is appropriate for the research objectives. The findings contribute to a deeper understanding of the performance of TDSI under different atmospheric conditions, particularly in neutral and convective boundary layers.

 

Overall, I find this study original and valuable to the field of atmospheric remote sensing and tomography. The results provide useful insights into the limitations and advantages of TDSI for turbulence reconstruction. However, there are a few areas where the authors could clarify their discussion to improve the manuscript.

 

1. The introduction effectively reviews the development of acoustic tomography (AT) and time-dependent stochastic inversion (TDSI). However, it does not explicitly highlight the challenges that currently exist in this research area. The authors discuss the evolution of AT and its advantages but do not clearly state what gaps still need to be addressed in the field of atmospheric turbulence reconstruction. It would strengthen the manuscript if the authors added a paragraph explaining the limitations of existing methods and how this study aims to overcome them. For example, does the primary limitation lie in the resolution of small-scale turbulence, the computational cost of inversion methods, or the sensitivity of AT to environmental noise?

 

2. The advantages of AT are well described in the introduction, but its inherent limitations are not sufficiently discussed. A brief discussion of these limitations would provide a more balanced introduction and further emphasize the necessity of refining TDSI.

 

3. In the results section, the authors report that TDSI reconstructs velocity fields more accurately than temperature fields in neutral conditions, but the reasons for this discrepancy are not fully explored. Is this due to weaker temperature gradients, making it more difficult to detect travel-time variations? Or could it be related to the spatial arrangement of the AT array and its ability to resolve fine-scale fluctuations?

 

4. The references in this paper are somewhat outdated. It is recommended to add more recent references from the past five years.