Distributed acoustic sensing as a tool for subsurface mapping and seismic event monitoring: a proof of concept

Piana Agostinetti, Nicola; Villa, Alberto; Saccorotti, Gilberto

doi:https://doi.org/10.5194/se-13-449-2022

Articles | Volume 13, issue 2

https://doi.org/10.5194/se-13-449-2022

© Author(s) 2022. This work is distributed under
the Creative Commons Attribution 4.0 License.

Special issue:

Fibre-optic sensing in Earth sciences

https://doi.org/10.5194/se-13-449-2022

© Author(s) 2022. This work is distributed under
the Creative Commons Attribution 4.0 License.

Articles | Volume 13, issue 2

Research article

|

03 Mar 2022

Research article |

| 03 Mar 2022

Distributed acoustic sensing as a tool for subsurface mapping and seismic event monitoring: a proof of concept

Nicola Piana Agostinetti, Alberto Villa, and Gilberto Saccorotti

Download

Final revised paper (published on 03 Mar 2022)
Supplement to the final revised paper
Preprint (discussion started on 15 Apr 2021)

Interactive discussion

Status: closed

RC1:
'Comment on se-2021-37', Martijn van den Ende, 01 May 2021

Please find my comments attached.

Citation: https://doi.org/10.5194/se-2021-37-RC1
- AC1: 'Reply on RC1', Nicola Piana Agostinetti, 04 Oct 2021
  
  The comment was uploaded in the form of a supplement: https://se.copernicus.org/preprints/se-2021-37/se-2021-37-AC1-supplement.pdf
  
  Citation: https://doi.org/10.5194/se-2021-37-AC1
CC1:
'Comment on se-2021-37', Herbert Wang, 07 May 2021

This paper is a relatively brief focused discussion on methods for how a DAS array can be used to 1) map subsurface heterogeneity (“exploration”) and 2) detect and locate seismic events (“monitoring”). The methods are validated and compared with geophone (“node”) data co-located with the array in the publicly available PoroTomo data set. The paper achieves its goals and is a useful contribution to understanding how to best to use DAS data. The paper should be particularly valuable to anyone planning a surface DAS array for one or both of the targeted applications.
While I recommend publication, I do have a list of editorial comments, suggestions, and questions. The most significant one is for readability and it is to consider combining the methods and results for the two categories of results (“exploration” and “monitoring”). I do also prefer that “exploration” be called “subsurface site effects” or “subsurface heterogeneity mapping” and the “monitoring” be called “seismic event and location detection”. These titles are clumsy but they don’t overpromise. After all, the basic result in the “exploration” category is relative site effect time delays of +/- 0.12 seconds. Therefore, I might also suggest a more specific title, e.g., Methods for subsurface mapping and event detection using Distributed Acoustic Sensing: Examples from Brady Hot Springs PoroTomo Experiment
Specific Comments, Suggestions, Questions.
l. 1. “PoroTomo” is short for “Poroelastic Tomography” so there is no reason to have the “TOMO” all caps.
l. 36 - Marra et al. is not DAS and is not relevant to this paper.
l. 103-104 - No node instrument response corrections were made. Is that relevant?
l. 115-117 – Given over 100 events in the Li and Zhan catalogue, maybe give a liitle more reason for choosing the 14 March 10:41 UTC event. Is it the closest to the array?
l. 118-150 – These paragraphs from “Methods” would most seamlessly be followed by the “Location” results (see l. 219).
l. 124-125- Is the meaning of “20-channel step” that channels 1-11, 31-41, 61-71, etc. are stacked if Channels 1-100 are a segment?
l. 179-180 Be explicit that “strictly compared” means that nodal horizontal components need to be vector summed in DAS direction.
L.207-209. Can you suggest a reason why the criterion for removing channels needs to be increased to 2 gage lengths from a corner?
l. 212-216 – Can you relate this result to local site effects presented in Parker et al. or Zeng et al.?
l. 219 – Note how out-of-context this statement is because its antecedent goes back to l. 118-135. It is best to discuss this result right after l. 135. Even though the paper is short, I don’t think the reader can be expected to keep this in mind. Consider keeping the “data and methods” and their corresponding results together.
l. 225 – Similarly to above, EDT was referred to back in “methods” section. Also, “Conversely” is a term of logic. The meaning here is “In comparison”.
l. 233-234 - “projection” implies vector, not tensor, projection. cos^2 is stronger than cos.
l. 250 - “PWF” Plane-wave fitting is never defined.
l. 270 – Don’t understand “(2) with a relatively poor areal coverage with respect to a standard seismic network” when DAS arrays up to 50-km in length can be deployed.
l. 271-272 - The statement about "old style" neglects recent deployment of large N node (e.g., Long Beach). So the proper comparison is between large N nodes and DAS.
l. 276 – How does scattering of late arriving surface waves affect P? In general, the wavelengths of P mean that segment-scale heterogeneity or topography should not be cause poor wavefield coherence. Variable coupling of the cable seems more likely.
l. 283-285 – Are “not-aligned segments” the same or different than “three separated segments.” i.e., is the difference 2 vs. 3, or parallel vs. not parallel? What is the point here?
l. 289-290 - Maybe add that teleseismic waves coming into low-velocity, near-surface sediments are nearly vertical and horizontal DAS sensitivity is attenuated by its cos^2 dependence. Also, emphasize that this is a consequence of the PoroTomo geology and is not necessarily a generalization about DAS.
l. 302 - Clarify meaning of “intra-profile channels”. Statement seems contradictory. How can very coherent channels be dominated by locally-scattered waves, unless the profile length is small compared with scatterers?
l. 311-315. Did not try to estimate magnitude of events, but maybe ML can be applied if there are a large number.

Citation: https://doi.org/10.5194/se-2021-37-CC1
- RC2:
  'CC1 again as RC', Herbert Wang, 11 May 2021
  
  This paper is a relatively brief focused discussion on methods for how a DAS array can be used to 1) map subsurface heterogeneity (“exploration”) and 2) detect and locate seismic events (“monitoring”). The methods are validated and compared with geophone (“node”) data co-located with the array in the publicly available PoroTomo data set. The paper achieves its goals and is a useful contribution to understanding how to best to use DAS data. The paper should be particularly valuable to anyone planning a surface DAS array for one or both of the targeted applications.
  
  While I recommend publication, I do have a list of editorial comments, suggestions, and questions. The most significant one is for readability and it is to consider combining the methods and results for the two categories of results (“exploration” and “monitoring”). I do also prefer that “exploration” be called “subsurface site effects” or “subsurface heterogeneity mapping” and the “monitoring” be called “seismic event and location detection”. These titles are clumsy but they don’t overpromise. After all, the basic result in the “exploration” category is relative site effect time delays of +/- 0.12 seconds. Therefore, I might also suggest a more specific title, e.g., Methods for subsurface mapping and event detection using Distributed Acoustic Sensing: Examples from Brady Hot Springs PoroTomo Experiment
  
  Specific Comments, Suggestions, Questions.
  
  l. 1. “PoroTomo” is short for “Poroelastic Tomography” so there is no reason to have the “TOMO” all caps.
  
  l. 36 - Marra et al. is not DAS and is not relevant to this paper.
  
  l. 103-104 - No node instrument response corrections were made. Is that relevant?
  
  l. 115-117 – Given over 100 events in the Li and Zhan catalogue, maybe give a liitle more reason for choosing the 14 March 10:41 UTC event. Is it the closest to the array?
  
  l. 118-150 – These paragraphs from “Methods” would most seamlessly be followed by the “Location” results (see l. 219).
  
  l. 124-125- Is the meaning of “20-channel step” that channels 1-11, 31-41, 61-71, etc. are stacked if Channels 1-100 are a segment?
  
  l. 179-180 Be explicit that “strictly compared” means that nodal horizontal components need to be vector summed in DAS direction.
  
  L.207-209. Can you suggest a reason why the criterion for removing channels needs to be increased to 2 gage lengths from a corner?
  
  l. 212-216 – Can you relate this result to local site effects presented in Parker et al. or Zeng et al.?
  
  l. 219 – Note how out-of-context this statement is because its antecedent goes back to l. 118-135. It is best to discuss this result right after l. 135. Even though the paper is short, I don’t think the reader can be expected to keep this in mind. Consider keeping the “data and methods” and their corresponding results together.
  
  l. 225 – Similarly to above, EDT was referred to back in “methods” section. Also, “Conversely” is a term of logic. The meaning here is “In comparison”.
  
  l. 233-234 - “projection” implies vector, not tensor, projection. cos^2 is stronger than cos.
  
  l. 250 - “PWF” Plane-wave fitting is never defined.
  
  l. 270 – Don’t understand “(2) with a relatively poor areal coverage with respect to a standard seismic network” when DAS arrays up to 50-km in length can be deployed.
  
  l. 271-272 - The statement about "old style" neglects recent deployment of large N node (e.g., Long Beach). So the proper comparison is between large N nodes and DAS.
  
  l. 276 – How does scattering of late arriving surface waves affect P? In general, the wavelengths of P mean that segment-scale heterogeneity or topography should not be cause poor wavefield coherence. Variable coupling of the cable seems more likely.
  
  l. 283-285 – Are “not-aligned segments” the same or different than “three separated segments.” i.e., is the difference 2 vs. 3, or parallel vs. not parallel? What is the point here?
  
  l. 289-290 - Maybe add that teleseismic waves coming into low-velocity, near-surface sediments are nearly vertical and horizontal DAS sensitivity is attenuated by its cos^2 dependence. Also, emphasize that this is a consequence of the PoroTomo geology and is not necessarily a generalization about DAS.
  
  l. 302 - Clarify meaning of “intra-profile channels”. Statement seems contradictory. How can very coherent channels be dominated by locally-scattered waves, unless the profile length is small compared with scatterers?
  
  l. 311-315. Did not try to estimate magnitude of events, but maybe ML can be applied if there are a large number.
  
  Citation: https://doi.org/10.5194/se-2021-37-RC2
  - AC2: 'Reply on RC2', Nicola Piana Agostinetti, 04 Oct 2021
    
    The comment was uploaded in the form of a supplement: https://se.copernicus.org/preprints/se-2021-37/se-2021-37-AC2-supplement.pdf
    
    Citation: https://doi.org/10.5194/se-2021-37-AC2
RC3:
'Comment on se-2021-37', Anonymous Referee #3, 19 Jul 2021

Earthquake seismology is not my discipline. But I see how this paper pushes forward the comparison of DAS and nodal data at Brady's from the recent Wang and Van Eden papers. However, I would like to see if there are more quantitative ways to incorporate the velocity model of the area instead of always making generalities about local velocity changes. This area has great active seismic studies that should be cited (as suggested in the attached) and included in this paper.

Citation: https://doi.org/10.5194/se-2021-37-RC3
- AC3: 'Reply on RC3', Nicola Piana Agostinetti, 04 Oct 2021
  
  The comment was uploaded in the form of a supplement: https://se.copernicus.org/preprints/se-2021-37/se-2021-37-AC3-supplement.pdf
  
  Citation: https://doi.org/10.5194/se-2021-37-AC3

Peer review completion

AR: Author's response | RR: Referee report | ED: Editor decision | EF: Editorial file upload

AR by Nicola Piana Agostinetti on behalf of the Authors (09 Dec 2021) Author's response Author's tracked changes Manuscript

ED: Referee Nomination & Report Request started (03 Jan 2022) by Philippe Jousset

RR by Martijn van den Ende (20 Jan 2022)

Suggestions for revision or reasons for rejection

I would like to thank the authors for taking the time to address the concerns raised during my first assessment of the manuscript. I think that the redefinition of this study’s objectives and clarification of some misunderstandings on my part have improved the quality of the manuscript. I agree with most of the comments made by the authors in response to my assessment, but I still see one major issue that I think should be addressed. (The page/line numbers mentioned below refer to the tracked changes document.)

I still do not agree with the interpretation that DAS amplitudes “robustly correlate with near surface geology” (abstract, L22). In their rebuttal, the authors suggest that their interpretation is based on a somewhat arbitrary amplitude threshold of 0.1. If I apply the same threshold for other sections of the cable, then I would expect to see numerous subsurface features in Fig. 7 and one feature in Fig. 8, which do not seem to be present. The authors also mention that the correlation with the velocity model is contradictive. To explain this contradiction, they argue that the injection wells are concentrated in one particular area with high Vp, “and thus the fracture network is expected to be well developed” (P8 L244-245). I don’t quite follow this reasoning, since in this case I would expect to see a low instead of a high Vp.

Looking at Fig. 21 of Wang et al. (2018), which shows P/S-wave SNR for the entire DAS array, a similar picture appears. Even though there exist array-scale trends in the SNR, these trends do not seem to correlate immediately with mapped geological features (faults, warm grounds, fumaroles). Of course, there could be deeper geological features with no surface expression that could be the cause for the trends in site effects, which should then manifest themselves in the 3D velocity models.

Given that fluctuations in the DAS amplitudes are seen in several places across the array that do not consistently correlate (in my opinion) with mapped near-surface geological features, I suspect that the correlation between the amplitude variation and one feature seen in Fig. 10 is due to chance. If the authors want to maintain their claim that DAS amplitudes robustly correlate with near-surface geology, they will have to quantify how robust this correlation is for the entire cable, and show that the null-hypothesis (no correlation) can be rejected with reasonable confidence. By giving only 1 example that qualitatively correlates well (and 1 example that doesn’t show a direct correlation), the reader cannot assess the significance of the correlation and evaluate what “robust” really means.

Alternatively, if the authors agree with me that the correlation is potentially due to chance, they could modify the abstract/main text to indicate that “the DAS and nodal arrays are in mutual agreement when it comes to site amplifications, but that it is not immediately clear which geological features are responsible for these amplifications. DAS could therefore hold potential for detailed mapping of shallow subsurface heterogeneities, but with the currently available information of the Brady Hot Springs subsurface geology, this potential cannot be quantitatively verified”. In this way, it is clear to the reader that more work remains to be done, instead expecting that DAS is a “robust” tool for subsurface mapping.

Kind regards,

Martijn van den Ende

Minor comments:

P2 L33, “derivative of strain”: maybe a more general way of phrasing would be that DAS measures the spatial derivative of particle motion. The interrogator used in the PoroTomo experiment measured strain rate (the temporal derivative of strain), but other interrogators measure just strain.

P2 L41: as Reviewer #2 already pointed out, Marra et al. did not use DAS

P10 L 315: DAS is not horizontal by definition, but by configuration.

Fig 11: I think the caption is mislabelled. Panel (a) shows the nodal array, panel (b) shows the DAS array.

PWF method: I think we can debate for quite some time about the (dis)similarities between PWF and beamforming, but it would be better to do this over a coffee at a conference than through peer-review reports/rebuttals.

Hide

ED: Reconsider after major revisions (20 Jan 2022) by Philippe Jousset

AR by Nicola Piana Agostinetti on behalf of the Authors (26 Jan 2022) Author's response Author's tracked changes Manuscript

ED: Publish subject to technical corrections (27 Jan 2022) by Philippe Jousset

ED: Publish as is (27 Jan 2022) by CharLotte Krawczyk (Executive editor)

AR by Nicola Piana Agostinetti on behalf of the Authors (31 Jan 2022)

Short summary

Sensing the Earth is a fundamental operation for the future where georesources, like geothermal energy and CO₂ underground storage, will become important tools for addressing societal challenges. The development of networks of optical fibre cables gives the possibility of a sensing grid with an unprecedented spatial coverage. Here, we investigate the potential of using portions of a optical fibre cable as a standard seismometer for exploring the subsurface and monitoring georesources.