Three-Dimensional Modeling of the Xichang Crust in Sichuan, China by Machine Learning
Round 1
Reviewer 1 Report
Dear authors,
I read your manuscript. Your approach is very interesting and scientifically sounding; even if the method has been applied on an area not of global interest, the study is intriguing as the methodology could be adopted in other areas of the world, were plenty of geological and geophysical data to be merged and integrated are available.
The paper is well written, even if some typos are present and the clarity of some sentences could be improved. The main comment I address to you is to enrich the interpretation of the final 3D geological model, moving it to the discussion paragraph. The quality of figures is sufficient: more information should be provided in order to allow the reader a better comprehension of the images.
I here provide a line-by-line revision of the manuscript, with all the minor comments requested.
Best regards,
Title: What do you mean with community?
Abstract, line 2: “of” instead of “in”
Paragraph2: In my opinion, another sub-section titled “Seismotectonic setting” could be added. The first part of the paragraph 2.1, indeed, contains descriptive information on the tectonics and seismic activity of the area.
Figure 1. Please, substitute "Deep" with "Depth" in the respective axis.
Figure 1. I do not see on the map the JHF (Jinghe Fault) label. Based on what written in the paragraph 4.1, may I suppose that the Jinghe Fault coincides with Qinghe fault, and that the respective label is QHF?
Pag. 4: Line 78: Information on the magnitude range of the adopted earthquakes should be provided.
Line 82: “except the granite body in Shimian area is cut by the Daliangshan fault”. Remove “is”
Line 101: What do you mean with “fault occurrence data”?
Page 5, Line 136 – 145: Too long sentence. Please, fix it.
Table 1: Substitute “tomographic” with “tomography”.
Page 6, Line 149 – 175: This description of the S-wave velocity model is very difficult to me to be understood, may be due to the particular choice of color palette of figure 3a. In my opinion, the reader should be better guided in finding the highlighted features of the velocity model, and it is all but trivial by using a “recursive” color palette. You should find a way to better highlight in the image the features you are describing.
Page 6, Line 152-153: “The velocity of 1.28-2.8km/s with large velocity variation and shallow depth”. Please, reformulate
Page 6, Line 173. Substitute “discontinuous” with “discontinuity”.
Page 6, Line 180: remove “corresponding”.
Page 7, Figure 3: The color palette are incomplete, as the numbers are adimensional and without a description of what they represent. The caption of the figure should be enriched as well. Please consider this last comment for many of the figures of the manuscript.
Page 8, Line 214: As I said before, is “Qinghe” fault the “Jinghe” fault named in Figure 1?
Page 8, Line 236: Referring to the “principal compressive stress”, is this information coming from focal mechanisms or from regional tectonic studies? Please, specify.
Page 8, Line 239 – 241 “Since one of ….” Please, reformulate.
Line 243: What do you mean with "occurrence information of fault plane"?
Page 9, Line 251: “Double difference method”. The citation of the paper introducing the method misses.
Line 259: Why do you speak about “regional” earthquakes?
Line 267-269: Reformulate the sentence, please.
Page 10, Line 305: Why did you choose this combination of parameters? What did exactly the “repeated” experiment suggest you and which kind of information provided?
Page 10, Line 325: Why did you choose a third-order polynomial?
Figure 4: Please substitute, in the “South-East-Deep” reference system, “deep” with “Depth”. The caption should be enriched as well. Furthermore, I would write the values along the three axis in terms of km instead of m.
Page 11, Line 349: Please, fix “seismc”.
Page 11, Line 346: “distance range of samples as 0.1 – 1”. Is there a unit of measure, or the distance is adimensional?
Page 12, Line 383-384: “The release energy and influence range ...” I do not understand this sentence.
Page 13, Line 407: What do you mean here with “lower”?
Figure 5: The color palette is incomplete, as its title is missing. Please, enrich the caption as well.
Page 14, Line 449-450: “The three restrict and depend on each other”. I do not understand this sentence.
Page 15, Figure 6: What do you mean with “external source structure” and “internal source structure”? The meaning of the arrows should be specified as well.
Page 16-17: In my opinion, the contents of the “Discussion” and “conclusion” paragraphs should be inverted. Furthermore, the description and interpretation of the geological model should be enriched. It seems too simplified. In addition, the main element of the geological model (i.e. the high speed body; the low velocity body etc.) should be better highlighted in this very complex 3D comprehensive geological model.
Page 16, Line 505: What do you mean with “misfractured”?
Author Response
Response to Reviewer 1 Comments
Dear reviewer,
Thank you very much for your support and detailed comments on the manuscript. Your valuable suggestions have played a very important role in the revision of the paper. I read all the comments carefully and corrected all the inadequacies point by point. Finally, the paper has been improved in the revision.
Although the purpose of this paper is to introduce a method to develop a unified crustal medium structure model using multi-source results, the interpretation and reliability of the model results are also very important. Therefore, I added the relevant content in the 'Discussion' section as requested.
Other response to reviewer 1 comments are as follows:
Point 1: Title: What do you mean with community?
Response 1: It means a regional model, and the title has been changed according to Reviewer 2.
Point 2: Abstract, line 2: “of” instead of “in”
Response 2: It has been modified as required.
Point 3: Paragraph2: In my opinion, another sub-section titled “Seismotectonic setting” could be added. The first part of the paragraph 2.1, indeed, contains descriptive information on the tectonics and seismic activity of the area.
Response 3: Accpted. This part mainly describes the reasons for selecting data and data sources. The former mainly discusses the importance of this region based on the structural background and research degree, while the latter mainly discusses the scope and reliability of dataset.
Point 4: Figure 1. Please, substitute "Deep" with "Depth" in the respective axis.
Response 4: It has been modified as required.
Point 5: Figure 1. I do not see on the map the JHF (Jinghe Fault) label. Based on what written in the paragraph 4.1, may I suppose that the Jinghe Fault coincides with Qinghe fault, and that the respective label is QHF?
Response 5: It is Qinghe fault, It has been modified as required.
Point 6: Pag. 4: Line 78: Information on the magnitude range of the adopted earthquakes should be provided.
Response 6: It has been modified as required.
Point 7: Line 82: “except the granite body in Shimian area is cut by the Daliangshan fault”. Remove “is”
Response 7: It has been modified as required.
Point 8: Line 101: What do you mean with “fault occurrence data”?
Response 8: It refers to fault orientation. It's made up of dip, dip angle and strike.
Point 9: Page 5, Line 136 – 145: Too long sentence. Please, fix it.
Response 9: It has been fixed as required.
Point 10: Page 6, Line 149 – 175: This description of the S-wave velocity model is very difficult to me to be understood, may be due to the particular choice of color palette of figure 3a. In my opinion, the reader should be better guided in finding the highlighted features of the velocity model, and it is all but trivial by using a “recursive” color palette. You should find a way to better highlight in the image the features you are describing.
Response 10: Partly agreement We have compared the viualization with the different color setting. If using contiuous color palette as below, it is difficult to highlight the characteristics of velocity structure with uniform color scale because the velocity in shallow layer varies greatly while the velocity in deep layer varies gently. The author tries several colormaps, and the display effect is not as obvious as that in this manuscript.
Point 11: Page 6, Line 152-153: “The velocity of 1.28-2.8km/s with large velocity variation and shallow depth”. Please, reformulate
Response 11: It has been modified as required.
Point 12: Page 6, Line 173. Substitute “discontinuous” with “discontinuity”.
Response 12: It has been modified as required.
Point 13: Page 6, Line 180: remove “corresponding”.
Response 13: It has been modified as required.
Point 14: Page 7, Figure 3: The color palette are incomplete, as the numbers are adimensional and without a description of what they represent. The caption of the figure should be enriched as well. Please consider this last comment for many of the figures of the manuscript.
Response 14: It has been modified as required.
Point 15: Page 8, Line 214: As I said before, is “Qinghe” fault the “Jinghe” fault named in Figure 1?
Response 15: It has been modified as required.
Point 16: Page 8, Line 236: Referring to the “principal compressive stress”, is this information coming from focal mechanisms or from regional tectonic studies? Please, specify.
Response 16: This information is coming from focal mechanisms, and references are indicated.
Point 17: Page 8, Line 239 – 241 “Since one of ….” Please, reformulate.
Response 17: It has been changed to ‘Due to one of the interfaces in the focal mechanism solution may be approximately parallel to the fault plane’.
Point 18: Line 243: What do you mean with "occurrence information of fault plane"?
Response 18: It mainly refers to the extended azimuth of fault plane in space. It consists of three elements: strike, dip and dip angle.
Point 19: Page 9, Line 251: “Double difference method”. The citation of the paper introducing the method misses.
Response 19: It has been modified as required.
Point 20: Line 259: Why do you speak about “regional” earthquakes?
Response 20: Earthquakes in the study area, and delete "regional" to eliminate misunderstanding
Point 21: Line 267-269: Reformulate the sentence, please.
Response 21: It has been modified as required.
Point 22: Page 10, Line 305: Why did you choose this combination of parameters? What did exactly the “repeated” experiment suggest you and which kind of information provided?
Response 22: The farther the earthquake is from the initial fault plane, the less likely it is to belong to the fault. By repeated experiments, more irrelevant earthquakes are deleted.
Point 23: Page 10, Line 325: Why did you choose a third-order polynomial?
Response 23: This is based on the latest research results of Brunsvik et al. (2021), who believe that the fault plane of interpolation in this way is more reasonable.
Point 24: Figure 4: Please substitute, in the “South-East-Deep” reference system, “deep” with “Depth”. The caption should be enriched as well. Furthermore, I would write the values along the three axis in terms of km instead of m.
Response 24: It has been modified as required.
Point 25: Page 11, Line 349: Please, fix “seismc”.
Response 25: It has been modified as required.
Point 26: Page 11, Line 346: “distance range of samples as 0.1 – 1”. Is there a unit of measure, or the distance is adimensional?
Response 26: This is the neighborhood radius after the feature transformation, no units. Distance Defaults to Euclidean distance.
Point 27: Page 12, Line 383-384: “The release energy and influence range ...” I do not understand this sentence.
Response 27: For the sake of understanding, the original sentence is changed to ‘As the magnitude of an earthquake increases, the energy it releases and the range of its effects increase exponentially.’
Point 28: Page 13, Line 407: What do you mean here with “lower”?
Response 28: It means that underneath a fault or other geologic body,
Point 29: Figure 5: The color palette is incomplete, as its title is missing. Please, enrich the caption as well.
Response 29: It has been modified as required.
Point 30: Page 14, Line 449-450: “The three restrict and depend on each other”. I do not understand this sentence.
Response 30: The velocity structure model, fault model and seismic structure model verify, depend on and constrain each other, and form the crustal medium structure model of Xichang region together.
Point 31: Page 15, Figure 6: What do you mean with “external source structure” and “internal source structure”? The meaning of the arrows should be specified as well.
Response 31: Both of them are two kinds of seismic source structures, "External source structure" refers to the area where an earthquake has occurred, which is the outer part of the source structure. "Internal source structure" refers to the region with frequent occurrence and high seismic density, which is inside the source structure. The arrows mean the normal direction of the surface.
Point 32: Page 16-17: In my opinion, the contents of the “Discussion” and “conclusion” paragraphs should be inverted. Furthermore, the description and interpretation of the geological model should be enriched. It seems too simplified. In addition, the main element of the geological model (i.e. the high speed body; the low velocity body etc.) should be better highlighted in this very complex 3D comprehensive geological model.
Response 32: The "Discussion" and "Conclusion" sections have been changed as requested, and the understanding and explanation of the model has been added to the discussion section. In addition, In addition, it is difficult to highlight the morphological characteristics of high-speed and low-speed bodies at the same time because they are inside this very complex 3D comprehensive geological model.
Point 33: Page 16, Line 505: What do you mean with “misfractured”?
Response 33: it is mean that once the crystalline substrate is broken or fractured, the earthquake will focus on the fault section.
Thank you again for your valuable advice. We would be very grateful if the modification can meet your requirements. Your kind considerations will be greatly appreciated. If you have any queries, please don’t hesitate to contact me at the address below.
With best regards,
Your sincerely,
Professor Shi Chen
Institute of Geophysics, China Earthquake Administration
Address: No.5 Minzu University South Road, Haidian District, Beijing
Tel: +86-010-68729401; Fax: +86-010-68729401
E-mail: chenshi@cea-igp.ac.cn
27 February 2022
Author Response File: 
Author Response.docx
Reviewer 2 Report
see attached detailed comments
Review of the paper entitled
The community 3D crustal model using machine learning method: A case study in Xichang area
By Li-Wen Gong, Huai Zhang, Shi Chen and Li-Juan Chen
This paper addresses important issues concerning the geometric 3D modeling of the deep crust in the high seismic risk Xichang area, China. In particular the authors extract a 3D model of the seismic (Vp) velocities from the processing of the seismicity of the area to be used to constrain the construction of a 3D model, including the geological formations, and the faulted structures at depth. The final aim is to use the resulting 3D model to better predict potential seismic risk deep zones, to try to understand the mechanisms that originate earthquakes, and to use it in rock mechanic simulations. This type of studies has been addressed by the Stanford group in the 90’s (not referenced) to better predict earthquakes in California, and then, in many other parts of the world. Several improvements have been done since these early works in particular in the acquisition, processing, inversion and visualization techniques of the seismic signal, increasing precision and filtering blurring noises. Similar improvements have been done in the seismic exploration techniques for the oil and gas industry (not referred in the paper).
The originality of the paper is a rigorous approach for extracting, processing, inverting the huge amount of seismic data collected on the areas since more than 10 years, and in the neural network (NN) and machine learning (ML) techniques used for clustering the relevant seismic events by fault zones. Little details are given on the implementation of the used processing methods (how many layers for the NN, what are the performances, uncertainty …). Faults are then extended at depth from their surface traces using these seismic events clusters; no uncertainties are given on the fault positions / locations, nor new unrecognized faults are identified at depth from the seismic cloud. No uncertainty analysis (for instance in the clustering procedure, depth location, fault position…) is given on the results presented in the case study. How much sensitive to errors is the final model? These points need discussion.
In addition to the spatial location, magnitude, mode, moment, seismic information gives also a dynamical view of the occurrence of the seismic events; the time variable is not exploited in this work; it can be used to decipher the earthquake mechanisms and the role of faults (i.e. post-event micro-seismicity, temporal sequence of earthquakes along the fault structures …). The derived model is only static and no dynamic. The early works by Nur and Cline (2000) [5] have demonstrated that seismic events / earthquakes have occurred moving along the Anatolian fault following a temporal sequence due to the transfer of the strains to adjacent fault segments. Can this phenomenon be observed in the Xichang area?
The methodology used is original, correctly implemented, and no errors have been noticed in the derived conclusions. However, some comments / improvements should be made on some detailed aspects in the implementation issues / typing before publication:
- It is suggested to shorten the title into “3D modeling of the Xichang crust (China) using machine learning (ML)” (see detailed comments bellow)
- Number of misprints wording should be fixed together with the English writing (see detailed comments bellow)
- Some references have been suggested to be added in the list (see detailed comments bellow)
- English writing needs improvement before publication
- The fonts of some legends should be enlarged for better readability (see detailed comments bellow);
- Some references are incomplete and should be completed (see list in detailed comments)
The authors identify seismic risk near fault junctions “with high terrestrial heat flow value, partial melting of granite may exist.” Values of such high surface heat flow should be given (how much 150mW/m2? higher?). The arguments for evoking partial melting are not clearly enough explained. On what arguments the authors say that? Are they some evidence of volcanic events? If granite / material partial melting occurs, the Vp velocity should decrease, the Vs should be near 0 (in liquid), and a thermal modeling of the deep crust should be in agreement with external conditions for beginning fusion (650-750°C, 2kbar depending on H2O content).
Another explanation would be an accumulation of compressive stresses in the surrounding rocks (should increase the Vp or decrease if it reaches the limit of rupture) adjacent to the fault, or along the faults, and at their intersections. When sufficient energy is accumulated to move the fault / break the rock (with expel of fluids -> seismic valves mechanism), the earthquakes occur. Nur et al. [5] suggested such an unzipped mechanism to explain the sequence occurrence in time of earthquakes along the North Anatolian fault. Could such observations be made along the faults of the Xichang area?
A number of misprints / English typing errors need revision; it is suggest checking the English writing before the final publication.
Similitudes with other works have not been noticed and no plagiarism with other papers has been identified.
This interesting paper needs revisions before publication.
Detailed review
See the amended enclosed pdf.
Suggestion for additional references:
- Grada, M.; Polkowskia, M.; Ostaficzukb S.R. (2016) High-resolution 3D seismic model of the crustal and uppermost mantle structure in Poland. Tectonophysics, 666, 188-210.
- Moschetti, M.P., M.H. Ritzwoller, and F.C. Lin, (2010) Seismic evidence for widespread crustal deformation caused by extension in the western USA, Nature, 464(7290), 885-889
- Moschetti, M.P., M.H. Ritzwoller, F.C. Lin, and Y. Yang, (2010) Crustal shear velocity structure of the western US inferred from amient noise and earthquake data, Geophys. Res., 115, B10306, doi:10.1029/2010JB007448, 2010
- Manu-Marfo, D., Aoudia, A., Pachhai, S. et al. (2019) 3D shear wave velocity model of the crust and uppermost mantle beneath the Tyrrhenian basin and margins. Sci Rep 9, 3609. https://doi.org/10.1038
- Nur, A.; Cline, E. (2000). Poseidon's Horses: Plate Tectonics and Earthquake Storms in the Late Bronze Age Aegean and Eastern Mediterranean. Journal of Archaeological Science. 27. 43-63. 10.1006/jasc.1999.0431.
When citing references please put a blank to separate the text: i.e. improved [1] and (not improved[1]) make changes throughout the paper.
- 1 Suggested title -> The community 3D crustal model using machine learning method: A case study in Xichang area-> 3D modeling of the Xichang crust (China) using machine learning (ML).
p.1 abstract line 1: The seismicity and earthquakes relocation can show the regional structure in-formation in crust -> Seismicity and earthquakes relocation can provide structural information on the crust at a regional scale.
p.1 abstract line 2: The fault morphology can be traced from epicentral distribution of small shocks -> The morphology of faults can be deciphered from the epicentral distribution of small shocks /events.
p.1 abstract line 4: information to modelling the 3D crustal model -> information for modelling in 3D the crust in
p.1 abstract line 7:This new community 3D crustal model included the -> This new 3D crustal model includes the
p.1 abstract last line: This community crustal model can also be used for the geodynamic analysis and strong motion simulation -> This crustal model can also be used for geodynamic analysis and strong motion simulations
p.2 abstract line 3: arrays have been deployed in Sichuan and Yunnan province-> arrays have been deployed in the Sichuan and Yunnan provinces
p.2 line 14-16: However, in the numerical simulation of large-scale regional deformation, the crustal medium structure model used is relatively simple -> However, the crustal medium structure model used for simulating numerically large-scale regional deformations is relatively simple
p.2 line 18: structure, which affects -> structure, and affects
p.2 line 46: The Xichang area
p.2 line 49: between Xianshuihe fault system and Xiaojiang fault system[32] -> between the Xianshuihe fault system and Xiaojiang fault systems
p.3 line 53: the Anninghe fault
p.3 line 68: in Xichang area. -> in the Xichang area.
p.3 Figure 1: change label Deep/km by -> Depth/km, add a 3D view of the seismic cloud
p.4 line 79: monitored in study area -> monitored in the study area
p.4 line 82-85: except the granite body in Shimian area is cut by the Daliangshan fault, the other faults are distributed along the boundary of magmatic body or Xichang basin, obviously constrained by the geometric framework of basement ->
except for the granite body in the Shimian area that is cut by the Daliangshan fault, the other faults are distributed along the boundary of the magmatic body or the Xichang basin, obviously constrained by the geometric framework of the basement.
p.4 line 88: such as Shimian area. -> such as the Shimian area.
p.4 line 114: please give the reference for the Gempy library.
p.4 line 116: The modeling flow chart is shown -> The workflow is shown
p.5 line 117 figure 2: Workflow
p.5 line 122: velocity structure of different scales - velocity structure at different scales
p.5 line 136: by Shen et al.(2016) is - > by Shen et al. (2016) [12] is
p.8 line 225 Table 2: give unit for length, give also the refrences by number i.e. Li et al. [7]
p.9 Table 2: Deep -> Depth (km) or Z altitude
p.9 Table 2: Column names should be aligned with figures, Table 2 must be reedited.
p.11 figure 4: Enlarge legends for redability. The three legends seem to be identical, so you can put only one (with more readability) for the three pictures. Unit along axes should be in km X(km), Y(km) Z(km)
p.12 line 384: What’s more, in order -> In order
p.14 line 439: in Xichang area (repetition) line 441 in the Xichang area
p.14 line 446: the authors say “with high terrestrial heat flow value, partial melting of granite may exist.” Values of such high surface heat flow should be given (150mW/m2 ? higher? ). On what arguments the authors say that? Are they some evidence of volcanic events? If partial melting of granite occurs, the Vp velocity should decrease and the Vs should be near 0 (in liquid).
Another explanation would be an accumulation of compressive stresses in the rocks (should increase the Vp) adjacent to the fault, or along the faults and at their intersections. When sufficient energy is accumulated to move the fault / break the rock the earthquakes occur. Nur et al. [5] suggested such an unzipped mechanism to explain the sequence occurrence in time of earthquakes along the North Anatolian fault. Could such observations be made along the faults of the Xichang area ?
p.15 figure 6: enlarge fonts in legend.
p.16 line 521: add page number to the following references [1], [4], [8], [11], [17], [20], [22], [24], [25], [28],
Comments for author File: Comments.zip
Author Response
Response to Reviewer 2 Comments
Dear reviewer,
Thank you very much for your support and detailed comments on the manuscript. Your valuable suggestions have played a very important role in the revision of the paper. I read all the comments carefully and corrected all the inadequacies point by point. Finally, the paper has been improved in the revision.
Since the 90 s, as you know, the Stanford team began to study velocity structure model, velocity inversion method is becoming more and more mature in recent years, there are many different kinds of inversion method, and a lot of achievements in research and velocity structure in the area, but how to make use of existing velocity model, combined with other recent research results to establish the reasonable structure of the crust medium model, is a scientific problem solved in this paper, So it can provide geometric model for seismic physical prediction and geodynamic analysis.
The machine learning methods adopted in this paper are mainly cluster analysis and supervised classification, which are used to extract seismic clusters. Then the fault plane structure model is fitted by the morphological characteristics of seismic clusters. In addition, the source region can be well delineated by the density isosurface of equivalent seismic particles, that is, the source structure. In summary, this paper uses velocity structure to build a frame model, and combines fault structure model and focal structure model to obtain a more precise crustal medium structure model.
The main purpose of this paper is to establish a crustal medium structure model, so more attention is paid to the three-dimensional spatial characteristics of seismic clusters. However, through the analysis of the earthquake occurrence time, it is found that there is no obvious migration law of earthquakes in this region, and only in 2013-2014 and 2018-2019, the frequency of seismic activity within the range of each earthquake cluster is high , which is related to the stress background of the whole region. In addition, there is a large earthquake cluster beneath Mount Gongga, which has no faults exposed on the surface, so this does not mean that there is a fault in this area, because the fault would need other evidence to support it.
Other response to reviewer 2 comments are as follows:
Point 1: Suggested title -> The community 3D crustal model using machine learning method: A case study in Xichang area-> 3D modeling of the Xichang crust (China) using machine learning (ML).
Response 1: It has been modified as required.
Point 2: p.1 abstract line 1: The seismicity and earthquakes relocation can show the regional structure in-formation in crust -> Seismicity and earthquakes relocation can provide structural information on the crust at a regional scale
Response 2: It has been modified as required
Point 3: p.1 abstract line 2: The fault morphology can be traced from epicentral distribution of small shocks -> The morphology of faults can be deciphered from the epicentral distribution of small shocks /events.
Response 3: It has been modified as required.
Point 4: p.1 abstract line 4: information to modelling the 3D crustal model -> information for modelling in 3D the crust in
Response 4: It has been modified as required.
Point 5: p.1 abstract line 7:This new community 3D crustal model included the -> This new 3D crustal model includes the
Response 5: It has been modified as required.
Point 6: p.1 abstract last line: This community crustal model can also be used for the geodynamic analysis and strong motion simulation -> This crustal model can also be used for geodynamic analysis and strong motion simulations
Response 6: It has been modified as required.
Point 7: p.2 abstract line 3: arrays have been deployed in Sichuan and Yunnan province-> arrays have been deployed in the Sichuan and Yunnan provinces
Response 7: It has been modified as required.
Point 8: p.2 line 14-16: However, in the numerical simulation of large-scale regional deformation, the crustal medium structure model used is relatively simple -> However, the crustal medium structure model used for simulating numerically large-scale regional deformations is relatively simple
Response 8: It has been modified as required.
Point 9: p.2 line 18: structure, which affects -> structure, and affects
Response 9: It has been modified as required.
Point 10: p.2 line 46: The Xichang area
Response 10: It has been modified as required.
Point 11: p.2 line 49: between Xianshuihe fault system and Xiaojiang fault system[32] -> between the Xianshuihe fault system and Xiaojiang fault systems
Response 11: It has been modified as required.
Point 12: p.3 line 53: the Anninghe fault.
Response 12: It has been modified as required.
Point 13: p.3 line 68: in Xichang area. -> in the Xichang area..
Response 13: It has been modified as required.
Point 14: p.3 Figure 1: change label Deep/km by -> Depth/km, add a 3D view of the seismic cloud
Response 14: It has been modified as required.
Point 15: p.4 line 79: monitored in study area -> monitored in the study area
Response 15: It has been modified as required.
Point 16: p.4 line 82-85: except the granite body in Shimian area is cut by the Daliangshan fault, the other faults are distributed along the boundary of magmatic body or Xichang basin, obviously constrained by the geometric framework of basement -> except for the granite body in the Shimian area that is cut by the Daliangshan fault, the other faults are distributed along the boundary of the magmatic body or the Xichang basin, obviously constrained by the geometric framework of the basement.
Response 16: It has been modified as required.
Point 17: p.4 line 88: such as Shimian area. -> such as the Shimian area
Response 17: It has been modified as required.
Point 18: p.4 line 114: please give the reference for the Gempy library.
Response 18: References to the Gempy Library have been added as required
Point 19: p.4 line 116: The modeling flow chart is shown -> The workflow is shown
Response 19: It has been modified as required.
Point 20: p.5 line 117 figure 2: Workflow
Response 20: It has been modified as required.
Point 21: p.5 line 122: velocity structure of different scales - velocity structure at different scales
Response 21: It has been modified as required.
Point 22: p.5 line 136: by Shen et al.(2016) is - > by Shen et al. (2016) [12] is
Response 22: It has been modified as required.
Point 23: p.8 line 225 Table 2: give unit for length, give also the refrences by number i.e. Li et al. [7]
Response 23: It has been modified as required.
Point 24: p.9 Table 2: Deep -> Depth (km) or Z altitude
Response 24: It has been modified as required.
Point 25: p.9 Table 2: Column names should be aligned with figures, Table 2 must be reedited.
Response 25: Table 2 has been reedited
Point 26: p.11 figure 4: Enlarge legends for redability. The three legends seem to be identical, so you can put only one (with more readability) for the three pictures. Unit along axes should be in km X(km), Y(km) Z(km)
Response 26: Figure 4 has been reworked as required
Point 27: p.12 line 384: What’s more, in order -> In order
Response 27: It has been modified as required.
Point 28: p.14 line 439: in Xichang area (repetition) line 441 in the Xichang area
Response 28: The repetition have been deleted as required
Point 29: p.14 line 446: the authors say “with high terrestrial heat flow value, partial melting of granite may exist.” Values of such high surface heat flow should be given (150mW/m2 ? higher? ). On what arguments the authors say that? Are they some evidence of volcanic events? If partial melting of granite occurs, the Vp velocity should decrease and the Vs should be near 0 (in liquid).
Response 29: Combined with the results of previous studies, it is believed that the granite (S-type granite)beneath Mt. Gongga was formed by partial melting of the crust. Combined with the distribution of hot springs around mt. Gongga and the anomaly of terrestrial heat flow, it is believed that the granite is not completely cooled, resulting in the low-velocity anomaly.
Point 30: Another explanation would be an accumulation of compressive stresses in the rocks (should increase the Vp) adjacent to the fault, or along the faults and at their intersections. When sufficient energy is accumulated to move the fault / break the rock the earthquakes occur. Nur et al. [5] suggested such an unzipped mechanism to explain the sequence occurrence in time of earthquakes along the North Anatolian fault. Could such observations be made along the faults of the Xichang area ?
Response 30: This unzipped mechanism may explain some faults in Xichang area, such as the Daliangshan fault, because the fault is arcuate and misbreaks the north-south crystalline basement, which may form lock-in and cause moderate and strong earthquakes. However, for small earthquakes in other regions, as long as the stress increases, small earthquakes will occur in the weak parts, and no obvious migration law is found.
Point 31: p.15 figure 6: enlarge fonts in legend.
Response 31: It has been modified as required.
Point 32: p.16 line 521: add page number to the following references [1], [4], [8], [11], [17], [20], [22], [24], [25], [28],
Response 32: The reference pages have been supplemented as required
Please see the attachment for the revised manuscript
Thank you again for your valuable advice. We would be very grateful if the modification can meet your requirements. Your kind considerations will be greatly appreciated. If you have any queries, please don’t hesitate to contact me at the address below.
With best regards,
Your sincerely,
Professor Shi Chen
Institute of Geophysics, China Earthquake Administration
Address: No.5 Minzu University South Road, Haidian District, Beijing
Tel: +86-010-68729401; Fax: +86-010-68729401
E-mail: chenshi@cea-igp.ac.cn
27 February 2022
Author Response File: Author Response.docx
Round 2
Reviewer 2 Report
Review of the paper entitled
3D modeling of the Xichang crust (China) using machine learning
By Li-Wen Gong, Huai Zhang, Shi Chen and Li-Juan Chen
This second version is considerably improved and is quite ready for publication as most of previous misprint corrections / comments / recommendations have been made / included / accounted for.
However, some further little improvements should be considered before publication:
- References given in Table 1 are not referenced by number i.e. Han et al. (2021) should be Han et al. (2021) [16]. This should be fixed for all references.
The authors identify low velocity anomalies as partial melting zones (line 486 p. 16):
“Moreover, there are low velocity anomalies under the Mt. Gongga with high terrestrial heat flow value, possibly due to incomplete cooling of the granite formed by partial melting [13].”
The age of the Gongga granite is not given. However as far as I know, formations in the Xichang are dated from the Permian or Triassic, or even older (ophiolite). Any granite of that age should be cold now (time for granite cooling is within the range 2 to 10 Ma (max)). High surface heat flow could be due to a thermal conductivity contrast between the basement and sedimentary basin, fluid up-coming along faults, or crust thinning. I am not convinced by the incomplete cooling hypothesis which seems for me unrealistic. Decrease in seismic velocity can be due to more hydrate material located beneath the Mt. Gongga.
This interesting paper can be published after minor revisions.
Comments for author File: 
Comments.pdf
Author Response
Dear reviewer,
I am very glad to see your revision opinions again. Your affirmation of my first modified article has given me great encouragement. I read the comments carefully, and consulted the relevant references, then revised them point by point. The response to comments are as follows:
Point 1: References given in Table 1 are not referenced by number i.e. Han et al. (2021) should be Han et al. (2021) [16]. This should be fixed for all references.
Response 1: It has been modified as required.
Point 2: “Moreover, there are low velocity anomalies under the Mt. Gongga with high terrestrial heat flow value, possibly due to incomplete cooling of the granite formed by partial melting [13].”
The age of the Gongga granite is not given. However as far as I know, formations in the Xichang are dated from the Permian or Triassic, or even older (ophiolite). Any granite of that age should be cold now (time for granite cooling is within the range 2 to 10 Ma (max)). High surface heat flow could be due to a thermal conductivity contrast between the basement and sedimentary basin, fluid up-coming along faults, or crust thinning. I am not convinced by the incomplete cooling hypothesis which seems for me unrealistic. Decrease in seismic velocity can be due to more hydrate material located beneath the Mt. Gongga.
Response 2: After reading your explanation of low velocity body and thermal anomaly under Mount Gongga, I have a new understanding of its formation. I also checked some other references which had different explanations for its cause, such as the root effect of high mountains; The crustal flow is blocked by the eastern high-speed body and converges here; Incomplete cooling of the granite formed by partial melting, etc. This is a complex scientific question , and more evidence is needed to explain the specific causes. In this paper, more attention is paid to its morphological characteristics and rock mechanical properties for later numerical simulation. Therefore, in order to avoid some misunderstanding, the corresponding description has been modified
Please see the attachment for the revised manuscript
Thank you again for your valuable advice. We would be very grateful if the modification can meet your requirements. Your kind considerations will be greatly appreciated. If you have any queries, please don’t hesitate to contact me at the address below.
With best regards,
Your sincerely,
Professor Shi Chen
Institute of Geophysics, China Earthquake Administration
Address: No.5 Minzu University South Road, Haidian District, Beijing
Tel: +86-010-68729401; Fax: +86-010-68729401
E-mail: chenshi@cea-igp.ac.cn
3 March 2022
Author Response File: Author Response.docx
