the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 18 Sep 2024
Passive seismic imaging of ore deposits using coda wave interferometry: a case study of Akanvaara V-Cr-PGE deposit in Northern Finland
Abstract. In this study, we present an innovative method to image the inner structure of orthomagmatic ore deposits using P-wave coda of regional seismic events. We combine data processing and interpretation schemes from conventional passive seismic interferometry and teleseismic receiver function (RF) method. We hypothesize that correlation of P-wave coda recorded by three-component sensors can be used to evaluate body wave part of empirical Green's tensor, from which arrivals of reflected and converted waves could be extracted. To test our hypothesis, we installed a high-resolution seismic array (profile) with 606 seismic instruments on the Akanvaara V-Cr-PGE deposit in Northern Finland above the inclined zones of V-Cr mineralization, placed inside ultramafic intrusion. From the regional seismic catalogue, provided by the Institute of Seismology, University of Helsinki, we selected the P-wave coda of 363 regional seismic events to evaluate body wave part of empirical Green's tensor by passive seismic interferometry. Further interpretation of the tensor allowed us to identify arrivals of PS and SP waves, converted at Cr and V mineralization zones. We conducted numerical simulation of plane wave interaction with the synthetic Akanvaara deposit model compiled from geological and drilling data and found that Green's tensors evaluated from synthetic seismograms and from seismic data contain similar converted PS and SP arrivals. To calculate depths to the conversion boundaries, we obtained S-wave velocity model using MASW method. According to calculated depths and geological model compiled from drilling data we suggest that the converted arrivals correspond to continuation of the Cr and V mineralized zones. Therefore, using the empirical Green's tensor, evaluated from P-wave coda of regional seismic events can be an effective tool for orthomagmatic ore deposits exploration in both greenfield and brownfield cases. In this paper we are describing details of the passive seismic experiment, numerical simulation, data processing and interpretation.
- Preprint
(14611 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2024-2637', Mariusz Majdanski, 09 Oct 2024
Review of the manuscript Passive seismic imaging of ore deposits using coda wave interferometry: a case study of Akanvaara V-Cr-PGE deposit in Northern Finland by Afonin et al. submitted to Solid Earth journal.
General comment
The manuscript is an extensive and detailed case study of the use of coda waves combined with the MASW technique to recognise mineral deposits in a hard rock environment. The presented technique is innovative and uses high-resolution 3C seismic and correlation of components to image inclined boundaries. The study explains the method using a synthetic example and follows with a field data case. The manuscript is rather long but has a logical structure. It is written in understandable English, and the figure's quality meets the research article criteria.
Main problems to fix
1) the synthetic example is very detailed and tries to mirror the study site geometry. Still, the results, especially in Figure 14, are difficult to follow. If, in such idealised cases, these results are not very convincing, how will this method work with real data?
2) There are some significant differences in processing synthetic and field data. The filtering is very different. For the field data 2-20 Hz, for synthetics 10-15 Hz. I would expect the synthetic case should be as close to real case thus the processing parameters should be the same. Please explain those differences.
3) The final results (Figure 20) are not convincing. How the results look like without interpretation. I cannot see any arrivals as they are marked with interpretation lines.
Small suggestions
Figure 2b to small fonts, it is hard to read any information
Figure 4. caption - add N and E for latitude and longitude for clarity
Figure 4 quality of these plots are horrible, I would like to see some waveforms. Maybe it would be better do decimate the number of traces and show some useful signal. The horizontal scale is not described.
Figure 5 I am guessing those are Z, N and E components. Please mark them, and add event info (date?) too, for clarity
Something is wrong with table 1 – there are P and S rows, but velocity is defined for both. So what for the separate rows with sub-tasks exist? This is explained in the text (page 9) and is not needed in the table.
Page 9, line 200: we select the frequency equal to 10 Hz was used for these waves -> we select the frequency equal to 10 Hz for these waves
Figure 12 these results indeed show some weak signals, but this is a synthetic case when we know exactly what we are looking for. How will it work for field data?
Figure 17 shows application of very narrow filter 17-18 Hz. Is this correct?
My recommendation
I am convinced this manuscript presents an interesting approach to study hard rock environment. The results are not very convincing, but it is still an innovative method. I recommend publishing it after fixing some minor problems and explaining the mentioned issues.
Citation: https://doi.org/10.5194/egusphere-2024-2637-RC1 -
AC1: 'Reply on RC1', Nikita Afonin, 02 Dec 2024
Referee: Review of the manuscript Passive seismic imaging of ore deposits using coda wave interferometry: a case study of Akanvaara V-Cr-PGE deposit in Northern Finland by Afonin et al. submitted to Solid Earth journal.
General comment
The manuscript is an extensive and detailed case study of the use of coda waves combined with the MASW technique to recognise mineral deposits in a hard rock environment. The presented technique is innovative and uses high-resolution 3C seismic and correlation of components to image inclined boundaries. The study explains the method using a synthetic example and follows with a field data case. The manuscript is rather long but has a logical structure. It is written in understandable English, and the figure's quality meets the research article criteria.
Main problems to fix
- the synthetic example is very detailed and tries to mirror the study site geometry. Still, the results, especially in Figure 14, are difficult to follow. If, in such idealised cases, these results are not very convincing, how will this method work with real data?
Authors: The figure 14 represents not synthetic case, but the correlation functions (empirical Green’s tensor, EGT), calculated from the real three-component data, recorded along the profile in our field experiment in Akanvaara deposit area. There are clearly visible PS and SP converted arrivals, originated on the mineralized zones (pointed by yellow arrows in non-diagonal componts of the tensor). Of course these arrivals are not so ideally visible as in synthetic example of EGT (figures 10 and 11). The main reason for this is that syntetic EGT was evaluated from noise-free synthetic data, while the EGT in Fig. 14 was evaluated from real data that is not noise-free. On Figure 15 we compare the synthetic EGT and the EGT calculated from the real data. As seen, they do not contradict each other and are in a good correlation.
Referee: 2) There are some significant differences in processing synthetic and field data. The filtering is very different. For the field data 2-20 Hz, for synthetics 10-15 Hz. I would expect the synthetic case should be as close to real case thus the processing parameters should be the same. Please explain those differences.
Authors: This comment is because of misunderstanding of our processing workflow. We used different frequency bands to process surface waves and P-wave coda
The prefiltering by bandpass filter of 2-20 Hz was used in the surface wave processing:
(Line 299: “We selected surface wave parts of these events and pre-processed them with prefiltering by bandpass filter of 2-20 Hz”)
The prefiltering by bandpass filter of 10-15 Hz was used in processing of P-wave coda
(Line 272: “To obtain empirical Green’s tensor from seismograms we applied prefiltering by bandpass filter of 10-15 Hz and one-bit normalization…”)
The same frequncy band (10-15 Hz) was used in the synthetic case.
Referee:
- The final results (Figure 20) are not convincing. How the results look like without interpretation. I cannot see any arrivals as they are marked with interpretation lines.
Authors: The same components of Empirical Greens tensors are presented also in Figure 14 (withoput interpretation, but with arrows pointing to converted arrivals). Also in Figure 15 we compare the Empirical Green’s tensors caluclated from the real data and those evaluated from synthetic seismograms. Especially in Figure 15 (EGT calculated from real data) there are clearly visible converted arrivals (blue and green arrows in zoomed parts of EGTs ZR components). Of course they are not so clear as in the noise-free synthetic tests (pointed by arrows of the same colors on the right subplots of the figure). But they are visible and interpretable, although not all the traces show the signal of the same quality. But the different quality of traces is quite normal situation even in the controlled-source experiments and also in classical RF studies that use teleseismic events. As the distance between sensors in our experiment was small enough, a certain number of traces with low SNR does not affect general correlation between arrivals of converted waves. We would like to note that correlation and picking of arrivals of converted waves at EGT was not done using paper plots, but using computer software with visualisation of traces, in which it is possible to increase the amplitudes and correlate the nearby traces.
Referee:
Small suggestions
Figure 2b to small fonts, it is hard to read any information
Authors: We improved the figure 2b
Referee: Figure 4. caption - add N and E for latitude and longitude for clarity
Authors: Done
Referee: Figure 4 quality of these plots are horrible, I would like to see some waveforms. Maybe it would be better do decimate the number of traces and show some useful signal. The horizontal scale is not described.
Authors: We improved the figure by adding pointers to Pn, Sn and scattered arrivals in P-wave coda as well as labels on horizontal scale. We added a three component seismograms, recorded by one of the instruments in the profile.
Referee: Figure 5 I am guessing those are Z, N and E components. Please mark them, and add event info (date?) too, for clarity
Authors: We modified the figure by adding notations of components. This figure show not a spectrogram of a single event, but spectra of P-wave coda of all seismic events (separate spectra are denoted by indexes in horizontal scale), which we selected for further processing. Horizontal axis means spectra of P-wave coda of a particular event numbered by some value that we called “Index of seismic event in the sample”.
Referee: Something is wrong with table 1 – there are P and S rows, but velocity is defined for both. So what for the separate rows with sub-tasks exist? This is explained in the text (page 9) and is not needed in the table.
Authors: We agree that these separate rows for P and S waves are not needed. The table 1 has been modified.
Referee: Page 9, line 200: we select the frequency equal to 10 Hz was used for these waves -> we select the frequency equal to 10 Hz for these waves
Authors: Corrected.
Referee: Figure 12 these results indeed show some weak signals, but this is a synthetic case when we know exactly what we are looking for. How will it work for field data?
Authors: This figure shows synthetic seimograms where converted arrivals are weak. But correlation functions of these seimograms contain clear converted arrivals (Figure 13). This example demonstrates the main advantage of interferometric approach and justifies its application to processing of real data.
Referee: Figure 17 shows application of very narrow filter 17-18 Hz. Is this correct?
Authors: Yes. We used the set of such narrow filters to extract dispersion curve of high frequency signal because using MASW method did not allow us to do that in this frequency band.
Referee: My recommendation
I am convinced this manuscript presents an interesting approach to study hard rock environment. The results are not very convincing, but it is still an innovative method. I recommend publishing it after fixing some minor problems and explaining the mentioned issues.
Authors: Thank you very much for your comments, which were helpful to improve the manuscript. We hope that after clarifications in our replies and after making the correspondent modifications in the text, our results became more clear for understanding.
Citation: https://doi.org/10.5194/egusphere-2024-2637-AC1
-
AC1: 'Reply on RC1', Nikita Afonin, 02 Dec 2024
-
RC2: 'Comment on egusphere-2024-2637', Anonymous Referee #2, 24 Oct 2024
General Comments
I enjoyed reading this study. It aims to tackle the important challenge of mineral exploration using passive seismic techniques in a novel and potentially exciting way. The study includes a geological background to the Akanvaara deposit, an overview of the methods, a set of synthetic models, and then processing and interpretation of results from a month-long deployment around the Akanvaara deposit. The introduced method is innovative for mineral exploration, but unfortunately I am not convinced by the results and there are a number of questions I have regarding the validity of the method in this case study. Whilst I would not recommend publishing this exact manuscript, the study could be re-written to clarify/better support the method and improve the quality of the results. Alternatively, perhaps a renewed focus on synthetic case studies from a variety of deposit types with more realistic crustal structures would enable this study to better showcase the suitability of the method for exploration.
Specific Comments
I have given detailed comments below, separated into each section of the manuscript. However, my main questions can be summarised as follows:
P-wave coda
P-wave coda for regional earthquakes are likely dominated by conversions at the Moho, as well as other major crustal reflectors and reverberations in cover lithologies (e.g., Wang et al., 2020). The synthetic models (e.g., Fig. 7) do not contain lower crustal boundary layers and therefore it is difficult to assess the impact these large-scale structures might be having on the results. The geological cross-section only goes down to 100m so it is difficult to assess the impact major upper crustal interfaces, like the basement-cover contact, may have on the case study example. Without addressing these potential issues, I worry whether the method is actually able to reliably resolve shallow structures with thicknesses of only metres to 100s metres.Phases of interest
The study uses SP and SS phase arrivals, among others. I do not understand how it is possible to calculate accurate time delays for SP and SS waves using P-wave coda. The SP and SS waves originate from the initial S-wave, and therefore must be compared relative to the S-wave arrival rather than the P-wave. In the synthetic example, an initial S-wave is used (e.g., Fig. 11), but the S-wave coda is not used in the real case-study data.Depth conversion for Figure 20
The depth conversion of Figure 20 relies on the knowledge of S-wave and P-wave velocities, as mentioned in the study. It is unclear where the required P-wave velocities came from. Additionally, the surface wave analysis of Fig. 19 constrains the S-wave structure to a maximum of 550m, yet Figure 20 contains depths in excess of 5 km. It is unclear where these additional velocity constraints came from, and how the depth conversion is performed.Interpretation of Figure 20
Without referring to the geological cross-section provided in Figure 2, I think it would be difficult to persuasively arrive at the interpretation of the seismic data presented in Figure 20. The geophysics depth conversion, as well as the significant difference in depth of the geophysical cross-section (>5 km in Figure 20) vs. the geological cross-section (~100m in Fig. 2) also raises some questions.Technical Comments
Introduction
- The introduction could do with a few more details on the deposit itself. For example, Lines 66-69 are too sparse in detail (layer thicknesses, rheological contrasts between layers) to allow me to assess whether the method is reasonable to start with.
Geological characteristics etc.
- More references required in the geological background section (e.g., Line 74).
- Figure 1: Can you please show where (b) is on (a), perhaps by highlighting Akanvaara label on (a). Also, does ‘m’ mean metres on (b)? The scale isn’t particularly clear as I am not familiar with the co-ordinate system.
- Perhaps make it clearer that the Akaanvaara intrusion belongs to the 2.44–2.5 Ga suite of layered mafic intrusions.
- Please state the aim more clearly: do you aim to resolve the mineralised layers within the Lower and Middle Zone (thickness of ~cm. to several metres), the magnetite gabbro of the Upper Zone (thickness of 100s metres), or just the general dipping structure of the area? This is important for deciding where the tool would be useful for exploration.
- Fig 2b: is this cross-section in line with the drill-holes or is it a cross-section along the seismic profile onto which you have projected the drill-holes?
- It is unclear in Line 134 as to how the events were selected. Did you use the whole catalogue of 363 events or filter that down to a smaller sample-set?
- How local are the “local” earthquakes and will they still satisfy the vertical incidence assumption (Line 139)?
- Fig 4 and Fig 5: These figures are unclear. Please show some seismograms and perhaps PSD plots through time for individual events.
- Line 160: a few reference examples would be useful here.
Method:
- Line 176: Presumably a significant part of the P-wave coda The Ps will arrive after the P.
- It would be good to link the method description with Figure 6 more clearly.
- Line 189: Surely the SP arrival cannot be compared with the P-wave arrival, as it is a conversion from an original S wave with a different original travel time.
- 7: perhaps make it clearer that Y is equal to depth, and panel 2 represents a bird’s-eye view.
- More detail on the method used by SOF13D would be helpful here.
- Line 217: What is the third dimension?
- Line 224: What does “done with MATLAB scripts” mean?
- 8: A plot of individual seismograms would be useful here, with labelled phases.
- Line 227: Please make it clearer that you subtracted the seismograms from the homogenous model from the seismograms from the heterogenous model.
Data Processing:
- How long was each P-wave coda record? How did you define the start and end?
- I assume this was an autocorrelation, in which each station was correlated with itself. If so, please make this clearer.
- What was the signal-to-noise of the stacked cross-correlations? Were low SNR records removed prior to analysis?
- Surely SP and SS arrivals need to be correlated relative to the original S-wave arrival time. I do not think they can be easily related directly to the original P-wave arrival.
- Line 299: Please give details of these 37 seismic events, and their azimuth.
- Fig 16: (A) I am not sure why the moveout is not reflected about the 0s time. Normally, moveout should be symmetrical about the 0s highlighting anti-causal and causal signals. Please explain why your results are different. (B) The phase velocities seem too high, suggesting either azimuthal biases or perhaps a processing error.
- Fig 17: (A) Please make this figure clearer – the definition is too low. (B) I am not entirely sure what this figure means. I am unsure why there are three 200m distance values. Moveout plots are standardly plotted showing an increasing distance on the y-axis against time on the x-axis, symmetrical around 0s.
- Line 324: please be specific as to the frequency range. Presumably these are surface-wave phase velocities, not S-wave velocities. No phase velocity to S-wave velocity has been performed yet.
- Please plot the 5 dispersion curves, and show where they come from along the profile.
- How does Geopsy calculate the theoretical dispersion curves?
- Did you set the Vp/Vs ratio in the Monte-Carlo models?
- Which misfit function did you use?
- Fig 19: in Line 327, it is mentioned that dispersion curves could not be obtained for two parts of the profile. Where are those parts on Fig. 19? Also, why have the results from 20m–150m been removed if they are included within the 1D inversions?
- Please include some sensitivity kernel plots to demonstrate that the surface waves are indeed sensitive to geological structure at these depths.
Results:
- How did you perform the depth conversion? Did you use the S-wave 2D velocity profile or select a 1D average? What P-wave velocity did you use? The depths go down to 5.4 km, yet the S-wave velocity model only extends to 500m.
- Line 378: Could this be more quantitative. There are numerous boundaries into Fig. 20, 19 and 18. Which boundaries correspond to the 0.15s arrival time?
Citation: https://doi.org/10.5194/egusphere-2024-2637-RC2 -
AC2: 'Reply on RC2', Nikita Afonin, 02 Dec 2024
Referee: General Comments
I enjoyed reading this study. It aims to tackle the important challenge of mineral exploration using passive seismic techniques in a novel and potentially exciting way. The study includes a geological background to the Akanvaara deposit, an overview of the methods, a set of synthetic models, and then processing and interpretation of results from a month-long deployment around the Akanvaara deposit. The introduced method is innovative for mineral exploration, but unfortunately I am not convinced by the results and there are a number of questions I have regarding the validity of the method in this case study. Whilst I would not recommend publishing this exact manuscript, the study could be re-written to clarify/better support the method and improve the quality of the results. Alternatively, perhaps a renewed focus on synthetic case studies from a variety of deposit types with more realistic crustal structures would enable this study to better showcase the suitability of the method for exploration.
Authors: In practice, each type of mineral deposits in hard-rock terrains has own geometry and composition, which can result in different scattered wavefield produced by scattering of the P-wave coda. Our research was made in the framework of SEMACRET project that aims to study orthomagmatic mineral deposits, in which mineralised bodies are often observed as sub-vertical or steeply dipping structures. That is why namely this case we considered in our paper, and we indicated this limitation in our Introduction chapter. It is not possible to consider all the types of ore bodies in hard rock terrains in one single research. We would like to notice that even in the review paper by Bohlen et al. (2003), in which they analysed the synthetic wavefield produced by scattering of signals from controlled sources, they did not consider all the possible geometries and compositions of ore bodies in hard rock terrains. However, the case we show in our study demonstrates that a) P wave coda of regional seismic events could be used as a signal and b) scattered wavefield can be analysed using Empirical Green’s Tensor (EGT) instead of more traditional Empirical Green’s Functions. Despite the data processing scheme for EGT evaluation will be near the same, interpretation of obtained empirical Green’s tensor could be different for each particular case of mineral deposits geometry and composition. For example, massive sulfide ore bodies in Fennoscandia are often observed as sub-horisontal lenses, so the scattered waveield would be different from the case we considered in our study. We hope that researchers from the field of mineral exploration will use the general idea of scattered wavefield analysis in their case studies and this will help them in getting new knowledges about structure of mineral deposits.
Referee: Specific Comments
I have given detailed comments below, separated into each section of the manuscript. However, my main questions can be summarised as follows:
P-wave coda
P-wave coda for regional earthquakes are likely dominated by conversions at the Moho, as well as other major crustal reflectors and reverberations in cover lithologies (e.g., Wang et al., 2020). The synthetic models (e.g., Fig. 7) do not contain lower crustal boundary layers and therefore it is difficult to assess the impact these large-scale structures might be having on the results. The geological cross-section only goes down to 100m so it is difficult to assess the impact major upper crustal interfaces, like the basement-cover contact, may have on the case study example. Without addressing these potential issues, I worry whether the method is actually able to reliably resolve shallow structures with thicknesses of only metres to 100s metres.Authors We agree that the general content of the coda recorded by our high-resolution profile is superposition of arrivals from deep boundaries waves scattered on local heterogeneties. However, in our study we rely on properties of signals from regional and local seismic events analysed in numerous previous studies in the Fennoscandial Shield, from which one important property is the frequency content. For example, in the New Manual of Seismological Observatory Practice, Chapter 9, by Schweitzer et al., they show that the frequencies of the first arrivals and secondary arrivals of P- and S-waves recorded from regional events are in the range of 2-10 Hz. This is also demonstrated by controlled-source wide-angle reflection and refraction studies in our region (see, for example, Janik et al., 2009). They demonstrated that not only first arrivals of P- S- Pn and Sn have frequencies within this range, but also intracrustal secondary arrivals like PcP, ScS have similar frequencies. As we use the prefiltering of the coda by 10-15 Hz bandpass filter, these intracrustal secondary arrivals are eliminated from the signal we use for evaluation of EGT.
Moreover, all these P- and S-arrivals from deep structures can be considered as plane waves due to their frequency content. When they propagate through the near-surface heterogeneilies, they are producung scattered wavefield. This scattered wavefield is superposition of converted and multiple reflected waves and such a case was considered in the studies by Draganov (2004) and Thorbecke and Draganov (2011), in which also results of numerical modelling of scattered wave produced by interaction of the waves from multiple sources with near-surface heterogeneities were demonstrated. In our study we demonstrated that such a scattered wavefield is clearly visible not only in synthetic data, but also in the real seimograms recorded in our high-resolution field experiment (Figure 4.). Using interferometric approach in that case is justified as we can get time lag between different phases of scatterd waves. That is why in our synthetic tests we used P and S plane waves arrived vertically to the ground surface, which represents well the situation with real coda.
Referee: Phases of interest
The study uses SP and SS phase arrivals, among others. I do not understand how it is possible to calculate accurate time delays for SP and SS waves using P-wave coda. The SP and SS waves originate from the initial S-wave, and therefore must be compared relative to the S-wave arrival rather than the P-wave. In the synthetic example, an initial S-wave is used (e.g., Fig. 11), but the S-wave coda is not used in the real case-study data.Authors: As we tried to explain in reply to comment above, we do not use P-wave as is but scattered wavefiled, produced by interaction of this wave with local heterogeneities and present in the P-wave coda. This wavfield contains both scattered P and S waves among which are converted and multiple reflected waves. They are coherent to each other as they are secondary waves of the same primary wavefield. Therefore we can use interferometric principle to evaluate time lag between them. We did not use S-wave coda because this is too short for calculation of correlation finctions.
Referee: Depth conversion for Figure 20
The depth conversion of Figure 20 relies on the knowledge of S-wave and P-wave velocities, as mentioned in the study. It is unclear where the required P-wave velocities came from. Additionally, the surface wave analysis of Fig. 19 constrains the S-wave structure to a maximum of 550m, yet Figure 20 contains depths in excess of 5 km. It is unclear where these additional velocity constraints came from, and how the depth conversion is performed.Authors: We took P wave velocites values from petrophysical information about typical rocks of the same type in Fennoscandia (Dortman, 1992; Schön, 2015). Concerning S-wave velocities, we assumed that at depths of 5 km they are of the same values as in the depth of 500 m. This assumption is based on geological information about main types of rocks in Akanvaara intrusion, namely mafic and ultramafic rocks.
Referee: Interpretation of Figure 20
Without referring to the geological cross-section provided in Figure 2, I think it would be difficult to persuasively arrive at the interpretation of the seismic data presented in Figure 20. The geophysics depth conversion, as well as the significant difference in depth of the geophysical cross-section (>5 km in Figure 20) vs. the geological cross-section (~100m in Fig. 2) also raises some questions.Authors: Of course is difficult to interpret the results without referring to geological information as in any geophysical method. Despite we do not know exactly properties of the medium deeper than several hundreed meters, we know that the medium in depth of 500 m is presented by ultramafic voulcanic rocks and we can reasonably assume that seismic velocities there should be near the same as at depths of 5 km, because we are dealing with orthomagnmatic deposit. Geophysical methods have always been used to extend knowledge about geology, although with some restrictions. In our study we did the same: we extended existing knowledge about geology using results of new seismic experiment method and previous geological and petrophysical information about that area.
Referee: Technical Comments
Introduction
- The introduction could do with a few more details on the deposit itself. For example, Lines 66-69 are too sparse in detail (layer thicknesses, rheological contrasts between layers) to allow me to assess whether the method is reasonable to start with.
Authors: The text about the lithology and thickness of layers has been added to the manuscript. ‘In the case study of the Akvanvaara intrusion, the different layered rocks below the experiment area include chromitite (1-2 meters), anorthosite (10s of meters), and gabbro (10s of meters) and magnetite gabbro (>100 meters)’.
Referee: Geological characteristics etc.
- More references required in the geological background section (e.g., Line 74)
Authors: References have been added to this paragraph (line 74-85)
Referee:
- Figure 1: Can you please show where (b) is on (a), perhaps by highlighting Akanvaara label on (a). Also, does ‘m’ mean metres on (b)? The scale isn’t particularly clear as I am not familiar with the co-ordinate system.
Authors: The figure has been improved (Akanvaara highlighted in the subplot (a)). Coordinate system, used in subplot (b) is ETRS 1989 TM35FIN NE. X axis means East direction, Y axis means North direction (mentioned in the figure caption)
Referee:
- Perhaps make it clearer that the Akaanvaara intrusion belongs to the 2.44–2.5 Ga suite of layered mafic intrusions.
Authors: Revised in geological section to make it clear that Akanvaara belongs to the 2.44-25 Ga layered intrusions suite. line 82-85.
Referee:
- Please state the aim more clearly: do you aim to resolve the mineralised layers within the Lower and Middle Zone (thickness of ~cm. to several metres), the magnetite gabbro of the Upper Zone (thickness of 100s metres), or just the general dipping structure of the area? This is important for deciding where the tool would be useful for exploration.
Authors: We were trying to resolve the mineralised layers within the Lower and Middle Zone among the general dipping structure as well as to get general information about the medium.
Referee:
- Fig 2b: is this cross-section in line with the drill-holes or is it a cross-section along the seismic profile onto which you have projected the drill-holes?
Authors: This cross-section was obtained from drill-holes (302,303,307,308 in subplot (a)). This line is parallel to our seismic profile (black curve in subplot (a)).
Referee:
- It is unclear in Line 134 as to how the events were selected. Did you use the whole catalogue of 363 events or filter that down to a smaller sample-set?
Authors: We selected 363 seismic events from all seismic events preented in the bulletin for considered period.
Referee:
- How local are the “local” earthquakes and will they still satisfy the vertical incidence assumption (Line 139)?
Authors: We mentioned in the test that we used seismic events, originated at epicentral distances of 250-800 km.
Referee:
- Fig 4 and Fig 5: These figures are unclear. Please show some seismograms and perhaps PSD plots through time for individual events.
Authors: The main reason why we include the Figure 4 is to show duration of P-wave coda of typical seismic event we used in our study, as well as scattered arrivals in this coda. This is not possible to show in the figure of lower resolution or in seismogram recorded by single seismic station. We think that the seismic record by single station as well as its spectrogram is not so informative in this particular case. We improved the Figure 4 by adding pointers to Pn, Sn and scattered arrivals in P-wave coda.
The Figure 5 show frequency content of P-wave coda of all seismic events, which we selected for further processing. Horizontal axis means spectra of P-wave coda of a particular event that we indicated by a number that we called “Index of seismic event in the sample”.
Referee:
- Line 160: a few reference examples would be useful here.
Authors: We added some clarification there
Referee:
Method:
- Line 176: Presumably a significant part of the P-wave coda The Ps will arrive after the P.
Authors: We mean there not coda but single wave. According to scattering matrix presented in Aki and Richards, 2002, the amplitude of converted PS wave will be of near the same value as amplitude of incidence P wave in our case.
Referee:
- It would be good to link the method description with Figure 6 more clearly.
Authors: Figure 6 is just an illustration of interferometric redatuming principle used in our proposed method, but not illustration of the method itself.
Referee:
- Line 189: Surely the SP arrival cannot be compared with the P-wave arrival, as it is a conversion from an original S wave with a different original travel time.
Authors: We rephrased the sentence: “In that case t_ps is a time lag on correlation function that corresponds to the time lag between direct P wave originated on the boundary and PS wave converted on the same boundary”
Referee:
- 7: perhaps make it clearer that Y is equal to depth, and panel 2 represents a bird’s-eye view.
Authors: Done
Referee:
- More detail on the method used by SOF13D would be helpful here.
Authors: In our manuscript we provided reference to this sofrtware, where the method is described in details. The sentence was rephrased as:
“To calculate synthetic seismograms, we used finite-difference approach described by Virieux, 1986 and Levander, 1988 and realized in SOFI3D software”
Referee:
- Line 217: What is the third dimension?
Authors: The third dimention is depth, in addition to two horizontal directions.
“The simulation was run for a 400x400x400 grid corresponding to 4000 m in horizontal directions and 4100 m in vertical direction”
Referee:
- Line 224: What does “done with MATLAB scripts” mean?
Authors: This means that for SOFI3D models of the medium (Vp,Vs and Rho distributions) defined on a grid are needed. We used MATLAB to make these grids and define parameters in the nodes of the grid.
Referee:
- 8: A plot of individual seismograms would be useful here, with labelled phases.
Authors: We added pointers to phases we want to demonstrate on this figure.
Referee:
- Line 227: Please make it clearer that you subtracted the seismograms from the homogenous model from the seismograms from the heterogenous model.
Authors: Done.
Referee:
Data Processing:
- How long was each P-wave coda record? How did you define the start and end?
Authors: We defined start and end of the coda using regional travel-time curves of Pn and Sn waves. In most cases, lengths of coda are of 30-60 s
Referee:
- I assume this was an autocorrelation, in which each station was correlated with itself. If so, please make this clearer.
Authors: We can call autocorrelation functions only in case of correlating the same components of the same station (to evaluate diagonal EGT components). In the case of diferent components they can not be called autocorrelation functions. That is why in our work we use the term “correlation functions”.
Referee:
- What was the signal-to-noise of the stacked cross-correlations? Were low SNR records removed prior to analysis?
Authors: This ratio was good enough to evaluate dispersion curves because we used only records of seismic events with the large magnitudes. An example of seismic record by single station is presented in Figure 4, where high SNR is obvious. The Figure 16 also demonstrates frequency-velocity spectra of high quality. Therefore, we did not estimate SNR in data processing.
Referee:
- Surely SP and SS arrivals need to be correlated relative to the original S-wave arrival time. I do not think they can be easily related directly to the original P-wave arrival.
Authors: Non diagonal components of EGT contain delays between secondary P and S waves in coda, which originated when scattering original P-wave on heterogeneities and they do not relate directly to original P-wave arrival. Please see replies to the previous comments, related to the content of P-wave coda.
Referee:
- Line 299: Please give details of these 37 seismic events, and their azimuth.
Authors: Additional information has been included to the text
Referee:
- Fig 16: (A) I am not sure why the moveout is not reflected about the 0s time. Normally, moveout should be symmetrical about the 0s highlighting anti-causal and causal signals. Please explain why your results are different. (B) The phase velocities seem too high, suggesting either azimuthal biases or perhaps a processing error.
Authors: These functions are asymmetric because we analysed waves, coming from narrow range of azimuths. These velocities cannot be high because of azimuths, as we used seismic waves arrived from direction aligned with our profile (+/-10 deg.). They are high because they propagate through ultramafic rocks, where Vs can exceed even 5000 m/s (Shön, 2015)
Referee:
- Fig 17: (A) Please make this figure clearer – the definition is too low. (B) I am not entirely sure what this figure means. I am unsure why there are three 200m distance values. Moveout plots are standardly plotted showing an increasing distance on the y-axis against time on the x-axis, symmetrical around 0s.
Authors: The figure 17 (a) have been improved; (b) Y axes labels have been corrected
Referee:
- Line 324: please be specific as to the frequency range. Presumably these are surface-wave phase velocities, not S-wave velocities. No phase velocity to S-wave velocity has been performed yet.
Authors: This was typo. Of course we mean phase velocities of surface waves there. The sentence has been corrected.
Referee:
- Please plot the 5 dispersion curves, and show where they come from along the profile.
Authors: We show an example of dispersion curve in Figure 18 a. Four other dispersion curves are near the same and their presentation in the manuscript is not so important for the main topic we discussing there.
Referee:
- How does Geopsy calculate the theoretical dispersion curves?
Authors: Geopsy is an open-access software that is widely used. It was not developed by ourself for solving the inversion problem for surface waves. In this case inclusion of the details of the software into our manuscript is not ethical, because a detailrd user guide exists for this software, and it is open access. So it is more correct to provide a reference recommended by the authors of this product. We provided a reference to this software in our manuscript.
Referee:
- Did you set the Vp/Vs ratio in the Monte-Carlo models?
Authors: Yes.
Referee:
- Which misfit function did you use?
Authors: The root mean square deviation of experimental and theoretical phase velocities is divided by the squared experimental phase velocities. (see Geopsy user guide)
Referee:
- Fig 19: in Line 327, it is mentioned that dispersion curves could not be obtained for two parts of the profile. Where are those parts on Fig. 19? Also, why have the results from 20m–150m been removed if they are included within the 1D inversions?
Authors: We corrected the sencence in the text, related to Fig.19. We corrected the Fig.18.
Referee:
- Please include some sensitivity kernel plots to demonstrate that the surface waves are indeed sensitive to geological structure at these depths.
Authors: In our case such plots are not useful and will overload the manuscript. We planned our experiment taking into account target resolution and depths.
Referee:
Results:
- How did you perform the depth conversion? Did you use the S-wave 2D velocity profile or select a 1D average? What P-wave velocity did you use? The depths go down to 5.4 km, yet the S-wave velocity model only extends to 500m.
Authors: For this conversion, we used averaged velocities, taken from our obtained S-wave velocity models as well as petrophysical and geological information about this deposit. Taking into account that we study orthomagmatic deposit, in which mafic and ultramafic volcanic rocks prevail, we assumed that velocities at depths of 5000 m should be equal to those at depths of 500 m. Of course, the depths we obtained in this case are dependent on these assumptions, but they allow us to get estimations about extensions of mineralized zones in considered deposit, as these zones are reaching the surface.
Referee:
- Line 378: Could this be more quantitative. There are numerous boundaries into Fig. 20, 19 and 18. Which boundaries correspond to the 0.15s arrival time?
Authors: We corrected this text.
Editor comment 1
Editor:
Dear Authors,
please proceed with your responses to the reviewer's comments.
I would also like to add to please extend your discussions and conclusions section. At the moment the latter is unbalanced with respect to the other sections of the manuscript, and a comprehensive discussion on the implication of your work is lacking.
Thank you very much, and looking forward to read the revised version of the manuscript.
Kind regards,
Irene Bianchi
Authors: We extended discussion and conclusion part
Refernces
Aki, K., and Richards, P.G.: Quantitative Seismology. 2nd Edition, CA: Univ. Sci. Books, Sausalito, 700, 2002.
Bohlen,T., Miller, C. and Milkereit, B.: Elastic Seismic Wave Scattering from Massive Sulfde Orebodies: On the Role of Composition and Shape. In: Eaton, D. W., Milkereit, B. and Salisbury, M H. (eds) Hardrock Seismic Exploration. Geophysicsl Development Series, SEG, https://doi.org/10.1190/1.9781560802396, 2003.
Dortman, N. B.: Handbook Petrophysics, Nedra, Moscow, 390 pp, 1992.
Janik, T., Kozlovskaya, E., Heikkinen, P., Yliniemi, J., and Silvennoinen, H.: Evidence for preservation of crustal root beneath the Proterozoic Lapland‐Kola orogen (northern Fennoscandian shield) derived from P and S wave velocity models of POLAR and HUKKA wide‐angle reflection and refraction profiles and FIRE4 reflection transect. Journal of Geophysical Research: Solid Earth, 114(B6), 2009.
Schweitzer, J., Fyen, J., Mykkeltveit, S., & Kværna, T. (2002). Manual of seismological observatory practice.
Schön, J. H.: Physical properties of rocks: Fundamentals and principles of petrophysics. Elsevier, 2015.
Citation: https://doi.org/10.5194/egusphere-2024-2637-AC2
-
EC1: 'Comment on egusphere-2024-2637', Irene Bianchi, 05 Nov 2024
Dear Authors,
please proceed with your responses to the reviewer's comments.
I would also like to add to please extend your discussions and conclusions section. At the moment the latter is unbalanced with respect to the other sections of the manuscript, and a comprehensive discussion on the implication of your work is lacking.
Thank you very much, and looking forward to read the revised version of the manuscript.
Kind regards,
Irene Bianchi
Citation: https://doi.org/10.5194/egusphere-2024-2637-EC1 -
AC3: 'Reply on EC1', Nikita Afonin, 02 Dec 2024
Editor:
Dear Authors,
please proceed with your responses to the reviewer's comments.
I would also like to add to please extend your discussions and conclusions section. At the moment the latter is unbalanced with respect to the other sections of the manuscript, and a comprehensive discussion on the implication of your work is lacking.
Thank you very much, and looking forward to read the revised version of the manuscript.
Kind regards,
Irene Bianchi
Authors: We extended discussion and conclusion part
Citation: https://doi.org/10.5194/egusphere-2024-2637-AC3
-
AC3: 'Reply on EC1', Nikita Afonin, 02 Dec 2024
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 166 | 63 | 38 | 267 | 5 | 4 |
- HTML: 166
- PDF: 63
- XML: 38
- Total: 267
- BibTeX: 5
- EndNote: 4
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1