Estimation of Avalanche Development and Frontal Velocities Based on the Spectrogram of the Seismic Signals Generated at the Vallée de la Sionne Test Site

Round 1

Reviewer 1 Report

In this paper, Surinach and colleagues analyze seismic measurements performed at the Vallée de la Sionne test site in order to extract information on the avalanche approaching velocity at different positions along the avalanche path. For this aim they use the part of the seismic signal anticipating the arrival of the avalanche over the sensors and propose a theoretical approach to extract the avalanche front velocity from the measurements. They separate avalanches into three avalanche flow regime populations, namely: wet, powder and transitional. To validate the velocity values from the seismic analysis they compare the results with GEODAR measurements.

In my opinion, the idea to estimate of the avalanche “approaching“ velocity from a single seismic signal is original and scientifically sound. Nevertheless, the paper is plagued by many problems and should not be published in the present form.

I will not provide detailed suggestions to improve the paper because the work quantity is too large for a reviewer. Few major suggestions to improve the paper are listed below.  

English: The English spelling and form need to be substantially improved in all paper. In general, the quality of the English is not sufficient for publication.

Paper title: it is very generic and does not describe the content of the paper. Something like “ Estimation of avalanche front velocities from seismic measurements at the Vallée de la Sionne test site” would be more precise.

Abstract: also here there are very generic sentences that need to be improved: For example sentences like “ Different characteristics related to the type an devolution of the flow behavior were obtained” are typical sentences with few meaning, which need to be avoided in a scientific paper. Concerning the English: passive forms are also to be avoided. Further, the results highlighted in the abstract are not consistent with the result of the paper. For example: “Our results show the feasibility of classifying the type of avalanche using the characteristics of the time series and the frequency-in- time evolution of seismic signals of avalanches.” This was already shown in previous publications, and thus cannot be listed between the result of this paper. Further, it appears that the proposed method to extract the approaching velocity can maybe work for wet avalanches but not for powder ones, etc….

Introduction: The reading is difficult. Sentences are often clumsy. There are numerous, irrelevant (for the paper aims) details. A general structure is missing. For example, the lines 79-92 are more suited for a method section than an introduction. The sentences are often too long and not spitted into paragraphs. Please, stick to the information which is relevant for the paper.

Section 3: methodology. This should be one of the most important part of the paper, since you claim that the method to extract the approaching velocity from the seismic is new. The method should be described in detail here and not in the appendix, and clear references should distinguish between what has already been done and what is new. Not being a seismic specialist, this part needs to be checked again by a researcher from the field to verify that the procedure is scientifically sound.

Section 4: Avalanche evolution. The paragraphs 4.1, 4.2, 4.3 and 4.4 are too much detailed for the aim of the paper. As already stated, it is already known from previous publications that the seismic sensors are comparable to other measurements in term of timing and length of signals or flow regimes. I suggest to move large part of this info into an appendix, and just summarize the essential detail in the main text.

I have a major concern is relation to the flow regime separation. Transitional avalanches have not a specific rheology which can be detected by the seismic sensors as univocal. A transitional avalanche is just an avalanche which has or the characteristics of a powder avalanche or that of a warm/dense flow depending on the position along the path. This means that, for your specific analysis, the transitional avalanches have to be ranked in one of the two categories (wet/dense or powder), depending on the position of the sensor and corresponding flow regime!  

Discussion: One of the main concerns I have about the discussion is that it appears that avalanches which do not fit into the expected population are simply left out of the analysis, or moved in another avalanche population. I do not understand really the criteria for these choices. The elimination of the category transitional flows may help in this sense.

Further, has been the GEODAR velocity data projected along the avalanche path to be really comparable with the seismic measurements?

The discussion paragraph is long. Try to split it into subsection.

Conclusions: Do not reflect the real results of the paper.  Please, reflect again on what is really new in this paper.

Author Response

Thank you very much for your comments.

 In our opinion, you seem to have another idea of our paper, which differs from our aim. We have attempted to adapt the content of our manuscript to your idea considering also the constructive comments of reviewer 2. Given that the estimation of the velocities is only one part of our contribution we have maintained the parts that we considered essential in this manuscript.

We have kept the description of the evolution of the avalanches comparing the spectrograms and time series. As suggested by reviewer 2, we have split and enlarged Figure 3.  However, following your suggestion part of the section 4 has been moved to Appendix B. We maintain this information because, despite your suggestion, the evolution of the seismic signals for the three different types of avalanche using jointly time series and spectrograms has not been fully described in the literature to date. Our specific and unique distribution of the seismic sensors along a path allows us to do this.    

The paper has undergone a large reorganization because of your requirements.

As regards the types of avalanche mentioned in the manuscript, we have used the classification in Pérez-Guillén et al, (2016). We maintain this classification. This is only a manner to identify them. We observed that the seismic signals (time series and spectrograms) present differences in the different sections according to the avalanche type. You are right that a Transitional avalanche does not have a specific rheology. A Transitional avalanche is just an avalanche which has either the characteristics of a Powder avalanche or that of a warm/dense flow depending on the position along the path. However, we observe a different behavior in the spectrogram sections of the seismic signals along the path according to the type of avalanche. This is precisely what we show in our study. The three avalanche types maintain their own identities along the path. You will find a detailed explanation below in response to your comments.

In this paper, Surinach and colleagues analyze seismic measurements performed at the Vallée de la Sionne test site in order to extract information on the avalanche approaching velocity at different positions along the avalanche path. For this aim they use the part of the seismic signal anticipating the arrival of the avalanche over the sensors and propose a theoretical approach to extract the avalanche front velocity from the measurements. They separate avalanches into three avalanche flow regime populations, namely: wet, powder and transitional. To validate the velocity values from the seismic analysis they compare the results with GEODAR measurements.

In my opinion, the idea to estimate of the avalanche “approaching“ velocity from a single seismic signal is original and scientifically sound. Nevertheless, the paper is plagued by many problems and should not be published in the present form.

I will not provide detailed suggestions to improve the paper because the work quantity is too large for a reviewer. Few major suggestions to improve the paper are listed below. 

English: The English spelling and form need to be substantially improved in all paper. In general, the quality of the English is not sufficient for publication.

There is an inconsistency between the comments of the two reviewers. Whereas for review 1 the English is acceptable, reviewer 2 indicates that the English is not of sufficient quality for publication.

 In this new version, the manuscript has undergone an exhaustive revision of English spelling and style. 

Paper title: it is very generic and does not describe the content of the paper. Something like “ Estimation of avalanche front velocities from seismic measurements at the Vallée de la Sionne test site” would be more precise.”

The estimation of the velocities is one of the outputs of the paper. For us, the avalanche evolution observed through the generated seismic signals is significant. The estimation of the velocities is an output of this evolution. We mentioned the changes earlier in this part in accordance with the suggestions of the two reviewers.

We have changed the title including Vallée de la Sionne test site as demanded by the reviewer. The new title is the following:

Estimation of avalanche evolution and frontal velocities based on the spectrograms of the seismic signals generated at the Vallée de la Sionne test site.

Abstract: also here there are very generic sentences that need to be improved: For example sentences like “ Different characteristics related to the type an devolution of the flow behavior were obtained” are typical sentences with few meaning, which need to be avoided in a scientific paper.

Phrase changed

Concerning the English: passive forms are also to be avoided.

We intend to avoid passive forms in this new version, despite the fact that passive is extensively used in scientific publications.  Moreover, the rules of the journal concerning the references do not make it easy to use the active voice.

Further, the results highlighted in the abstract are not consistent with the result of the paper. For example: “Our results show the feasibility of classifying the type of avalanche using the characteristics of the time series and the frequency-in- time evolution of seismic signals of avalanches.” This was already shown in previous publications, and thus cannot be listed between the result of this paper. Further, it appears that the proposed method to extract the approaching velocity can maybe work for wet avalanches but not for powder ones, etc….

The sentence was changed incorporating the role of the spectrograms.  

As mentioned earlier, an exhaustive study of spectrograms and time series of the different avalanche types has not been published to date given that we have a specific and unique distribution of seismic sensors along a path. We maintain this information in the paper and we emphasize that the different sections (SON, SOB and TAIL (STA, SEN)) of the spectrograms can be used to study the avalanche evolution for different types.  

The aim of the paper is twofold: a) to highlight the role of the different sections of the spectrograms in order to observe the evolution of the avalanche and b) to estimate the speed of the front of the avalanche. For this  purpose, the analysis of the curves is necessary.

Additionally, as mentioned in the paper, GEODAR images do not reflect the same phenomena as the seismic signals do, and in some cases the start of the avalanche was not detected by GEODAR. In this situation, the comparison of the results is meaningless. Therefore, we have not compared the measurements. This does not imply that our method does not work for powder avalanches.

In our opinion, seismic signals can furnish some additional information to the other measurements.

We have incorporated changes into the abstract

Introduction: The reading is difficult. Sentences are often clumsy. There are numerous, irrelevant (for the paper aims) details. A general structure is missing. For example, the lines 79-92 are more suited for a method section than an introduction. The sentences are often too long and not spitted into paragraphs. Please, stick to the information which is relevant for the paper.

We have eliminated the first paragraphs of the introduction devoted to the history of our studies of seismic signals. However, we have maintained the information related to the different attempts carried out to estimate the avalanche velocity because we believe it is a synthesis that could be useful in the special issue in which our contribution will be included.  In our opinion, it is important to highlight that the velocity can be obtained from our method with only one seismic signal. This constitutes a significant difference from other published works.

Additionally, we have incorporated into the introduction a short description of the HT method that we used in the quantification of the spectrograms.

Section 3: methodology. This should be one of the most important part of the paper, since you claim that the method to extract the approaching velocity from the seismic is new. The method should be described in detail here and not in the appendix, and clear references should distinguish between what has already been done and what is new. Not being a seismic specialist, this part needs to be checked again by a researcher from the field to verify that the procedure is scientifically sound.

We did not include the method in the main text so as not to interfere in the presentation of the analysis of the avalanches. This new method is generic, and can be applied to other seismic signals of mass movements. However, as demanded by you we have moved the content of Appendix B to the main text as Section 3. This also meets the requirement of reviewer 2, who asked for an explanation of how the κ and β parameters were obtained. As a result of this, a reorganization of the paper has been necessary.  We hope this new version will be to your satisfaction.

Section 4: Avalanche evolution. The paragraphs 4.1, 4.2, 4.3 and 4.4 are too much detailed for the aim of the paper. As already stated, it is already known from previous publications that the seismic sensors are comparable to other measurements in term of timing and length of signals or flow regimes. I suggest moving large part of this info into an appendix, and just summarize the essential detail in the main text.

We are not aware of the publications that explain in detail the different roles of the sections of the spectrograms for the different types of avalanche at three different locations along a path.  

According to your suggestion, but also in agreement with reviewer 2, we have moved part of the information to the appendix, whilst maintaining the description in the main text.

I have a major concern is relation to the flow regime separation. Transitional avalanches have not a specific rheology which can be detected by the seismic sensors as univocal. A transitional avalanche is just an avalanche which has or the characteristics of a powder avalanche or that of a warm/dense flow depending on the position along the path. This means that, for your specific analysis, the transitional avalanches have to be ranked in one of the two categories (wet/dense or powder), depending on the position of the sensor and corresponding flow regime! 

We used the description and classification employed in previous SLF related papers. You are right: Transitional avalanches have not a specific rheology. A transitional avalanche is just an avalanche which has either the characteristics of a powder avalanche or that of a warm/dense flow depending on the position along the path and the quality of the entrained snow. This is precisely what we sought to show.

In our spectrograms, we observe that (you can find the information in Table A1). Note that the different sections of the spectrograms are different depending on the avalanche type and site. Powder snow (POW) (Figures 3 and A3e) and transitional (TRANS )(Figures 4 and A3d ) avalanches show a clear SON section at B and C locations in contrast  to the WET avalanche  (Figures 5 and A3f) in which the SON section is not clear as it shows peaks of energy associated with impacts. No specific SON section is observed at any type of avalanches at D.

The SOB section in the POW avalanche is short (22s) at B and C (10s) and no section is observed at D.  The length of this section in the TRANS avalanche at B is longer (35s) than the section in the POW and the WET (28s) avalanches. At C, in contrast to the POW avalanche, the SOB section in the TRANS avalanche is very long (173s) and energetic. The SOB section in the WET avalanche is not very well observed at C, and it is similar to the SOB section in the TRANS avalanche at D. No SOB section is observed at any of the three avalanches at D.

The STA-SEN section is longer in the WET (121s) and TRANS avalanches (123s) than in the POW avalanche (68s) at B.  This section is almost absent in the POW and the WET avalanches at C. In the TRANS avalanche, it disappears at the expense of the SOB section at C. The spectrogram in the TRANS avalanche at D is similar to that of the WET avalanche at C, showing mainly energy of low frequency content. Note, also, the energetic phase at the end of the signals observed in the three avalanches that corresponds to the stopping phase [ 22].

Although as mentioned, a TRANS avalanche is a POW avalanche that evolves into a WET avalanche along the path. The specific characteristics of the spectrograms enable us to distinguish between the different avalanche types given that the different sections appear at different locations (and times) along the path (Note the discrepancy in the SOB section at C in both avalanches (TRANS and WET) and the similarity of the spectrograms at D (for TRANS) and C (for WET)).

 What we seek to show with the spectrograms is that it is possible to discriminate between the avalanche types exclusively from the evolution of the spectrograms at different altitudes/locations of the path.

As regards the curves, we observe the avalanche evolution at different states at the different locations. The TRANS avalanche has characteristics of WET at D, whereas at C still it does not follow this regime. In the case of the WET avalanche, according to the spectrograms, they maintain their regime along the path. We do not calculate the curves at D because of the lack of data.

You mentioned that transitional avalanches do not have a specific rheology which can be detected by the seismic sensors as univocal. Since we obtained a different behavior in the seismic signals for the different avalanche types, we should perhaps consider that it is not only rheology that seismic signals detect. Note that GEODAR and seismic signals reflect a different phenomenon.

Discussion: One of the main concerns I have about the discussion is that it appears that avalanches which do not fit into the expected population are simply left out of the analysis, or moved in another avalanche population. I do not understand really the criteria for these choices. The elimination of the category transitional flows may help in this sense.

The three avalanches that we eliminated in the calculations have a release location different from that of the others as indicated in Table 1 and they descend not only Gully 1. We have clarified this in the text.  To avoid misunderstanding we have eliminated the curves in the plots.

Further, has been the GEODAR velocity data projected along the avalanche path to be really comparable with the seismic measurements?

YES, the values of both measurements are comparable. GEODAR presents problems when the avalanche descends Gully 2 (A. Khöler personal comm.), but this is not our case since we only selected avalanches descending Gully 1.

The discussion paragraph is long. Try to split it into subsection.

Thank you.   Done.

Conclusions: Do not reflect the real results of the paper.  Please, reflect again on what is really new in this paper.

Thank you. Done.

Author Response File: Author Response.pdf

Reviewer 2 Report

Interesting paper that requires some work to be more clear and readable.

Comments:

1) All the figures in the paper must be improved. Labels for axis (in particular the vertical one is always  missed so that is hard to understand the meaning (check, among other, Fig. 3)

2) Fig. 3 should be split in two and enlarged, otherwise is very hard to understand. Again, what is the meaning of the axis?

3) Fig. 10 and 11 (among others) should be enlarged and improved. The same for the pictures in the Appendix A and B.

4) Text refers to Appendix A1 or A2  but the Appendix are label A and B. Please check your manucsript.

5) Eqs. 1 and 2 are presented without explanation about quantities. For example, what are A(t,f), and f? The speed is sa or Sa? If this speed is used in some figures, please add the proper symbol. Check and describe the parameters used in througout the paper.

6) A short description of the instrumentation could be added (see pag. 3).

7) It is not clear how parameter K' and beta are extracted (see par. 5). A comment can be added.

8) Parameter alpha depends on the seismic characteristics of the ground (pag. 14). Is it possible to obtain it from some geological and geotechnical analysis? Are there any data about the site?

This part is very important (crucial) and should be improved. The procedure "is circular", so that it should be explained better. It is sometime difficult to understand if the paper refers to an avalanche speed from GEORADAR or from calculation.

Author Response

Dear reviewer 2,

Thank you very much for your interesting comments. Below you will find our answer.

The manuscript has been corrected in accordance with your suggestions. Moreover, following the suggestion of rev.1 we have reorganized the manuscript. We now hope that this meets with your requirements.

Interesting paper that requires some work to be more clear and readable.

Comments:

1)          All the figures in the paper must be improved. Labels for axis (in particular the vertical one is always missed so that is hard to understand the meaning (check, among other, Fig. 3)

Thank you.  In all figures of seismograms, the axis of time (s) was omitted expressly because the time scale is the same for all the plots and the label “Time(s)” was indicated in the figure at the bottom. As suggested, we include labels in all the seismograms. We also revised the other labels carefully.

2) Fig. 3 should be split in two and enlarged, otherwise is very hard to understand. Again, what is the meaning of the axis?

 We have included all the information in the same figure for a good comparison between the avalanches. Moreover, for us, the comparison of the time series and spectrograms is important because their information is complementary and should be read together. As suggested, we have split the figure into three figures according to the avalanche type showing time series and spectrograms in the same figure including the most important information.  Additionally, we have incorporated most of the information related to this figure into the appendix B as suggested by rev 1.

Fig. 10 and 11 (among others) should be enlarged and improved. The same for the pictures in the Appendix A and B.

Thank you. Done.

4) Text refers to Appendix A1 or A2  but the Appendix are label A and B. Please check your manuscript.

This has been corrected. Note, however, that we were referring to the subsections A-A1, and A-A2.

5) Eqs. 1 and 2 are presented without explanation about quantities. For example, what are A(t,f), and f? The speed is sa or Sa? If this speed is used in some figures, please add the proper symbol. Check and describe the parameters used in throughout the paper.

We have included the explanation of the quantities with their units as suggested.  

The speed in line 200 is s_a. It was a typing error. Thank you. However, to avoid any misunderstanding we replace sa by S (of speed) everywhere.

6) A short description of the instrumentation could be added (see pag. 3).

Done. In fact, this information is contained in the works referred. However, we have included it as suggested.

7) It is not clear how parameter K' and beta are extracted (see par. 5). A comment can be added.

The explanation of the method has been moved from Appendix A to the main text as suggested by rev.1. We hope that this problem has been resolved in this new version.

8) Parameter alpha depends on the seismic characteristics of the ground (pge. 14). Is it possible to obtain it from some geological and geotechnical analysis? Are there any data about the site?

No. To our knowledge no seismic ground characteristics related to alpha are known for the specific VdLS experimental site. Pilz and Fäh (2017, JOSE)  presented values of k (s^-1) (spectral decay factor) for the Swiss territory. Although the alpha and this parameter are both related to the intrinsic attenuation, our alpha has units of m^-1 and k has units of s^-1. As a result, these values are not comparable.

 This part is very important (crucial) and should be improved. The procedure "is circular", so that it should be explained better. It is sometime difficult to understand if the paper refers to an avalanche speed from GEORADAR or from calculation.

Note that we have adopted the different terms, speed and velocity as indicated in line 535 of the original manuscript to distinguish between them. However, in this new version we specify the difference explicitly.   Additionally, we have incorporated more explanations and one diagram to describe the “circular” procedure better.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

%COMMENTS FOR THE AUTHORS
The paper has been substantially improved. The applied method is now more clearly explained. Instead, the validation of the method remains questionable. Practically the paper concludes that the geodar data (the validation data) are not relevant for powder snow avalanches, but neither for some wet avalanches. In summary, 5 experimental measures remain, compared to hundreds of cases that can be found in the VDLS databases. Further, the paper still contains a myriad of small typos and inaccuracies.  I found some of them, probably they are much more. Please check the consistency of symbols in the all paper.

Here the  details:

Line 14 . check VdlS trough the text sometimes is VdLS sometime VdlS…
Line 16:  powder, transitional……,  and wet snow avalanches.
Line 18-19 : Improve formulation: the speed is contained in B
Line 21: Improve formulation: a rationale was established
Line 26: What do you mean for estimate the avalanche evolution exactly?
Line 107: test site
Line 124: Square brackets are missing [30], [48]
Line 135:s.p.s.? Hz?
Line 142-150: why is now the text bold?
Figure 1 has too low quality. The legend text should also be improved.  Remove the white inbox. This information is already in the caption.
Figure 2: This is not a DTM, It is a terrain profile. Half of the text is bold.  Missing a parenthesis (
Table 1: What does it mean B&900 or B&1000…?
Line 192:193:  Not clear what you mean here….
Line 320: becomes eq. 1 or 7?
Line 336; Why Ai anb Bi are without () and a and b are with ()?
Appendix B : line 844-856 ; again the text is in Bold
Line 923-924: line jump
Line 325. What is figure S3?
Paragraph 5.1: please check again this paragraph looking for consistency in equation numbers, variables definitions and be sure to have included all necessary references.
Line 386-387: what do you exactly mean with problematic values?
Line 435: at this position, location
Figure 8: the color of figure 8 are different from the color of Figure 7 and this make difficult to compare the avalanches.
Line 417, line 421-422, line 440….the behavior will be discussed below…. where?
Line 440-441: Why you eliminate the avalanche? Why the other ones are treated with cautions? The fact that the avalanches do not fit into the expected range is not the problem of the avalanche, but probably the problem of the analysis you adopt. You really need to discuss this point in the paper.
Figure 9 - Figure 10 -Figure 11:  the graphics titles are not consistent. The title of figure 9 should be “evolution powder avalanches”. Why you use a title for your graphics? Normally the caption should contain the necessary information. Please, introduce again the “problematic” avalanches in your graphs. The reader should see them also if they do not fit to your theory.
Figure 10 should display transitional avalanches but it is a mix between wet and transitional. The caption may be used to explain why. It is not clear to me which is your criteria to split avalanches.
Line 519: you cannot ignore data without a clear explanation why you are doing so.
Figure 12: what is “alfa e”? is not in the figure
Line 558: VBGDR , VCGDR or VB_GDR…….control all other variable in the caption and text.
Line 561: what is alfa e? It is mentioned in the caption but not included in the table  
Line 578-581. I do not understand from where this consideration related to entrainment are coming from. Do you have any evidence that the avalanches entrained/ not entrained snow at these locations? Why should be the entrainment on a wet avalanche negligible?  
Line 592-595: this reasoning does not make much sense to me. The flow regimes cannot change just because the avalanche released by slightly different locations. There should be other reasons for that. What would be the effect of preexisting deposits and high snow depth over the sensors, for example? Other reasons?
Line 597-599: what do you mean, we did not consider the avalanches not to alter the results?  
Figure         13: a, b, c ,d should be integrated into the figure….
Line: 679 38
Line 685, line 698 , line 717: VB_GDR … ? The symbol is changed …..

Author Response

Please see the attachement

Author Response File: Author Response.pdf