the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 26 Sep 2023
Real-time Monitoring and Analysis of Debris Flow Events: Insight from seismic signal features and dynamic flow characteristics
Abstract. Debris flows are among the most dangerous natural hazards worldwide because they start abruptly, move quickly, and transport large boulders, causing great loss of life and infrastructure. The most important approach to preventing and mitigating debris flows is through monitoring and early warning. In recent years, environmental seismology has emerged as a powerful method for monitoring debris flows because it allows non-contact observation over large areas and can provide extensive information on debris flow dynamics. However, further research is required on combining debris flow imagery with seismic signal analysis, incorporating information from post-disaster surveys, and the inversion of seismic signals into dynamic parameters of debris flows. Here, we aim to explore the basic parameters, development process, and magnitude of debris flows based on seismic signal analysis combined with other information recorded in real time during the formation and development of three debris flows in Wenchuan, China. The analysis involves three stages. First, we compensate for the energy loss of the seismic signal due to the absorption attenuation effect and restore the signal to an unchanged state as far as possible. Second, we identify the start and end time of the debris flow from the seismic signal, analyze the rainfall data to determine that the debris flow was triggered by the test rain, and determine that changes in the energy and frequency ranges of the seismic signal are highly consistent with the development of the debris flow. Third, a comprehensive analysis of debris flow images, the power spectral density (PSD) of the seismic signal, and forward modeling of the PSD of the seismic signal of the debris flow are used to reveal the relationship between the seismic signal and the development process of the debris flow and clarify the feasibility of debris flow analysis from the time-frequency characteristics of the seismic signal. Debris flow exhibits the characteristics of fast excitation and slow recession. Using the cross-correlation algorithm and verifying Manning's formula, a maximum velocity of 7.027 m/s was calculated for the second debris flow. A comparison of the frequency characteristics of the seismic signal allowed the relative magnitude of the three debris flows to be assessed. The study provides a theoretical basis and a case study exemplar for the reconstruction of the debris flow process and peak velocity estimation using debris flow seismology, offering a framework for upscaling debris flow monitoring networks and the determination of early warning thresholds.
- Preprint
(2784 KB) - Metadata XML
- BibTeX
- EndNote
Status: closed
-
RC1: 'Comment on egusphere-2023-2015', Anonymous Referee #1, 09 Nov 2023
-
AC1: 'Reply on RC1', Yan Yan, 26 Jan 2024
Thank you for the professional and pertinent advice on the manuscript structure and key point of this manuscript. After changing structure of the manuscript, we cannot achieve the validation of the methodology, the interpretation of the data, the quantitative description of the events. The quality of the manuscript (redaction and figures) has been improved.
After carefully analyzing the constructive guidance given by the reviewer regarding the manuscript’s innovations, objective, etc., we determined the innovations of this study “a theoretical basis and a case study exemplar for the real-time monitoring, analyzing the debris flow by a debris flow monitoring system based on the core of seismic monitoring, offering a framework for extreme environment upscaling debris flow monitoring networks, the determination of early warning thresholds and hazard assessment and analysis”.Please see the attachment.
-
AC1: 'Reply on RC1', Yan Yan, 26 Jan 2024
-
RC2: 'Comment on egusphere-2023-2015', Anonymous Referee #2, 27 Nov 2023
General comments
In this study, seismic sensors have been used to investigate three debris flow events that occurred in the Fotangbagou and Ergou catchments, in Wenchuan (China). The authors combined seismic observations with rainfall measurements, photos by infrared cameras, and a post-event survey to get additional information on the debris flows. The velocity of the debris flows has been estimated through the seismic signals recorded at different stations and then compared with results from the application of the Manning formula, while the relative magnitude of the events has been inferred after taking into account the decay of the seismic energy during propagation. The topic addressed here is relevant and within the scope of ESurf, as although seismology has been used to monitor debris flows for decades, the link between the seismic signature and the properties of the events still needs to be properly understood. The quality of the seismic data is good, and the study areas seem interesting. However, I find the current state of the manuscript far from being considered for publication. My major concerns are (i) the large gap between the aims raised by the authors and what is actually shown afterwards, (ii) the lack of accuracy in most of the methods and analyses shown herein, and (iii) the quality of the writing.
The core of the abstract and the introduction is the aim of inverting the seismic signals into dynamic parameters of debris flows to provide a “theoretical basis for reconstruction and inversion of the debris flow process”, and offer “a framework for upscaling debris flow monitoring networks and the determination of early warning thresholds (e.g. lines 31-33 and 52-56 in the abstract, and lines 124-128 and 134-137 in the introduction). However, among the many debris flow parameters (e.g. flow height, flow volume or mass, velocity, solid concentration) only the velocity of one debris flow out of the three is estimated. I believe that this gap between scientific questions and results is mainly due to a lack of independent information on the debris flow events. The rainfall measurements have been used to propose rainfall as the triggering factor of the debris flows, but they cannot help deciphering the flow characteristics. The grain size distribution’s estimation of one debris flow through the post-event survey can been exploited only partially, and the authors use different diameters for the modelling section (please see my specific comments on this topic below). Images from infrared cameras usually give useful insights for monitoring debris flows, but they are available here only for the debris flow occurred in the daytime, and the camera frame rate of 5 minutes seems too low to me to catch the highly variable nature of debris flows and the associated seismic signals. Moreover, at this stage there is no mention along the text about how this work can be used for early monitoring systems, and as underlined by the other reviewer I don’t see applications to the “real-time monitoring” mentioned in the title.
My second concern is the reliability of the analyses, since I cannot say whether all the methods are applied correctly. As an example, the method for the compensation of seismic energy dissipation is not clear to me. The authors should provide more thorough explanations, as almost all references are not present or not accessible. Similarly, the choice of the parameters used in most of the equations ((5), (6), (8), (9), and (10)) is vague and not straightforward. How do you estimate the channel slope and roughness, and the flow lengths and heights? You only briefly mention the procedure in the results section without details, and most importantly you often do not show the values. What about the seismic parameters (attenuation factor Q, Rayleigh wave velocities and seismic travel time)? It is understandable that you couldn’t estimate them in the field, but you should at least give some references and discuss the errors associated with your choices, as your conclusions rely on them. Moreover, I find most of the interpretations of the results as speculative, meaning I often barely see what the authors claim to observe in the figures.
Finally, I agree with the other reviewer that the quality of the writing should also be strongly improved. In several parts it is hard to follow the text, the vocabulary is not correct and some physical quantities are called with different terms along the manuscript. A thorough revision is therefore required, and I believe this could improve the clarity of the work.
For all these raisons, I recommend major revisions before this manuscript can be considered for publication. I think that an important effort must be made by the authors in order to address these points. My opinion is that the authors have the data to write a nice contribution, but most of the analyses need to be revised and the structure redesigned.
My main suggestion is to change the aims of this work. Since I find it too speculative to invert the debris flows dynamic parameters in this context, this manuscript should be rather presented as a “case study” where to show the preliminary results of two new monitoring stations in catchments prone to debris flow events. To do so, the authors could first compare the debris flows they observe with existing observations in the same catchments (e.g. Guo et al, 2016), for example with respect to the triggering rainfall, and underline the potential additional information gathered with the seismic sensors (e.g., without the seismic sensors, how do we know if debris flows occur in these catchments? This is not clear to me and should be clarified). Within this new structure, I believe that some first inputs for the development of early warning for the study areas would fit better. I suggest the authors to analyze the seismic signals associated with rainfall events that did not trigger debris flows. In this way they could propose some seismic thresholds (e.g. Coviello et al., 2019). Regarding this latter point, the authors should not over-interpret the seismic signals, frankly acknowledge the limitations of the monitoring stations, and discuss what could be improved (e.g. are the authors sure that the camera frame rate is high enough to get the debris flows dynamics? Why didn’t the camera work well at night? Could it be possible to install flow height sensors in such gullies?,…). Some of these limitations appear in the discussion section, but if the authors were really aware of them, several analyses shouldn’t have been carried out. Comparisons with other studies on the seismic monitoring of debris flows are not present in the discussion section, yet I consider them as necessary. I acknowledge that to follow my suggestions all the structure of the manuscript has to be modified, and it may require a huge work by the authors, but I honestly believe that this is still the simpler way to valorize their findings. This is also in line with the criticisms raised by the other reviewer.
Below are my specific comments to the authors on the different sections of the manuscript.
Specific comments
Abstract
I think that the abstract should be re-written and shortened. Although it is important to explain the methods, the main findings of the work should appear more clearly and not only in 3-4 lines at the end of it (lines 48-52).
Lines 27-28: these lines are not clear to me, because in my view early warning is a consequence of monitoring. I would say that it is important to monitor debris flows to better understand their dynamics and also define thresholds for early warning systems.
Lines 28-29: I would change “non-contact observations” with “remote observations”
Lines 31-37: according to my general comment, the scientific questions and aims of the work should be soften. The main subject could be the Wenchuan area and the high frequency of debris flows events, which need to be monitored.
Line 32: what do you mean by “imagery”?
Line 34: what do you mean by “basic parameters”? Please tell us which ones
Line 35: I propose to change “other” with “additional”
Line 38: “absorption attenuation effect” is a bit hard to digest. I would simply talk about the energy loss of the seismic signal during propagation in the ground
Line 39: what do you mean by “as far as possible”?
Line 41: what is the “test rain”? Do you mean the rainfall observed?
Lines 41-43: I’m not sure that this is shown in the text. Please see my comments on the result section
Lines 43-48: I find these lines not clear. I would remove the part which starts from “clarify the feasibility”.
Line 48: instead of saying “fast excitation and slow recession” I would be more clear, saying that the seismic signature of the debris flows is characterized by an abrupt increase of seismic power and a slower decrease. Is the increase in seismic power related to the passage of the front? This aspect needs to be discussed later on
Line 49: it is too strong to say “verifying Manning’s formula”, also because in the text you seem to say the opposite (the Manning’s formula confirms the cross-correlation). I would rather say that you estimate velocities with two independent methods
Line 50: three significant digits seem to many to me, given the errors associated with the method. I would also remove “maximum” velocity, since it is more a mean velocity
Lines 52-55: this end should be modified if you follow my suggestion
Keywords
I would put seismic methods instead of seismic wave
Introduction
Lines 60-61: I would make the distinction between mobilization from landslide or surface runoff. Please add some references
Lines 66-68: “disaster reduction measures” is too vague to me, since it would also include check dams and deposition basins. I suggest you to be more explicit, here and along the text.
Line 68-70: the sentence is a bit repetitive. Please reformulate with something like “On-site monitoring provides information on the triggering mechanisms (e.g. rainfall events) and the characteristics of debris flows such as flow depth,…, which can be used to develop warning systems”. I don’t see how flow velocity can be used for early warning systems
Line 102: remove “or so”
Line 104: please comment briefly on how environmental seismology works, saying that natural processes generate ground vibrations
Line 109: I propose you to change “non-contact monitoring” with “remote monitoring”. What do you mean by “rich information”? I would remove it.
Line 111: if I’m correct, Marchetti et al. (2019) use infrasound measurements
Line 115-119: if you want to talk about debris flow models, you need to introduce all of them and not only the one by Lai et al. (2018). Please see Farin et al. (2019), Zhang et al. (2021)
Line 122: flow is not a parameter
Line 123: what do you mean by “identification”?
Line 132: please change “comprising” with “composed of”
Lines 131-134: these sentences are repetitive. Please change them with “The in-gully monitoring systems are composed of seismic sensors, rainfall gauges, and infrared cameras.
Lines 134-137: following my general comment, these lines are too strong. At this stage, this study cannot offer “a framework for establishing a debris flow identification, monitoring, and early warning systems”. In my view, several aspects of your monitoring systems can be improved, and there is no mention about early warning in the current text.
Study site and field monitoring system
Line 151: please add some references about the occurrence of debris flows in this area
Figure 1: I’m not sure that Figure 1c is necessary, since you do not mention the geology of the area with this precision. If you remove the panel and all the legend, Figure 1a and 1c could gain some space. In the caption of the figure, say that the catchments of interest are in red. I don’t get Google Earth 2015/2018: why these two years?
Lines 163-164: “a narrow and winding channel” before “abundant water sources”
Line 161-164 and so on: please decide between degrees and percentage for the slope
Line 171: what are “adequate” water sources?
Line 172: what is the “average slope ratio”? Isn’t it just “slope”?
Line 181: please remove “etc.”: either you mention all the quantities measurements, or you mention only what you use in this work
Line 182: “seismic monitoring” is better than “seismic signal monitoring”
Line 185: it is not clear to me what is the eigenfrequency of 150 Hz, given that your sampling frequency is 100 Hz
Line 189: “other data” is too vague. I would say that infrared cameras give insights on the debris flow processes. Also “verify the seismic reconstruction” is a bit too strong. I would say rather say “compare with the seismic observations”
Methodology
The workflow in Figure 3 is not clear to me. Please follow the suggestions of the other reviewer
Line 206-2012: I’m not sure if all these details are needed because the short-time Fourier transform is a common method in signal processing. I propose to just mention the language you use
Lines 214-222: this part can be misleading. You need to clarify that you are talking about seismic signals, since the works by Arattano and Marchi (2005) and Comiti (2014) apply this method also to flow stage measurements. I suggest you to take inspiration from Arattano and Marchi (2005) to be more clear in the explication.
Have you computed the signal time delay with equation (4), and then you divide by the distance between the stations? Please be more explicit at line 221. Have you considered the distance between the stations along the channel or the straight distance?Line 225: I think that you must acknowledge that the Manning formula has been originally developed for hydraulics problems (Open channel flow, F.M. Handerson (1966)). Add to the references also (Rickenmann, 1999)
Line 231: what do you mean by “slope ratio”? Please define the hydraulic radius and tell us how you estimate it
Line 233: d50 is the median particle size of the channel bed? I’m not sure that you have used equation (6) to estimate the roughness coefficient, since the n=0.05 you use afterwards corresponds to a d50=1.33 m if I’m not wrong. How have you estimated this d50?
Equation 7: Is S(t,f) the same as X(n,w) in equation (1)? If yes, please keep the language consistent.
Line 239: how do you define fmin and fmax?
Lines 242-257: why have you chosen the model by Lai et al. (2018)? Since several models exist, you need to justify your choice (Farin et al., 2019; Zhang et al., 2021)
Lines 249-259: as stated in my main comment, this method is not clear to me as most of the references are not accessible. Please add more details, especially on equation (10). From what I understand, you could have estimated the energy loss during propagation through the Green’s function as it has been done in several works you mention (e.g. Tsai et al., 2012; Lai et al., 2018). What is the spreading time and what is its value in equation (9)? How have you chosen Q? Is this the same Q you use while applying Lai et al. (2018)? If so, you should call it in the same way. What is sigma in equation (10)? More in general, I haven’t understood how you combine equation (9) and (10) to restore the signal. At this stage, I cannot say if I’m convinced by the estimation of the relative magnitude of the events based on these equations.
Results
Section 4.1
It is not clear to me if you know that these three debris flows occurred from other independent observations or just from the seismic observations. In the latter case, you should convince the reader that all these events are debris flows and not just intense sediment transport events. It would be also a way to highlight the usefulness of the seismic sensors. Maybe from some critical rainfall thresholds already observed in these catchments? On the other hand, if you know they were debris flows from other observations, please clarify it. For this reason, I should start the section by commenting on the rainfall measurements, and only then on the spectrograms. I propose you to merge Figure 4 and 5 in order to see (or not) the match between rainfall and peak seismic power, because now it is hard to follow your comments.
Table 2: how do you define the starting and ending time of the debris flows? Please explain your method. For example, an increase of decibels over a certain threshold, a rapid rise, the decrease below a certain threshold. I believe that the times shown are quite approximate (3:00, 7:30, 2:00): can’t you be more precise?
Are you sure that the first peak of seismic power is related to the passage of a debris flow? Shouldn’t it be just sediment transport + rainfall (e.g. Rindraharisaona et al., 2022)? That could be the reason why the frequency band is narrower compared to the second event at 7:20, which really looks like a debris flow. More comments are needed on Figure 4. You also never mention the difference between the different parts of a debris flow (front, or body). For instance, the peak of seismic power is usually associated with the passage of the front as it contains the biggest particles. Comments are needed about this aspect. I think that power is more correct than energy next to the spectrogram.
Line 266: the subplots are wrong, the frequency bands are visible on the other ones
Line 269: here and along the text, I propose you to change “time-frequency spectrum” with “spectrogram” and to keep consistency along the text
Line 291: isn’t the cumulative rainfall of the first debris flow event in the Fotangbagou gully 15.6 mm?
Lines 298-321: I must admit that it is really hard for me to follow this part. For what I understand, your idea is to recover the energy loss during the propagation of the seismic wave. If this is true, lines 301-304 are describing another problem, that is the fact that what the sensor records are the seismic waves generated by the entire debris flow (e.g. the front and the body of it), which is not the focus of your approach.
What do you mean by frequency and velocity dispersion at line 299?
What do lines 305-306 mean? What do “river channels are about 10 m around the site during the processing signal” mean? Is 10 m the width?
How have you computed the average the average travel time?
How have you chosen the values of Q and Rayleigh wave velocities and why they change from a site to another? It is crucial to explain your choices, giving references or discussing them.
What do you mean by “its characteristics change is more obvious” at line 318?Figure 6: more comments are needed for this figure. Why the function h is so different between a site to another? How do you interpret it? Please check the caption, I think that the number of the monitoring stations are not always correct. I also suggest to use spectrogram instead of “time-frequency domain energy spectrum”, which is also different from the vocabulary used in Figure 4
Line 335: what do you mean by “effectiveness of the debris flow evolution process”?
Lines 340-342: please use the same terminology, you have changed again the term for the spectrogram and it is the first time that you mention the vertical direction
Section 4.2.1
The subplots in Figure 7 already appear in Figure 6, therefore I would remove this figure and refer to Figure 6. Are the spectrograms computed after restoring the signal? The caption makes the reader think they are.
More comments are needed to present the spectrograms. How have you computed the bandwidths you talk about at lines 347-348? At station 2, the spectrogram has no power under 8 Hz: why does it happen? Is it the result of a filtering process?
Why the seismic power remains relatively high for so much time (until 10:00)? Is it sediment transport?Lines 361-362: if the average amplitude at station 1 is higher than at station 2, it is trivial that the power is also higher, because the power is computed from the signals amplitude.
Section 4.2.2
I find the use of the infrared camera interesting and I acknowledge your effort to get the maximum information possible from the images. However, I find most comments on the figure too speculative. How can you observe an increase in particle content from Figure 8a to 8d? How can you say that the flow velocity increases if images are static? I cannot see signs of erosion of the left bank you mention at line 388, and how can you see that the channel is smooth at point C (line 394)? I don’t see a decreasing velocity after 7:39 (line 396), and the presence of a rock at point A is not sufficient to conclude that the transport capacity is low (lines 412-413).
I honestly believe that these images can be used to identify the passage of the debris flow, and maybe to get some insights on the flow stage, but any comment on flow velocity and concentration is too vague to me.
Why have you analyzed only images from 7:39 to 8:04 given that the debris flow lasts longer? I think it is crucial to see the condition of the channel before the development of the debris flow. I suggest you to mark different part of the section before the debris flow (e.g. banks, sediment deposits), so that we can better visualize the changes and the magnitude of the event. It is also important to see images after the event: I wonder if the big rocks we see in Figure 8f are carried by the flow or just deposited.Section 4.2.3
Post-event field investigations are important and it is nice that you made some measurements. However, I don’t see where you have used the information you got from the survey and how they can help answering your scientific questions.
Line 420: where is point C?
Lines 424-426: these sentences are vague. In Figure 9c I still see very big rocks. Moreover, the fact that at a very specific point of the fan there are some small particles, doesn’t mean that the carrying capacity of the debris flow sharply decreases. I suggest you to remove this part.
Lines 427-438: please specify that from your sample you can only quantify the small fraction of the deposit. For the same raison, it is not correct to say that the 94th percentile of the grain size distribution in 0.018 m: look at the big rocks in the photos.
Are you sure that you have used this value for equation (8)? In the following section you use much bigger values.Figure 9: In Figure 9a you use “deposition fan”, in the caption you use “accumulation fan”. Please choose one of them.
Section 4.2.4
I’m sorry but I’m not sure about the need of this section. The changes you claim to see in Figure 10 are really small, especially from 7:44 to 08:04, maybe less than 1 decibel. I believe that this can be within the errors associated with your computations. If you want to investigate the variation of frequency over time, maybe you could compute the frequency peak or the mean frequency as in Farin et al. (2018). However, it should be done continuously and not only for the 6 time intervals as you do.
Similarly, the application and interpretation of the model by Lai et al. (2018) is vague to me. How have you got D=0.5-0.6 m if the 94th percentile of your sample is just 0.018 m? Why have you chosen velocities ranging between 2 and 6 m/s? Have you taken these values from the literature? This choice looks weird also because later you do estimate the velocity, so why haven’t you tested your estimation? Maybe it would be interesting to test the model by Lai et al. (2018) with the velocity you estimate, and see what is the diameter that gives you a seismic power similar to observations. However, some issues still remain as it is not clear to me how you can estimate the debris flow length and the seismic parameters in equation (8). I believe that the uncertainties on these parameters are too high to interpret the result of the modelling.
The interpretation of your results looks also incorrect to me. In equation (8), both diameter and velocity are to the third power, therefore you cannot say that particle size has a minor effect than velocity. Your interpretation is due to the fact that the velocities you have tested vary much more than the diameters (you triple the velocity from 2 to 6 m/s, but the diameter only varies by a factor of 1.2 between 0.5 and 0.6 m) (lines 479-484). Moreover, at line 502 you say that “the particle content in one of the factors affecting the energy of seismic signals”, but the model by Lai et al. (2018) makes the opposite assumption of constant particle content. If you believe that particle content plays a role, why have you used this model?
For all these raisons, I would remove this section, unless you can solve the points I have raised.
Section 4.2.5
Line 518: what do you mean by “horizontal distances”?
Line 519: Why have you chosen a gain factor of 1.8? Is it the sigma of equation (10)? Are the spectrograms resulting from the seismic signal restoration? If not, I don’t see how the restoration is useful here given that all your comments concern the spectrograms.
Lines 522-537: it is really hard to see what you observe in the figures. I propose to show the different times with lines in the spectrograms.
Figure 11: more comments are needed. As I have already said, why in Figure 11b and 11d the seismic power is zero below 10 Hz? Is the signal filtered or a different characteristic of the instruments compared to the ones in the Ergou gully? In both cases, this is an important aspect to clarify since it affects all the interpretation about the frequency bands. Again, there are no comments about the passage of the debris flow front: you assume that the first peak in seismic power is related to the passage of the debris flow, but it could also be the front approaching towards the seismic station. Finally, are you sure that in Figure 11f we see the seismic signature of a debris flow? It could be rainfall in my opinion (Rindraharisaona et al., 2022).
Lines 550-552: the two sentences repeat the same concept
Line 556: I cannot see the decay towards 23 Hz. Can you explain this comment?
Section 4.3: as I said before, I’m not convinced about the method you have used. However, I will make some comments on the text.
Lines 569-571: please try to be more clear. The decay of seismic power is not a problem by itself. You should say that you need to take into account the distance between the sensors and the debris flow if you want to estimate the relative magnitude of the events.
Line 575: what are these values? m/s? You should also mark the peaks in the figures (and mention the figures where you can see these values). Which station have you considered? This must be clarified
Lines 576-578: how do you compute the frequency bands? How can you use the frequency bands to get the magnitude of the debris flows? The larger the frequency band, the bigger? Who says that?
Section 4.4
This section is interesting, but several aspects must be clarified and you should convince me about the parameters you have used
Line 580: in my opinion you haven’t estimated the maximum velocity but rather a mean velocity of the debris flow, as you consider all the signal
Lines 583: even if velocities are shown in the table, you need to recall them in the text. What do you mean by “normal”? Give some references
Lines 583-586: from these lines it is not clear to me if you are talking about the cross-correlation method. Comments are needed also for the other debris flows
Line 586: three decimal digits seem too many to me given the uncertainties.
Figure 12: you should show all the three debris flows. It is not clear to me if the curves you show come from the cross-correlation routine or are just the amplitudes
Line 595: it is a bit strong to say that you use the Manning formula to verify the velocity calculations, because also the Manning formula has its own uncertainties. I would say that you have estimated the velocity with two independent methods
Line 599: please be more clear about the estimation of the roughness coefficient. Is it true that you have estimated it with equation (6)? What value of d50 have you used?
Lines 604-607: these sentences should appear in the methods. Please tell us more precisely how you have computed the hydraulic radius and the slope, as they are crucial terms in the Manning formula
Line 611: comments are needed as it means that between the two stations the velocity is quite constant
Lines 611-612: as already said, it is not correct to say that one method verifies the other
Figure 13: I’m confused by this figure, since in Figure 2 you show the infrared camera only at station 1. How many infrared cameras do you have on the Fotangbgou? If two, why have you used only one camera in the previous sections? If one, how have you estimated the flow stage at station 2? Please clarify this aspect
Discussion
As a general comment, a better job should be done in this section. The discussion is the place to compare your findings with existing works and you have done it only in section 5.2. If you follow my main suggestion, section 5.3 is good to discuss on the limitations of this monitoring system and on what could be improved, taking advance of the monitoring stations already existing around the world. Comparisons with other works are also needed on the values of velocity.
Section 5.1
I would remove this section, since you are not adding discussion points but only repeating your findings. Moreover, several parts are not clear to me: at line 633 I don’t see how the kinematic parameters vary with topography; I don’t understand how the distance between the sensor and the channel can affect kinematic parameters (maybe you wanted to say that the distance must be taken into account if one wants to use seismic sensors to estimate kinematic parameters?); at line 639-640, the meaning of “seismic features select representative analysis points” is obscure to me.
Conclusions
The conclusions should be adapted with the respect to the new structure of the manuscript
Line 708: you say “large difference”, but at lines 629-630 you seem to say the opposite
References
Coviello, V., Arattano, M., Comiti, F.,Macconi, P., & Marchi, L. (2019).Seismic characterization of debrisflows: Insights into energy radiationand implications for warning. Journal of Geophysical Research: Earth Surface,124, 1440–1463. https://doi.org/10.1029/2018JF004683
Farin, M., Mangeney, A., de Rosny, J.,Toussaint, R., & Trinh, P.-T. (2018).Link between the dynamics of gran-ular flows and the generated seismicsignal: Insights from laboratory experi-ments.Journal of Geophysical Research:Earth Surface,123, 1407–1429.https://doi.org/10.1029/2017JF004296
Farin, M., Tsai, V. C., Lamb, M. P., & Allstadt, K. E. (2019). A physical model of the high-frequency seismic signal generated by debris flows. Earth surface processes and landforms, 44(13), 2529–2543. https://doi.org/10.1002/esp.4677
Rickenmann, D. Empirical Relationships for Debris Flows. Natural Hazards, 19, 47–77 (1999). https://doi.org/10.1023/A:1008064220727
Rindraharisaona, E. J., Réchou, A., Fontaine, F. R., Barruol, G., Stamenoff, P., Boudevillain, B., et al. (2022). Seismic signature of rain and wind inferred from seismic data. Earth and Space Science, 9, e2022EA002328. https://doi.org/10.1029/2022EA002328
Zhang, Z., Walter, F., McArdell, B. W., Haas, T., Wenner, M., Chmiel, M., & He, S. (2021). Analyzing bulk flow characteristics of debris flows using their high frequency seismic signature. Journal of Geophysical Research: Solid Earth, 126(12). https://doi.org/10.1029/2021JB022755
Citation: https://doi.org/10.5194/egusphere-2023-2015-RC2 -
AC2: 'Reply on RC2', Yan Yan, 26 Jan 2024
Thank you for spending the time to review and assess our manuscript. After carefully analyzing the reviewers’ comments, we were deeply convinced that we had not been able to accurately summarize the innovations and research objectives of our current study in the previous manuscripts, and we carefully analyzed the content of our Methodology and research. We therefore determined the research purpose in this manuscript: a theoretical basis and a case study exemplar for the real-time monitoring, analyzing the debris flow by a debris flow monitoring system based on the core of seismic monitoring, the determination of early warning thresholds and hazard assessment and analysis. We have rewritten the abstract to emphasize the technical line of the study and the results obtained, highlighting the strengths of the study and eliminating “the large gap between the aims” as pointed out by the reviewer. We have rewritten the abstract, highlighted the technical line of the study and the results obtained highlights the strengths of the study and eliminates “the large gap between the aims”, as pointed out by the reviewer.
Regarding the reviewer’s suggestion of “the lack of accuracy in most of the methods and analyses shown herein”, we have made a targeted revision in this round of revision, and we have sorted out and refined the contents of section 4.1~4.3, rewritten the manuscript to avoid ambiguity, and corrected some of the errors, e.g., Fig.8a incorrectly labelled point A and C. The modification has improved the accuracy of analyses. We have made an overall introduction to the methodology and necessary modifications to highlight the feasibility of our research methodology.Please see the attachment.
-
AC2: 'Reply on RC2', Yan Yan, 26 Jan 2024
- AC3: 'Comment on egusphere-2023-2015', Yan Yan, 26 Jan 2024
Status: closed
-
RC1: 'Comment on egusphere-2023-2015', Anonymous Referee #1, 09 Nov 2023
-
AC1: 'Reply on RC1', Yan Yan, 26 Jan 2024
Thank you for the professional and pertinent advice on the manuscript structure and key point of this manuscript. After changing structure of the manuscript, we cannot achieve the validation of the methodology, the interpretation of the data, the quantitative description of the events. The quality of the manuscript (redaction and figures) has been improved.
After carefully analyzing the constructive guidance given by the reviewer regarding the manuscript’s innovations, objective, etc., we determined the innovations of this study “a theoretical basis and a case study exemplar for the real-time monitoring, analyzing the debris flow by a debris flow monitoring system based on the core of seismic monitoring, offering a framework for extreme environment upscaling debris flow monitoring networks, the determination of early warning thresholds and hazard assessment and analysis”.Please see the attachment.
-
AC1: 'Reply on RC1', Yan Yan, 26 Jan 2024
-
RC2: 'Comment on egusphere-2023-2015', Anonymous Referee #2, 27 Nov 2023
General comments
In this study, seismic sensors have been used to investigate three debris flow events that occurred in the Fotangbagou and Ergou catchments, in Wenchuan (China). The authors combined seismic observations with rainfall measurements, photos by infrared cameras, and a post-event survey to get additional information on the debris flows. The velocity of the debris flows has been estimated through the seismic signals recorded at different stations and then compared with results from the application of the Manning formula, while the relative magnitude of the events has been inferred after taking into account the decay of the seismic energy during propagation. The topic addressed here is relevant and within the scope of ESurf, as although seismology has been used to monitor debris flows for decades, the link between the seismic signature and the properties of the events still needs to be properly understood. The quality of the seismic data is good, and the study areas seem interesting. However, I find the current state of the manuscript far from being considered for publication. My major concerns are (i) the large gap between the aims raised by the authors and what is actually shown afterwards, (ii) the lack of accuracy in most of the methods and analyses shown herein, and (iii) the quality of the writing.
The core of the abstract and the introduction is the aim of inverting the seismic signals into dynamic parameters of debris flows to provide a “theoretical basis for reconstruction and inversion of the debris flow process”, and offer “a framework for upscaling debris flow monitoring networks and the determination of early warning thresholds (e.g. lines 31-33 and 52-56 in the abstract, and lines 124-128 and 134-137 in the introduction). However, among the many debris flow parameters (e.g. flow height, flow volume or mass, velocity, solid concentration) only the velocity of one debris flow out of the three is estimated. I believe that this gap between scientific questions and results is mainly due to a lack of independent information on the debris flow events. The rainfall measurements have been used to propose rainfall as the triggering factor of the debris flows, but they cannot help deciphering the flow characteristics. The grain size distribution’s estimation of one debris flow through the post-event survey can been exploited only partially, and the authors use different diameters for the modelling section (please see my specific comments on this topic below). Images from infrared cameras usually give useful insights for monitoring debris flows, but they are available here only for the debris flow occurred in the daytime, and the camera frame rate of 5 minutes seems too low to me to catch the highly variable nature of debris flows and the associated seismic signals. Moreover, at this stage there is no mention along the text about how this work can be used for early monitoring systems, and as underlined by the other reviewer I don’t see applications to the “real-time monitoring” mentioned in the title.
My second concern is the reliability of the analyses, since I cannot say whether all the methods are applied correctly. As an example, the method for the compensation of seismic energy dissipation is not clear to me. The authors should provide more thorough explanations, as almost all references are not present or not accessible. Similarly, the choice of the parameters used in most of the equations ((5), (6), (8), (9), and (10)) is vague and not straightforward. How do you estimate the channel slope and roughness, and the flow lengths and heights? You only briefly mention the procedure in the results section without details, and most importantly you often do not show the values. What about the seismic parameters (attenuation factor Q, Rayleigh wave velocities and seismic travel time)? It is understandable that you couldn’t estimate them in the field, but you should at least give some references and discuss the errors associated with your choices, as your conclusions rely on them. Moreover, I find most of the interpretations of the results as speculative, meaning I often barely see what the authors claim to observe in the figures.
Finally, I agree with the other reviewer that the quality of the writing should also be strongly improved. In several parts it is hard to follow the text, the vocabulary is not correct and some physical quantities are called with different terms along the manuscript. A thorough revision is therefore required, and I believe this could improve the clarity of the work.
For all these raisons, I recommend major revisions before this manuscript can be considered for publication. I think that an important effort must be made by the authors in order to address these points. My opinion is that the authors have the data to write a nice contribution, but most of the analyses need to be revised and the structure redesigned.
My main suggestion is to change the aims of this work. Since I find it too speculative to invert the debris flows dynamic parameters in this context, this manuscript should be rather presented as a “case study” where to show the preliminary results of two new monitoring stations in catchments prone to debris flow events. To do so, the authors could first compare the debris flows they observe with existing observations in the same catchments (e.g. Guo et al, 2016), for example with respect to the triggering rainfall, and underline the potential additional information gathered with the seismic sensors (e.g., without the seismic sensors, how do we know if debris flows occur in these catchments? This is not clear to me and should be clarified). Within this new structure, I believe that some first inputs for the development of early warning for the study areas would fit better. I suggest the authors to analyze the seismic signals associated with rainfall events that did not trigger debris flows. In this way they could propose some seismic thresholds (e.g. Coviello et al., 2019). Regarding this latter point, the authors should not over-interpret the seismic signals, frankly acknowledge the limitations of the monitoring stations, and discuss what could be improved (e.g. are the authors sure that the camera frame rate is high enough to get the debris flows dynamics? Why didn’t the camera work well at night? Could it be possible to install flow height sensors in such gullies?,…). Some of these limitations appear in the discussion section, but if the authors were really aware of them, several analyses shouldn’t have been carried out. Comparisons with other studies on the seismic monitoring of debris flows are not present in the discussion section, yet I consider them as necessary. I acknowledge that to follow my suggestions all the structure of the manuscript has to be modified, and it may require a huge work by the authors, but I honestly believe that this is still the simpler way to valorize their findings. This is also in line with the criticisms raised by the other reviewer.
Below are my specific comments to the authors on the different sections of the manuscript.
Specific comments
Abstract
I think that the abstract should be re-written and shortened. Although it is important to explain the methods, the main findings of the work should appear more clearly and not only in 3-4 lines at the end of it (lines 48-52).
Lines 27-28: these lines are not clear to me, because in my view early warning is a consequence of monitoring. I would say that it is important to monitor debris flows to better understand their dynamics and also define thresholds for early warning systems.
Lines 28-29: I would change “non-contact observations” with “remote observations”
Lines 31-37: according to my general comment, the scientific questions and aims of the work should be soften. The main subject could be the Wenchuan area and the high frequency of debris flows events, which need to be monitored.
Line 32: what do you mean by “imagery”?
Line 34: what do you mean by “basic parameters”? Please tell us which ones
Line 35: I propose to change “other” with “additional”
Line 38: “absorption attenuation effect” is a bit hard to digest. I would simply talk about the energy loss of the seismic signal during propagation in the ground
Line 39: what do you mean by “as far as possible”?
Line 41: what is the “test rain”? Do you mean the rainfall observed?
Lines 41-43: I’m not sure that this is shown in the text. Please see my comments on the result section
Lines 43-48: I find these lines not clear. I would remove the part which starts from “clarify the feasibility”.
Line 48: instead of saying “fast excitation and slow recession” I would be more clear, saying that the seismic signature of the debris flows is characterized by an abrupt increase of seismic power and a slower decrease. Is the increase in seismic power related to the passage of the front? This aspect needs to be discussed later on
Line 49: it is too strong to say “verifying Manning’s formula”, also because in the text you seem to say the opposite (the Manning’s formula confirms the cross-correlation). I would rather say that you estimate velocities with two independent methods
Line 50: three significant digits seem to many to me, given the errors associated with the method. I would also remove “maximum” velocity, since it is more a mean velocity
Lines 52-55: this end should be modified if you follow my suggestion
Keywords
I would put seismic methods instead of seismic wave
Introduction
Lines 60-61: I would make the distinction between mobilization from landslide or surface runoff. Please add some references
Lines 66-68: “disaster reduction measures” is too vague to me, since it would also include check dams and deposition basins. I suggest you to be more explicit, here and along the text.
Line 68-70: the sentence is a bit repetitive. Please reformulate with something like “On-site monitoring provides information on the triggering mechanisms (e.g. rainfall events) and the characteristics of debris flows such as flow depth,…, which can be used to develop warning systems”. I don’t see how flow velocity can be used for early warning systems
Line 102: remove “or so”
Line 104: please comment briefly on how environmental seismology works, saying that natural processes generate ground vibrations
Line 109: I propose you to change “non-contact monitoring” with “remote monitoring”. What do you mean by “rich information”? I would remove it.
Line 111: if I’m correct, Marchetti et al. (2019) use infrasound measurements
Line 115-119: if you want to talk about debris flow models, you need to introduce all of them and not only the one by Lai et al. (2018). Please see Farin et al. (2019), Zhang et al. (2021)
Line 122: flow is not a parameter
Line 123: what do you mean by “identification”?
Line 132: please change “comprising” with “composed of”
Lines 131-134: these sentences are repetitive. Please change them with “The in-gully monitoring systems are composed of seismic sensors, rainfall gauges, and infrared cameras.
Lines 134-137: following my general comment, these lines are too strong. At this stage, this study cannot offer “a framework for establishing a debris flow identification, monitoring, and early warning systems”. In my view, several aspects of your monitoring systems can be improved, and there is no mention about early warning in the current text.
Study site and field monitoring system
Line 151: please add some references about the occurrence of debris flows in this area
Figure 1: I’m not sure that Figure 1c is necessary, since you do not mention the geology of the area with this precision. If you remove the panel and all the legend, Figure 1a and 1c could gain some space. In the caption of the figure, say that the catchments of interest are in red. I don’t get Google Earth 2015/2018: why these two years?
Lines 163-164: “a narrow and winding channel” before “abundant water sources”
Line 161-164 and so on: please decide between degrees and percentage for the slope
Line 171: what are “adequate” water sources?
Line 172: what is the “average slope ratio”? Isn’t it just “slope”?
Line 181: please remove “etc.”: either you mention all the quantities measurements, or you mention only what you use in this work
Line 182: “seismic monitoring” is better than “seismic signal monitoring”
Line 185: it is not clear to me what is the eigenfrequency of 150 Hz, given that your sampling frequency is 100 Hz
Line 189: “other data” is too vague. I would say that infrared cameras give insights on the debris flow processes. Also “verify the seismic reconstruction” is a bit too strong. I would say rather say “compare with the seismic observations”
Methodology
The workflow in Figure 3 is not clear to me. Please follow the suggestions of the other reviewer
Line 206-2012: I’m not sure if all these details are needed because the short-time Fourier transform is a common method in signal processing. I propose to just mention the language you use
Lines 214-222: this part can be misleading. You need to clarify that you are talking about seismic signals, since the works by Arattano and Marchi (2005) and Comiti (2014) apply this method also to flow stage measurements. I suggest you to take inspiration from Arattano and Marchi (2005) to be more clear in the explication.
Have you computed the signal time delay with equation (4), and then you divide by the distance between the stations? Please be more explicit at line 221. Have you considered the distance between the stations along the channel or the straight distance?Line 225: I think that you must acknowledge that the Manning formula has been originally developed for hydraulics problems (Open channel flow, F.M. Handerson (1966)). Add to the references also (Rickenmann, 1999)
Line 231: what do you mean by “slope ratio”? Please define the hydraulic radius and tell us how you estimate it
Line 233: d50 is the median particle size of the channel bed? I’m not sure that you have used equation (6) to estimate the roughness coefficient, since the n=0.05 you use afterwards corresponds to a d50=1.33 m if I’m not wrong. How have you estimated this d50?
Equation 7: Is S(t,f) the same as X(n,w) in equation (1)? If yes, please keep the language consistent.
Line 239: how do you define fmin and fmax?
Lines 242-257: why have you chosen the model by Lai et al. (2018)? Since several models exist, you need to justify your choice (Farin et al., 2019; Zhang et al., 2021)
Lines 249-259: as stated in my main comment, this method is not clear to me as most of the references are not accessible. Please add more details, especially on equation (10). From what I understand, you could have estimated the energy loss during propagation through the Green’s function as it has been done in several works you mention (e.g. Tsai et al., 2012; Lai et al., 2018). What is the spreading time and what is its value in equation (9)? How have you chosen Q? Is this the same Q you use while applying Lai et al. (2018)? If so, you should call it in the same way. What is sigma in equation (10)? More in general, I haven’t understood how you combine equation (9) and (10) to restore the signal. At this stage, I cannot say if I’m convinced by the estimation of the relative magnitude of the events based on these equations.
Results
Section 4.1
It is not clear to me if you know that these three debris flows occurred from other independent observations or just from the seismic observations. In the latter case, you should convince the reader that all these events are debris flows and not just intense sediment transport events. It would be also a way to highlight the usefulness of the seismic sensors. Maybe from some critical rainfall thresholds already observed in these catchments? On the other hand, if you know they were debris flows from other observations, please clarify it. For this reason, I should start the section by commenting on the rainfall measurements, and only then on the spectrograms. I propose you to merge Figure 4 and 5 in order to see (or not) the match between rainfall and peak seismic power, because now it is hard to follow your comments.
Table 2: how do you define the starting and ending time of the debris flows? Please explain your method. For example, an increase of decibels over a certain threshold, a rapid rise, the decrease below a certain threshold. I believe that the times shown are quite approximate (3:00, 7:30, 2:00): can’t you be more precise?
Are you sure that the first peak of seismic power is related to the passage of a debris flow? Shouldn’t it be just sediment transport + rainfall (e.g. Rindraharisaona et al., 2022)? That could be the reason why the frequency band is narrower compared to the second event at 7:20, which really looks like a debris flow. More comments are needed on Figure 4. You also never mention the difference between the different parts of a debris flow (front, or body). For instance, the peak of seismic power is usually associated with the passage of the front as it contains the biggest particles. Comments are needed about this aspect. I think that power is more correct than energy next to the spectrogram.
Line 266: the subplots are wrong, the frequency bands are visible on the other ones
Line 269: here and along the text, I propose you to change “time-frequency spectrum” with “spectrogram” and to keep consistency along the text
Line 291: isn’t the cumulative rainfall of the first debris flow event in the Fotangbagou gully 15.6 mm?
Lines 298-321: I must admit that it is really hard for me to follow this part. For what I understand, your idea is to recover the energy loss during the propagation of the seismic wave. If this is true, lines 301-304 are describing another problem, that is the fact that what the sensor records are the seismic waves generated by the entire debris flow (e.g. the front and the body of it), which is not the focus of your approach.
What do you mean by frequency and velocity dispersion at line 299?
What do lines 305-306 mean? What do “river channels are about 10 m around the site during the processing signal” mean? Is 10 m the width?
How have you computed the average the average travel time?
How have you chosen the values of Q and Rayleigh wave velocities and why they change from a site to another? It is crucial to explain your choices, giving references or discussing them.
What do you mean by “its characteristics change is more obvious” at line 318?Figure 6: more comments are needed for this figure. Why the function h is so different between a site to another? How do you interpret it? Please check the caption, I think that the number of the monitoring stations are not always correct. I also suggest to use spectrogram instead of “time-frequency domain energy spectrum”, which is also different from the vocabulary used in Figure 4
Line 335: what do you mean by “effectiveness of the debris flow evolution process”?
Lines 340-342: please use the same terminology, you have changed again the term for the spectrogram and it is the first time that you mention the vertical direction
Section 4.2.1
The subplots in Figure 7 already appear in Figure 6, therefore I would remove this figure and refer to Figure 6. Are the spectrograms computed after restoring the signal? The caption makes the reader think they are.
More comments are needed to present the spectrograms. How have you computed the bandwidths you talk about at lines 347-348? At station 2, the spectrogram has no power under 8 Hz: why does it happen? Is it the result of a filtering process?
Why the seismic power remains relatively high for so much time (until 10:00)? Is it sediment transport?Lines 361-362: if the average amplitude at station 1 is higher than at station 2, it is trivial that the power is also higher, because the power is computed from the signals amplitude.
Section 4.2.2
I find the use of the infrared camera interesting and I acknowledge your effort to get the maximum information possible from the images. However, I find most comments on the figure too speculative. How can you observe an increase in particle content from Figure 8a to 8d? How can you say that the flow velocity increases if images are static? I cannot see signs of erosion of the left bank you mention at line 388, and how can you see that the channel is smooth at point C (line 394)? I don’t see a decreasing velocity after 7:39 (line 396), and the presence of a rock at point A is not sufficient to conclude that the transport capacity is low (lines 412-413).
I honestly believe that these images can be used to identify the passage of the debris flow, and maybe to get some insights on the flow stage, but any comment on flow velocity and concentration is too vague to me.
Why have you analyzed only images from 7:39 to 8:04 given that the debris flow lasts longer? I think it is crucial to see the condition of the channel before the development of the debris flow. I suggest you to mark different part of the section before the debris flow (e.g. banks, sediment deposits), so that we can better visualize the changes and the magnitude of the event. It is also important to see images after the event: I wonder if the big rocks we see in Figure 8f are carried by the flow or just deposited.Section 4.2.3
Post-event field investigations are important and it is nice that you made some measurements. However, I don’t see where you have used the information you got from the survey and how they can help answering your scientific questions.
Line 420: where is point C?
Lines 424-426: these sentences are vague. In Figure 9c I still see very big rocks. Moreover, the fact that at a very specific point of the fan there are some small particles, doesn’t mean that the carrying capacity of the debris flow sharply decreases. I suggest you to remove this part.
Lines 427-438: please specify that from your sample you can only quantify the small fraction of the deposit. For the same raison, it is not correct to say that the 94th percentile of the grain size distribution in 0.018 m: look at the big rocks in the photos.
Are you sure that you have used this value for equation (8)? In the following section you use much bigger values.Figure 9: In Figure 9a you use “deposition fan”, in the caption you use “accumulation fan”. Please choose one of them.
Section 4.2.4
I’m sorry but I’m not sure about the need of this section. The changes you claim to see in Figure 10 are really small, especially from 7:44 to 08:04, maybe less than 1 decibel. I believe that this can be within the errors associated with your computations. If you want to investigate the variation of frequency over time, maybe you could compute the frequency peak or the mean frequency as in Farin et al. (2018). However, it should be done continuously and not only for the 6 time intervals as you do.
Similarly, the application and interpretation of the model by Lai et al. (2018) is vague to me. How have you got D=0.5-0.6 m if the 94th percentile of your sample is just 0.018 m? Why have you chosen velocities ranging between 2 and 6 m/s? Have you taken these values from the literature? This choice looks weird also because later you do estimate the velocity, so why haven’t you tested your estimation? Maybe it would be interesting to test the model by Lai et al. (2018) with the velocity you estimate, and see what is the diameter that gives you a seismic power similar to observations. However, some issues still remain as it is not clear to me how you can estimate the debris flow length and the seismic parameters in equation (8). I believe that the uncertainties on these parameters are too high to interpret the result of the modelling.
The interpretation of your results looks also incorrect to me. In equation (8), both diameter and velocity are to the third power, therefore you cannot say that particle size has a minor effect than velocity. Your interpretation is due to the fact that the velocities you have tested vary much more than the diameters (you triple the velocity from 2 to 6 m/s, but the diameter only varies by a factor of 1.2 between 0.5 and 0.6 m) (lines 479-484). Moreover, at line 502 you say that “the particle content in one of the factors affecting the energy of seismic signals”, but the model by Lai et al. (2018) makes the opposite assumption of constant particle content. If you believe that particle content plays a role, why have you used this model?
For all these raisons, I would remove this section, unless you can solve the points I have raised.
Section 4.2.5
Line 518: what do you mean by “horizontal distances”?
Line 519: Why have you chosen a gain factor of 1.8? Is it the sigma of equation (10)? Are the spectrograms resulting from the seismic signal restoration? If not, I don’t see how the restoration is useful here given that all your comments concern the spectrograms.
Lines 522-537: it is really hard to see what you observe in the figures. I propose to show the different times with lines in the spectrograms.
Figure 11: more comments are needed. As I have already said, why in Figure 11b and 11d the seismic power is zero below 10 Hz? Is the signal filtered or a different characteristic of the instruments compared to the ones in the Ergou gully? In both cases, this is an important aspect to clarify since it affects all the interpretation about the frequency bands. Again, there are no comments about the passage of the debris flow front: you assume that the first peak in seismic power is related to the passage of the debris flow, but it could also be the front approaching towards the seismic station. Finally, are you sure that in Figure 11f we see the seismic signature of a debris flow? It could be rainfall in my opinion (Rindraharisaona et al., 2022).
Lines 550-552: the two sentences repeat the same concept
Line 556: I cannot see the decay towards 23 Hz. Can you explain this comment?
Section 4.3: as I said before, I’m not convinced about the method you have used. However, I will make some comments on the text.
Lines 569-571: please try to be more clear. The decay of seismic power is not a problem by itself. You should say that you need to take into account the distance between the sensors and the debris flow if you want to estimate the relative magnitude of the events.
Line 575: what are these values? m/s? You should also mark the peaks in the figures (and mention the figures where you can see these values). Which station have you considered? This must be clarified
Lines 576-578: how do you compute the frequency bands? How can you use the frequency bands to get the magnitude of the debris flows? The larger the frequency band, the bigger? Who says that?
Section 4.4
This section is interesting, but several aspects must be clarified and you should convince me about the parameters you have used
Line 580: in my opinion you haven’t estimated the maximum velocity but rather a mean velocity of the debris flow, as you consider all the signal
Lines 583: even if velocities are shown in the table, you need to recall them in the text. What do you mean by “normal”? Give some references
Lines 583-586: from these lines it is not clear to me if you are talking about the cross-correlation method. Comments are needed also for the other debris flows
Line 586: three decimal digits seem too many to me given the uncertainties.
Figure 12: you should show all the three debris flows. It is not clear to me if the curves you show come from the cross-correlation routine or are just the amplitudes
Line 595: it is a bit strong to say that you use the Manning formula to verify the velocity calculations, because also the Manning formula has its own uncertainties. I would say that you have estimated the velocity with two independent methods
Line 599: please be more clear about the estimation of the roughness coefficient. Is it true that you have estimated it with equation (6)? What value of d50 have you used?
Lines 604-607: these sentences should appear in the methods. Please tell us more precisely how you have computed the hydraulic radius and the slope, as they are crucial terms in the Manning formula
Line 611: comments are needed as it means that between the two stations the velocity is quite constant
Lines 611-612: as already said, it is not correct to say that one method verifies the other
Figure 13: I’m confused by this figure, since in Figure 2 you show the infrared camera only at station 1. How many infrared cameras do you have on the Fotangbgou? If two, why have you used only one camera in the previous sections? If one, how have you estimated the flow stage at station 2? Please clarify this aspect
Discussion
As a general comment, a better job should be done in this section. The discussion is the place to compare your findings with existing works and you have done it only in section 5.2. If you follow my main suggestion, section 5.3 is good to discuss on the limitations of this monitoring system and on what could be improved, taking advance of the monitoring stations already existing around the world. Comparisons with other works are also needed on the values of velocity.
Section 5.1
I would remove this section, since you are not adding discussion points but only repeating your findings. Moreover, several parts are not clear to me: at line 633 I don’t see how the kinematic parameters vary with topography; I don’t understand how the distance between the sensor and the channel can affect kinematic parameters (maybe you wanted to say that the distance must be taken into account if one wants to use seismic sensors to estimate kinematic parameters?); at line 639-640, the meaning of “seismic features select representative analysis points” is obscure to me.
Conclusions
The conclusions should be adapted with the respect to the new structure of the manuscript
Line 708: you say “large difference”, but at lines 629-630 you seem to say the opposite
References
Coviello, V., Arattano, M., Comiti, F.,Macconi, P., & Marchi, L. (2019).Seismic characterization of debrisflows: Insights into energy radiationand implications for warning. Journal of Geophysical Research: Earth Surface,124, 1440–1463. https://doi.org/10.1029/2018JF004683
Farin, M., Mangeney, A., de Rosny, J.,Toussaint, R., & Trinh, P.-T. (2018).Link between the dynamics of gran-ular flows and the generated seismicsignal: Insights from laboratory experi-ments.Journal of Geophysical Research:Earth Surface,123, 1407–1429.https://doi.org/10.1029/2017JF004296
Farin, M., Tsai, V. C., Lamb, M. P., & Allstadt, K. E. (2019). A physical model of the high-frequency seismic signal generated by debris flows. Earth surface processes and landforms, 44(13), 2529–2543. https://doi.org/10.1002/esp.4677
Rickenmann, D. Empirical Relationships for Debris Flows. Natural Hazards, 19, 47–77 (1999). https://doi.org/10.1023/A:1008064220727
Rindraharisaona, E. J., Réchou, A., Fontaine, F. R., Barruol, G., Stamenoff, P., Boudevillain, B., et al. (2022). Seismic signature of rain and wind inferred from seismic data. Earth and Space Science, 9, e2022EA002328. https://doi.org/10.1029/2022EA002328
Zhang, Z., Walter, F., McArdell, B. W., Haas, T., Wenner, M., Chmiel, M., & He, S. (2021). Analyzing bulk flow characteristics of debris flows using their high frequency seismic signature. Journal of Geophysical Research: Solid Earth, 126(12). https://doi.org/10.1029/2021JB022755
Citation: https://doi.org/10.5194/egusphere-2023-2015-RC2 -
AC2: 'Reply on RC2', Yan Yan, 26 Jan 2024
Thank you for spending the time to review and assess our manuscript. After carefully analyzing the reviewers’ comments, we were deeply convinced that we had not been able to accurately summarize the innovations and research objectives of our current study in the previous manuscripts, and we carefully analyzed the content of our Methodology and research. We therefore determined the research purpose in this manuscript: a theoretical basis and a case study exemplar for the real-time monitoring, analyzing the debris flow by a debris flow monitoring system based on the core of seismic monitoring, the determination of early warning thresholds and hazard assessment and analysis. We have rewritten the abstract to emphasize the technical line of the study and the results obtained, highlighting the strengths of the study and eliminating “the large gap between the aims” as pointed out by the reviewer. We have rewritten the abstract, highlighted the technical line of the study and the results obtained highlights the strengths of the study and eliminates “the large gap between the aims”, as pointed out by the reviewer.
Regarding the reviewer’s suggestion of “the lack of accuracy in most of the methods and analyses shown herein”, we have made a targeted revision in this round of revision, and we have sorted out and refined the contents of section 4.1~4.3, rewritten the manuscript to avoid ambiguity, and corrected some of the errors, e.g., Fig.8a incorrectly labelled point A and C. The modification has improved the accuracy of analyses. We have made an overall introduction to the methodology and necessary modifications to highlight the feasibility of our research methodology.Please see the attachment.
-
AC2: 'Reply on RC2', Yan Yan, 26 Jan 2024
- AC3: 'Comment on egusphere-2023-2015', Yan Yan, 26 Jan 2024
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 449 | 241 | 38 | 728 | 29 | 31 |
- HTML: 449
- PDF: 241
- XML: 38
- Total: 728
- BibTeX: 29
- EndNote: 31
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1