the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Characterizing the evolution of mass flow properties and dynamics through analysis of seismic signals: Insights from the 18 March 2007 Mt. Ruapehu lake-breakout lahar
- 1Department of Earth Sciences, Simon Fraser University, Burnaby, British Columbia, Canada
- 2Department of Earth Sciences, University of Durham, Durham, DH1 3LE, UK
- 3Volcanic Risk Solutions, Institute of Agriculture and Environment, Massey University, Palmerston North, New Zealand
- 1Department of Earth Sciences, Simon Fraser University, Burnaby, British Columbia, Canada
- 2Department of Earth Sciences, University of Durham, Durham, DH1 3LE, UK
- 3Volcanic Risk Solutions, Institute of Agriculture and Environment, Massey University, Palmerston North, New Zealand
Abstract. Monitoring for lahars on volcanoes can be challenging due to the ever-changing landscape which can drastically transform the properties and dynamics of the flow. These changes to the flows require the need for detection strategies and risk assessment that are tailored not only between different volcanoes, but at different distances along flow paths as well. Being able to understand how a flow event may transform in time and space along the channel is of utmost importance for hazard management. While visual observations and simple measuring devices in the past have shown how lahars transform along the flow path, these same features for the most part have not been described using seismological methods. On 18 March 2007 Mt. Ruapehu produced the biggest lahar in New Zealand in over 100 years. At 23:18 UTC the tephra dam holding the Crater Lake water back collapsed causing 1.3 x 106 m3 of water to flow out and rush down the Whangaehu channel. We describe here the seismic signature of a lake-breakout lahar over the course of 85 km along the Whangaehu river system using three 3-component broadband seismometers installed < 10 m from the channel at 7.4, 28, and 83 km from the crater lake source. Examination of 3-componennt seismic amplitudes, peak frequency content, and directionality combined with video imagery and sediment concentration data were used. The seismic data shows the evolution of the lahar as it transformed from a highly turbulent out-burst flood (high peak frequency throughout), to a fully bulked up multi-phase hyperconcentrated flow (varying frequency patterns depending on the lahar phase) to a slurry flow (bedload dominant). Estimated directionality ratios show the elongation of the lahar with distance down channel, where each recording station shows a similar pattern, but for differing lengths of time. Furthermore, using directionality ratios shows extraordinary promise for lahar monitoring and detection systems where streamflow is present in the channel.
Braden Walsh et al.
Status: final response (author comments only)
-
RC1: 'Comment on nhess-2022-31', Emma Surinach, 04 Mar 2022
The authors of the paper use seismic signal analysis from 3-component sensors placed at 3 different sites along the lahar channel generated by the Mt. Ruapehu lahar of March 18, 2007 in combination with earlier field observations to draw conclusions about the evolution of the mass and the properties of the flow and its dynamics. They use parameters and methodology already used in other contributions, to which they refer.
The paper needs a major revision.
I have carefully read the manuscript up to Line 308 in the Conclusions.
At that point I confirm myself that to draw valid conclusions the authors have to take into account the amplitude of the frequency peaks. Although different values of frequency peaks appear in the signal, their contribution to the signal should be clarified. The amplitude of the frequency peaks is independent of the signal amplitude. This gives information on the frequency content of the signal.
As an example, in Vázquez et al., (2016) that deals with the seismic characterisation of lahars at Volcán de Colima, Mexico, of 10 Hz the contribution to the seismic signals is superior to the others. It is up to the authors to use this representation or that of Roig-Lafon (2022, https://www.tdx.cat/handle/10803/672460#page=1) or another (e.g. size of the dots). The question is to evaluate the contribution of each frequency to draw accurate conclusions.
The inclusion of this approach will give value to this contribution and will differentiate it from the other works mentioned in it that deal with the same matter.
In addition, authors must include a data processing section/subsection as well as an explanation (e.g., equations, references...) of how the data results are obtained.
To facilitate the interpretation of the results, the authors must include a table with the values and characteristics observed in the figures, as well as indicate in the figures the different parts described in them. Then, in the discussion, refer to them.
In the following selected details that I consider will improve the manuscript, once corrected.
Line 68. Seismic instruments have been used since 1998 or earlier. I do not consider it relatively young (e.g., Sabot et al., 1998)
Line 69. Specify. There are more geophysical instruments. You are only referring to those relating to vibrations.
Line 83. You must include the purpose of this contribution because the previous ones also deal with this topic. Additionally, you can include directly without splitting subsection 1.1.
Line 130. Homogenize or clarify the terms channel, river... according to the figure, also throughout the text
Lines 144-145 As I understand it, the data you analyze also comes from measurements on the ground. Change the sentence.
Data
Line 155. In the instrument setup, is the vertical component of the instrument orthogonal to the slope or in the zenith direction? In our studies we have observed that the results are different. Part of the seismic energy orthogonal to the slope goes to the other components. Indicate, at least, the angle of the slope where the sensors are installed.
Line 162. Are other instruments co-located with the seismometer? How are the averaged speeds of the lahar determined? Also explain the procedure to fixe the arrival times of the lahar.
Results
The results are values after a data processing.
The description of the figures/results must be explained taking into account the amplitude of the frequency peaks and differentiate between those that are significant and those that are not. In the discussion the authors have to take this into account.
Line 193. In some cases, the event passes over the station, this is not the case. The lahar passes closer to the station. In fact, it is the record at the station of the waves generated by the passage of the lahar. In the figure the different parts are differentiated: head, forehead...
3.2 Directionality
Line 236. Are you considering amplitude of the signal or the energy? The energy of a wave depends on its amplitude and frequency.
Line 250. Note that it is shorter in RTMT (2 min) than in TRAN (3 min)
Line 251. Notice that there is a bend in both curves of the same wavelength but shifted in time. Note the similar concave behavior at RTMM 17-35 min. and TRAN 15-30 min.
Line 254. Describe properly minute 7.
Moreover
Complete the reference Lube et al (2012).
For known and general descriptions, cite previous articles. Avoid the most recent ones unless they add something new.
Emma Suriñach
-
AC1: 'Reply on RC1', Braden Walsh, 03 Jun 2022
Addressing Reviewer Comments: Reviewer in bold text and response in regular text.
Major comments:
At that point I confirm myself that to draw valid conclusions the authors have to take into account the amplitude of the frequency peaks At that point I confirm myself that to draw valid conclusions the authors have to take into account the amplitude of the frequency peaks
While the amplitude of each frequency range of the signal will yield differing content and the contribution amount will be different, the results and especially the discussion section of this manuscript are still valid. The use of PSF or dominant frequencies has been used by numerous studies on a variety of mass flows (of which many are cited in the manuscript) to estimate the dynamics. The “contribution” of the PSF or difference between the spectral centroid yields the same results and is also an indicator of dynamics based on frequency (i.e. when PSF is bimodal/”spread” is large (contribution small), multiple processes are occurring). That said we added a section in the discussion looking at the normalized spectrograms as well as the centroidal frequency to show when and where the PSF is most dominant.
Authors must include data processing section/subsection as well as an explanation of how the data results are obtained.
All of this information is already in the manuscript. Most is in the “Data” section as well as Lines 164-169 in the “Results” section. As for equations, we believe it is not necessary to include basic equations for converting to frequency domain or directionality (simple division). For the new SCF analysis we included equations, see new section 4.1.
To facilitate the interpretation of the results the authors must include a table with the values and characteristics observed in the figures as well as indicate in the figures the different parts described in them.
We describe and include multiple images from the lahar in figure 2 as well as created a model in figure 8 referencing the seismic data with the observations. Furthermore, throughout the discussion we reference each figure multiple times while explaining the results. See lines 367-377, 386-388 for example.
Line by line comments:
Line 68: seismic instruments have been used since 1998, I do not consider it relatively young. Changed to “However, in order to fully utilize these instruments, improved interpretation, comprehension, assessment, and universality is needed.“
Line 69: specify. There are more geophysical instruments, you are only referring to those relating to vibrations. See change for Line 68
Line 83: you must include the purpose of this contribution because the previous ones also deal with this topic. The inclusion of section 1.1 is to give the audience an introduction on the details of the parts of the conceptual lahar model on Mt. Ruapehu. Which then transitions into section 1.2 which talks about the 2007 event. The first paragraph in the manuscript is a brief introduction on mass flows and why it is important to monitor them.
Line 130: Homogenize or clarify the terms channel, river according to the figure also throughout text. We homogenized the wording throughout.
Line 144: change the sentence. Changed to “supplementary measurements”
Line 155: in the instrument setup is the vertical component of the instrument orthogonal to the slope or zenith direction. In our studies we have observed that the results are different. Indicate, at least the angle of the slope where the sensors are installed. Added “Furthermore, the seismometers were installed normal to horizontal to lessen the degree of vertical energy transfer to the horizontal components.”
Line 162: Are other instruments co-located with the seismometer? How are the average speeds determined. Also explain the arrival times of the lahar. Added “Arrival times are based off of images and eye witnesses at each of the monitoring stations. The flow velocity at RTMT and COLL were estimated from imagery and at TRAN from a flow meter.”
Line 193: in some cases, the event passes over the station, this is not the case. The lahar passes closer to the station, in fact it is the record at the station of the waves generated by the passage of the lahar. Deleted “passing the station”
Line 236: Are you considering amplitude of the signal or the energy. Changed to “amplitude”
Line 250: note that it is shorter in RTMT than in TRAN. Noted in text
Line 251: notice that there is a bend in both curves of the same wavelength but shifted in time. Note the similar concave behavior at RTMT and TRAN. Further analysis has shown no correlation with this “curve” in DR at RTMT and TRAN.
Line 254: Describe properly minute 7. Minute seven is described on lines 252-253
Line 568: fix citation for Lube et al., 2012: Fixed citation
-
AC1: 'Reply on RC1', Braden Walsh, 03 Jun 2022
-
RC2: 'Comment on nhess-2022-31', Anonymous Referee #2, 28 Apr 2022
This observational study investigates along-channel seismic signals of a lahar at Ruapehu caused by an outbreak event in 2007. The study focuses on using all three components of the data to investigate changes in peak frequency and directional energy in the high frequency energy and makes some interpretations about what the observed patterns mean about flow style and how it evolves as the flow progresses down the channel. The main takeaway the authors want to get across seems to be encouraging the use of all three components of seismic data instead of just the vertical component. The authors also present a conceptual model for what changes over time in flow style they think are responsible for their observations at each of the three stations.
My overall assessment is that the study is definitely an interesting additional datapoint to add to the lahar seismology literature and the directionality patterns and frequency changes over time are compelling. However, I found the discussion and interpretation to be quite speculative and many of the observations have been made in other papers at other sites already so this paper is not necessarily groundbreaking but does add more information to the literature. I don’t think the study makes good use of the other non-seismic data that was collected and their study could be strengthened and made less speculative by doing so (e.g. they don’t link their ideas about changes in flow behavior to the time-lapse photos, and on line 138 they mention many data types that aren’t used in the study but seem like they would be extremely useful for understanding changes in flow style). The part of the paper that does use some other non-seismic data, Figure 7b, is barely discussed in the paper despite being one of the most interesting observations. That alone could certainly be emphasized more. The conceptual model is definitely credible and reasonable, though not especially strongly or uniquely supported by the observations so it’s more of a hypothesis/discussion point.
I do have one major concern with the basic analysis though that should be addressed before the paper goes further, and that is the lack of any investigation into the site amplification due to the subsurface structure below each monitoring site that may be influencing the peak frequencies of the seismic data. I suspect the stations likely do all have strong site amplification at specific frequencies because the sites are probably installed on past lahar or alluvial deposits that overlie some older higher velocity materials which would give a good impedence contrast. Since the time window prior to the arrival of the lahar has the same peak frequency as the flow itself for many of the cases, I suspect that it is very possible that the peak frequency might be controlled by the site effects at each station more than the source. This may also explain why we see some jumping between two discrete frequencies, I suspect those sites where that happens may have two peaks in the spectral response due to multiple layers of contrasting sediment under the site. Differences in spectral peaks between the two horizontal components can indicate 2D/3D subsurface structure so it’s also possible any observed differences between the components are not due to the flow itself. Site effects and their influence on the observed peak frequencies needs to be investigated and accounted for before any detailed interpretations about frequency content can be made. For example, the authors could use H/V analysis (see Molnar et al 2022 https://doi.org/10.1007/s10950-021-10062-9 for a recent review), spectral ratio analysis using a nearby reference site on bedrock if there is one, and/or an analysis of the noise patterns at the sites over longer periods of time.
Along the same line, I also have concerns about the robustness of the frequency peaks. No information is given on how those are computed. Were they picked off a raw FFT (which is often quite noisy and unstable in exact peak locations)? Or was smoothing done prior to picking the peaks? Was any analysis of the robustness of the peaks and their stability done? Some of the temporal patterns, especially at COLL, are very subtle and I am not sure they are statistically significant, at least as presented. More detail needs to be added. Some more robust ways of estimating the frequency content to consider would be perhaps using the multitaper method to compute the spectra on which the peaks are picked, or using the dominant frequency (e.g., see Douma and Sneider 2006, https://doi.org/10.1111/j.1365-246X.2005.02807.x), or another weighted or smoothed method of estimating a more robust spectral peak. Another option to look at would be instantaneous frequency, which can be obtained by the Hilbert transform.
Also, especially in the discussion section but also earlier, I feel like the paper is lacking an explanation for why, physically, the different flow styles would be reflected differently in the directionality and frequency peaks. There are many speculations, like, to give one example “The reason DR decreases…could be due to the parallel component being more sensitive to flow processes than bedload forces” but I want to know why, physically, would the parallel component be more sensitive to flow processes than bedload forces? Is it a difference in the wave types they generate? The forcing directions? There are many places where I want to know the physical reasons behind why a given pattern is being invoked as an explanation for flow characteristics. Why would “underflow” cause lower frequencies? Why would a low frequency occur when a bow wave is passing? Why would the angle of the slope cause higher frequencies? Why would the vertical component be higher when bedload dominates? And so on…
Other comments:
I don’t see any information on how the data used in the study can be accessed, this is necessary for a reproducible study.
L61-63, a lot of these things are true of seismic instruments as well, and there is more ambiguity in interpretation for quantitative values. I also disagree that they can be used for “accurate” (L66) estimates of flow properties. Only in very limited situations is that true.
L69-70, using seismometers for flow monitoring is not young, it’s been used operationally since the 1990’s and there are even earlier papers that use seismometers for mass movements.
L77 previously not recorded by who? People have used three component sensors for recording lahars and other mass movements many times in the past. Maybe rephrase as “information about the flow that is not recorded if only vertical-channel sensors are used”
L88-89, perhaps it would be useful to explain what you mean by terms like plug-like and laminar here and elsewhere in the paper, many debris flows are still very agitated if not turbulent, is that still called plug-like by your definition?
L133 missing “the” and missing comma
L133-134 Please explain how velocities stated here and elsewhere in the paper were measured
L157 Same as above, and also please explain what the averaging represents, is it time averaging over the entire flow? Spatial averaging?
L166-167 Please give detail as to how the peak spectral frequency was estimated (and I assume the authors didn’t mean peak spectral frequency amplitude but just peak spectral frequency since they don’t present spectral amplitudes anywhere, so remove the word amplitude if that is true).
L169 Please explain how the arrival time is known, is this from the cameras? It doesn’t seem to be from the seismic records based on later figures and discussion.
L180-181 How do you know this is the arrival of the head for the lahar? Is the definition of the “head” that time when the peak amplitude occurs or is it determined independently from the cameras or flow height sensors? I also don’t know why the word streamflow is in parentheses here or what that is meant to convey.
Figure 4: it’s interesting that there is an upward sweep of frequency on the vertical component, but it jumps discretely on the horizontals. Any idea why?
L232: rephrase this sentence, such as “installed so that the North component of the sensor is instead aligned to be parallel to the flow”
L234: but site effects cannot be ignored, see my main comment
L236: give details on how the energy was computed, over what frequency etc. and how directionality is computed for reproducibility
L237-241: Add some information here about physical reasons why the directionality would contain information about rheology changes and other flow parameters.
L265-266: I don’t understand what multiple pulses has to do with bulking material that is different from collecting material from erosion, do the authors mean more pulses of material bring more material down and the flow fronts coalesce into one bigger one? Perhaps this sentence could be rephrased?
L284: “The ~10 Hz PSF may be explained by flow processes” is a very vague statement, can you be more specific?
L305: Here or wherever it was first introduced, can you explain what is meant by a 4-phase lahar? I don’t think that is common enough to not define.
L364-365: here and elsewhere, are the speculations about flow style corroborated by the camera images and/or other data types that was collected?
L405: Unclear what is meant by “at different distances away from source” – this study did not investigate changes with distance from the source so how does that come into play? Maybe this statement needs some clarification?
L435: This statement and the supporting evidence is one worth emphasizing more in the paper.
L444-446 and Figure 9, this discussion makes me wonder what the directionality looks like for long time periods of noise and earthquakes and whatever other signals the station records when a lahar is not passing by. Is the change in directionality unique to when a lahar is passing by? Could it be differentiated from other seismic sources?
L464-466 tilt is usually at much lower frequencies than was the focus of the investigation here. Since no details were given on how the energy was computed for each component (on raw data? Filtered? Etc.) it’s hard to assess whether this would have a substantial influence or not without more information.
-
AC2: 'Reply on RC2', Braden Walsh, 03 Jun 2022
Addressing Reviewer Comments: Reviewer in bold text and response in regular text.
Major comments:
Site amplification:
The time windows prior to the lahar arrival has similar PSF due to the fact that there was already streamflow in the channel and it has been shown in the past that streamflow has higher PSFs. The March lahar was sourced from a lake breakout and hence would show similar PSF to streamflow compared to a more traditional bulky debris flow. To your concern with site amplification, we went ahead and estimated H/V ratios for each station. This analysis of where the H/V frequency peaks are has been included in the revised manuscript.
Frequency analysis:
See response to first major comment from Reviewer 1. We calculated centroidal frequencies and normalized spectrograms to compare to the PSF results. More details on the actual data processing were added as well.
Physical reasons for signal properties
We cite many papers in which data came from natural and laboratory sources that describe the physical reasons why the seismic signals change. On this point, we describe these throughout the manuscript many times and how they relate to our signals. We have gone back through the manuscript and rewrote some areas to better clarify these statements.
Line by line comments:
Line 61-63: a lot of these things are true of seismic instruments as well, and there is more ambiguity in interpretation for quantitative values. I also disagree that they can be used for “accurate” (L66) estimates of flow properties. Only in very limited situations is that true. Changed to “and/or lack the capability to evaluate multiple pulses or flow events”
Line 69-70: Using seismometers for flow monitoring is not young: Changed to “However, in order to fully utilize these instruments, improved interpretation, comprehension, assessment, and universality is needed.“
Line 77: previously not recorded by who? People have used three components many times in the past: changed to “yield additional information about the flow that is not utilized if only the vertical component is used”
Line 88-89: Perhaps it would be useful to explain what you mean by terms like plug-like and laminar: These are explained in the references, as well when describing the lahar at COLL
L133: missing “the” and missing comma. Added
L133-134: please explain how velocities are measured. Added how the velocities were estimated. See replies to comment from Reviewer 1.
Line 157: explain what the averaging represents. Added “The flow velocity at RTMT and COLL were estimated from imagery and at TRAN from a flow meter.”
Line 166-167: give details as to how the PSF was estimated. Figures 3-5 show the peak spectral amplitude at its represented frequency. Line 166-167 describe how the points were estimated. To add value to this please see the new supplementary figures in which normalized spectrograms for each station and component are displayed.
Line 169: please explain how the arrival time is known. Added “Arrival times are based off of images and eye witnesses at each of the monitoring stations.” On line 162.
Line 180-181: how do you know this is the arrival of the head? Why is the word streamflow in parentheses? Line 180 states “prior to the arrival of the head (peak seismic amplitude).” The streamflow was meant to state streamflow is in the channel. We see the confusion. Deleted “(streamflow)”
Figure 4: it’s interesting that there is an upward sweep of frequency on the vertical component, any idea why? There is a sweep in all the components, but the vertical shows the most consistent, probably due to the better coupling with the ground compared to the horizontal directions and what controls them.
Line 232: rephrase this sentence: changed to “so that the North component is aligned to be parallel to the flow”.
Line 234: site effects cannot be ignored. See comment to major comment.
Line 236: give details on how the energy was computed and directionality. Changed to “The directionality ratio (DR) can be defined as the cross-channel amplitude divided by the flow parallel amplitude.”
Line 237-241: add information here about physical reasons why the directionality would contain information about rheology changes. Added information about differences in signal between streamflow and lahar.
Line 265-266: I don’t understand what multiple pulses has to do with bulking material that is differing from collecting material from erosion, maybe rephrase sentence. Changed to “or through the coalescing of multiple pulses to shorten the total length of the lahar”
Line 284: this is a vague statement, can you be more specific? The explanation is stated following this statement on lines 285-288.
Line 305: can you explain what is meant by a 4-phase lahar. Added “(see section 1.1)”
Line 364-365: here and elsewhere, are the speculations about flow style corroborated by the camera images or other data types that were collected. See figure 2 and refences to figure 2 throughout the manuscript. Also figure 7 for data relating to COLL. There were also eyewitnesses to confirm the camera images.
Line 405: unclear what is meant by “at different distance away from source”. Source is the starting point of the flow event. Changed to “the mass flow source”
Line 435: This statement and the supporting evidence is one worth emphasizing more in the paper. We added some lines throughout to emprise this.
Line 444-446: is the change in directionality unique to when a lahar is passing by? Could it be differentiated from other seismic sources? This is an interesting idea and needs to be looked into in the future, but is outside the realm of this study, but would be a great future study to do. That said, the channel in question in this manuscript always contains streamflow, so the DR is always at “streamflow” levels when only recording “background” noise. Furthermore, DR of background noise in a “dry” channel was recorded at Colima, Mexico and was always high and indistinguishable from lahar DR, see Walsh et al., 2020.
Line 464-466: tilt is usually at much lower frequencies. Since no details were given on how the energy was computed for each component it is hard to assess whether this would have an influence or not. Details are given in the Data and Results sections. Also see the citation given that lead to the statement of tilt. Tilt has also been used as a detection method for mass flows in the past as well.
-
AC2: 'Reply on RC2', Braden Walsh, 03 Jun 2022
Braden Walsh et al.
Braden Walsh et al.
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 285 | 107 | 20 | 412 | 9 | 10 |
- HTML: 285
- PDF: 107
- XML: 20
- Total: 412
- BibTeX: 9
- EndNote: 10
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1