I found re-reviewing this paper to be frustrating because I felt like I was doing the same review again, I had many of the same comments as the first time I read the paper. I came away with the impression that the revisions were minimal and rushed and that our reviews were not carefully taken into consideration. A few of my comments were skimmed over or left out of the response document entirely. Since this was a re-review, I also spent more time assessing the writing, and I found the technical writing quality to be lacking. The authors use imprecise language quite frequently, some sentences are so vaguely worded as to be nearly meaningless, and there are grammatical issues. I still don’t think this paper is publishable in its current state. I do think there are interesting observations here but only if the authors are interested in improving the analysis and communication to a level that is required for it to be a valuable and reproducible contribution to the scientific literature.
Major comments:
-As I said before, the authors still give no information on where the data used in this study are located and how they can be accessed. This study is not reproducible without that, and most journals require that.
-Also as I said before, the authors still give almost no detail on how they did some of their analyses, so again this is not a reproducible study. The peak spectral frequency in particular has no details still despite both me and the other reviewer asking for that. The authors just say they compute a frequency spectrum and take the peak, but there are many ways of estimating a frequency spectrum with various ways of smoothing as well that might give different peaks. Did they just compute the FFT with no smoothing and take the highest value, or did they use welch’s method, multitaper or some other approach that attempts to reduce the noise and stabilize the peaks? Same with directionality, they need to say exactly how the amplitudes used in those ratios are determined, did they just take the ratio of RMS amplitude for each component for each 10 s window or did they average ratios of envelopes from Hilbert transform? Was any filtering applied or is this all done with the full broadband data (hopefully they filtered because full broadband data can have a lot of very low frequency noise, especially in the horizontals)? No information at all is given on how the sediment concentration data was collected and very little on how velocities were estimated. All the details of the methods that someone would need to reproduce the study need to be stated explicitly in the text. Ideally they would have a methods section that details these things all in one place before the results rather than sprinkling these details throughout the text.
-The abstract and introduction in particular contain imprecise and sometimes exaggerated language, but there is imprecise or casual language throughout. I made some line by line comments below but I didn’t have time and it’s not my responsibility to do this in great detail so I’d suggest the authors do a very thorough re-reading to improve the quality of the scientific writing in general.
-The “evolution of lahar signals” section on L313-L437 is still extremely speculative and also very hard to follow, my head was spinning trying to digest it. The authors did not respond to this comment that I also made last time and did not improve this section. The authors are, in my opinion, often overinterpreting the meaning of very subtle features of the seismic signals by invoking hypotheses from different past studies that are, in themselves, sometimes speculations. I think this paper would be much stronger if the authors focused on making their own interpretations that are supported by the other non-seismic data they have that tell them what the flow was doing over time and space for this exact event rather than speculating wildly on the meaning of seismic attributes alone based on findings from other studies of other flows elsewhere. The authors say there are videos, flow monitors, stage height measurements, pore-pressure sensors etc. at 21 monitoring locations. No detail was given as to where these instruments were, but I presume they know how the flow behavior was evolving quite well over time at at least some of the seismic station locations. The authors should compare the time of those changes directly to the changes seen in the seismic data and use that for making interpretations just on this event. And then, those findings could be compared to some of these other studies to see if they are, in fact, also finding the same thing. That would provide a lot more scientific value than trying to overinterpret based only on what those past studies found. The comparison of directionality with sediment concentration is a great example of that, and I think it’s one of the more interesting parts of the paper. More of that is needed and it seems like they should have the data to do that.
-Along the same lines, in the same section, L313-437, the authors still do not give any physical explanations for their numerous speculations like “a low PSF could represent the supercharged stream flow pulse” or “the ~10 Hz PSF may be explained by flow processes” to name two of many speculations made in this section. The authors responded to my comment on this last time by saying that they cite papers that describe the physical reason the seismic signals change, and that they describe these physical reasons through the manuscript many times. I am not satisfied with their response. First of all, the authors should not require their readers to read all the citations to understand the physical reasons why different relationships between seismic signal properties and flow properties might occur, they should paraphrase that relevant information in this text. And secondly, they do not actually explain the physical reasons anywhere in this section as they state in their response. Perhaps they didn’t understand what I meant so I will try to clarify by giving just one example: on L339, the authors say “The reason the DR decreases during phase 1 for the 2007 lahar could be due to the parallel component being more sensitive to flow processes than bedload forces”. But they don’t say why this might be the case. That’s what I’m asking for. So in this case, for the DR to be variable, it requires a change in the polarization and/or the type of seismic waves and/or the direction those waves are coming from, so why could the parallel component of a seismometer be more sensitive to flow processes? To explain the physical reasons behind it (and I'm not saying this is right), one could say something like, for example, that granular fluid flow induces more shear forces on the bed and side channel and that might generate more SH waves that could convert into Love waves, and that might cause higher flow parallel amplitudes than cross-channel amplitudes when the highest amplitudes are coming from the part of the channel closest to the seismometer. Whereas for bedload, the impacts might be more vertical (assuming they just mean particles bouncing along in mainly fluid rather than bedload during a debris flow, which would have more shearing) and thus would generate more Rayleigh waves that might have a stronger cross-channel component if the signal were strongly dominated by energy coming from right next to the sensor and not other adjacent parts of the flow. The tricky thing with the elongated source of debris flows, and especially with stations being so close to the channel though, is that there is likely a lot of energy coming from all parts of the channel near the station and it may not always be dominated by processes happening right next to the station and that is probably muddying your directionality ratios, but I digress. In any case, explanations for why the seismic wavefield is altered due to these various explanations that are invoked is what I’m asking for. Give some logical explanations for why the invoked relationships between flow characteristics and various observed features of the seismic wavefield might make sense physically. What would be better is less speculation though, as described in my previous comment.
-While the authors did add an H/V analysis to investigate site effects, which I appreciate, they stuck it at the very end as an afterthought. It should have been when they introduced the frequency analysis in order to justify that their approach is valid before they start making interpretations. They also give conflicting information on how they did the analysis in the supplement. They said they used noise in the text and in the supplement text, but the label on the figure says they used the signal of the lahars.
-The authors added a new frequency estimation method, spectral centroid frequency, to address the concerns that myself and the other reviewer both had about the stability of the peak spectral frequency. I appreciate that, and I do appreciate that some details are given about how its computed, but it’s also added late in the paper and only as a way to support the use of the other spectral metric. If this is more stable, why not just replace the peak spectral frequency with this one? Or at least put it on the same plot as the other frequency metric, it didn’t seem to match up as well as the text leads one to think, the changes seem more subtle, but it’s hard to tell without being able to actually compare the two more easily.
Line by line:
L13-14: This sentence doesn’t make sense, the “changing landscape” implies long term volcanic processes but the part of the sentence saying it can “drastically transform the properties and dynamics of the flow” implies either changes that happen within the channel during a single given lahar due to erosion and deposition I presume or changes that happen along the course of the flow as it reaches lower slopes and wider channels etc. I’m guessing you mean the latter but I’m not sure. Consider rewriting for clarity.
L15-16: Odd English usage. The saying is “tailored to” not “tailored between” or “tailored at”, so change “between” to “to”, and change “but at” to “but to”.
L17-18: This seems like exaggerated language, I doubt any emergency manager would say the thing they need to understand first is how a lahar transforms over time. Tone down the language, don't use the word utmost, and find a more tempered way to say this is useful for hazard management. You could be more specific by saying how the lahar evolves influences how big it grows and how far it might go and what areas it might inundate or something like that.
L26-27: Used by who? Use active voice. I’d also say how these approaches were used, e.g., something like “We analyzed 3-component seismic amplitudes, frequency content…to investigate how the evolution of the lahar’s behavior is reflected in the seismic wavefield.”
Also say what directionality is very briefly in the abstract, (cross channel amplitudes/parallel amplitudes) since it’s not a standard seismic parameter.
L31-33: It would be incredibly hard to understand what you mean by this without reading the entire paper in detail first. I suggest being more specific about what the pattern is on each of the three stations. Also, why are you not mentioning the sediment concentration part of the study in the abstract? I found that to be one of the most compelling and novel parts of the paper.
L33: Extraordinary promise is exaggerated, I suggest removing the word extraordinary. I also haven’t seen in this paper any specific suggestions of how directionality ratios could actually and practically be used in monitoring so this statement isn’t really reflective of the paper.
L40-41: This sentence and the rest of the paragraph seems specific to debris flows/lahars, but the way this paragraph is written, you’re still referring to all volcanic mass flows. Why not just start this paragraph specifically talking about just lahars and forget about mentioning the PDC’s and debris avalanches?
L42: Larger than what? And what kinds of changes?
L44: A flood can’t “have” a warning. “outburst floods can occur with little or no warning”
L45: Eruption sources of what? “Lahars triggered by eruptions can be anticipated…”
L46: various methods is too vague
L48: How is “the intensity of rain” a technique? Setting alarm thresholds based on the intensity of rain would be a technique.
L52: change forestry to forests
L56: I suggest replacing “predict and investigate” with “characterize”
L57-58: How can this be the first step? People have been doing this for many many decades, and the next paragraph starts off talking about some of that work. This also implies that we don’t know much but we know a lot.
L59: Specify that you are talking about in-situ methods here.
L61-63: References needed.
L66: The paper cited here as showing that seismometers can be “capable warning systems” isn’t even about an actual warning system.
L66: I would disagree that they can “accurately” estimate flow properties, seismometers are quite inaccurate compared to other methods (depending on what flow properties & flow dynamics, it’d help if you were more specific about what you mean by flow properties) and prone to many limitations that aren’t really discussed here.
L68-70: This sentence is in direct conflict with the previous one that said geophysical instruments could already provide accurate estimates.
L71: Using all three components isn’t a “technique.” Rephrase to say something like “Using techniques that use information from all three components…” or similar
L72: Again, using all three components isn’t a technique in itself and can’t characterize anything by itself. Using all three components to do what?
L77-79: Provide a brief explanation here about what directionality analysis is, because it’s not a common term. The next sentence sort of implies what it is, but just say it directly. Also provide a bit of explanation for how it could provide the information listed. i.e., why would the polarization of the seismic waves be sensitive to changes in those things.
L86: Change “entrapped” to “entrained”
L88-89: As I said before, please give a few words explaining what you mean by laminar or plug-like. It isn’t enough that they are in the references already, paraphrase here. I am specifically asking for this because later in the text what you talk about as plug-like or laminar isn’t what I had in mind so you need to be sure everyone knows exactly what you mean.
L93-107: I don’t see why it’s relevant that Cronin et al had three models at different distances, the differences between those three models aren’t described here in any meaningful way. Just stick to the four phases. Also, it would be helpful to include here what part of this sequence you are referring to as the lahar “head”. That term is used later in the text and I don’t know what part you mean by that exactly or what characteristics it has.
L122: change “entering” to “to enter”
L125: Are there some missing words here? How can landslides be “along with” a channel? Perhaps you mean landslides contributed additional sediment to the channel?
L131: change “thus an” to “thus had”
L132: How can a debris flow be filled? Suggest changing “sediment-filled” to “sediment-rich”
L133: I flagged this sentence last time and it still makes no grammatical sense. Please fix. Also, provide detailed information on how velocities were estimated before presenting velocity results.
L141: Awkward transition, the way that it’s worded sounds like you’re talking about another lake-breakout than in the previous paragraph. Suggest changing “properties of a” to “properties of this” and maybe other changes to make the transition smoother.
L153: change “sampling” to “sampling rate”
L154-155: This is not accurately described. To rotate to flow parallel requires both the North and the East component, you can’t rotate just the North component to get flow parallel as this phrasing implies. I suggest just saying you rotated them to align with the flow parallel (P) and cross channel directions (T).
L155-156: This is an awkward way to say the seismometers were installed to be level. Broadbands have to be installed level anyway, not sure why this sentence is necessary or why you can’t just say the sensors were leveled.
L164: Arrival times of just the flow front or of different phases that passed? Also, discuss the temporal accuracy, especially of the eyewitness reports which are notoriously not accurately timed.
L165: Provide details on how flow velocities were estimated from imagery and some more info about the flow meter that was installed at TRAN. Do you have continuous flow velocity estimates at TRAN? If so, why aren’t these used in the paper? Also, this information should be before flow velocities are reported.
L167: Insert methods section here.
L171: Delete the word amplitude after PSF, you are not reporting the amplitudes of the frequency in this study as far as I can tell, just the frequency.
L172-174: This kind of thing should be labeled on the figure and in the caption, not necessary in the main text.
Figure 2: Give time stamps consistent with your figure time axes for the photos, especially for b,c,d,e. Possibly also label them on the figures. What is a “low PSF beginning of a lahar body”? Use more precise language.
L183: Spectra should be changed to spectrum if you’re talking about just one. Also, more detail should be given as there are many ways to get a frequency spectrum. I’m guessing you’re talking about computing the raw FFT and taking the absolute value but I shouldn’t have to guess.
Figure 3-5: Label the panels with letters and then use those to refer to them in the text instead of referring to them by the color of the dots. I’d also suggest labeling the arrival of all the lahar features you talk about in the text that you know about from the videos etc. on these figures. It would greatly help people follow what you are talking about.
L203: Define what you mean by lahar head here or earlier in the text.
L205: Mention upward sweep of vertical frequencies?
L224: What is the difference between the lahar front mentioned here and the lahar head mentioned earlier? Be sure all the terms you use are very clearly defined early in the text.
L232-234: Here or earlier such as in the introduction, talk about why directionality could potentially contain useful information, physically. i.e., why would some directions have stronger energy than others as the flow behavior progressed? What is changing that is changing the seismic wavefield? L244-246 sort of starts to scratch the surface, but that should be earlier in the text and also you need to get into why, physically, would the directionality of seismic waves be influenced by those flow properties.
L239: You should, however, discuss some caveats of this, including the fact that you will have energy coming from a large section of the flow that is more altered by attenuation, not just the portion right next to the station.
L240-241: Give details on how exactly you estimated the amplitudes for this, ideally in a methods section earlier (amplitude method, filtering, duration of windows etc.). I can piece together clues sprinkled through the text to guess what you did, but again, I shouldn’t have to guess. Just explicitly say the methods up front in enough detail that someone could do exactly the same thing you did.
L243: The paper cited here was not about a warning system so saying DR was used for warning purposes is misleading.
L253: TRAN did not behave similarly to RTMT as this sentence implies, TRAN started low and stayed low until 10 min. RTMT started around 1 and barely dropped before increasing again. Consider rephrasing.
L271: A frequency can’t describe lahar dynamics. Consider rephrasing maybe something like to understand “if changes in PSF are related to changes in lahar dynamics”?
L272: What is meant by frequency constraints is not clear. I assume you mean the stability of the frequency peaks needs to be analyzed? Consider rewording.
L274-275: Normalizing by what? By the peak amplitude? Total energy?
Figure 7: This plot needs to also have the context of the flow arrival times on it and the other frequency estimates. It’s also hard to compare to the other plots when it’s by itself like this.
L296-297: These features need to be labeled on the plot and/or times of these features need to be given, ideally both. I have no idea when these things are happening on those plots.
L300: What is “this” that explains the bimodal distribution? I’m guessing you’re implying that the peak spectral frequency is jumping for TRAN because there is energy in two bands and they are similar in amplitude so which one is higher and thus serves as the PSF varies slightly over time. Be more explicit, again I shouldn’t have to guess at what you mean.
L326: Do you know from your videos or other data that the supercharged stream pulse is passing at this point? I feel like with the amount of data available for this study, speculation about what is happening at various times during the signal is not needed and instead the authors should rely on what they know from observations directly from their non-seismic data instead of speculation based on other studies. I could make this comment several times in this section, but I won’t point them all out as this whole section needs a reworking based on my main comment above.
L331-332: I don’t understand the logic of this statement, or why a low frequency zone suggests lahar elongation. Please clarify.
L333: The term “flow processes” could mean pretty much anything. What else is there?
L334: I don’t understand what this means, a flow is discharge, how can it be more sensitive to discharge at one time than another? Do you mean the seismic signal is more sensitive?
L338: Why is phase 1 not labeled on the plots? Also, do you have confirmation from your other data when phase 1 is passing? If so, is it the same time period as when DR is behaving this way? See main comments.
L342-343: Do you know this is the case from your data from this event?
L350-351: What else is there besides turbulence or sediment transport? This doesn’t seem very meaningful. Also, why would one expect the frequencies to be the same between different monitoring locations? Frequency is very dependent on distance from the source, material properties, among other factors.
L354: Bulking up into a full 4-phase lahar doesn’t make sense to me. The way the 4-phase lahar is described earlier is an evolution over time and distance, not four phases that happen all at once as this implies. “Bulking up” seems to me like something that happens during that process, in phase 2 and 3, not all at once. Do you mean it has reached phase 3 at this site?
L367: Same comment as above. Do you mean it reached phase 3 or 4 at this point in space and time?
L373: This is the first time relative amplitudes of vertical vs. horizontal have been mentioned. If this is something of interest in this study, it should be presented in the results section. I also feel that this statement is made as if it is always true, I’m sure there are other reasons that vertical amplitudes could be higher than horizontal. It also probably varies with frequency.
L375: Bimodal pattern of what? I assume you mean of the peak frequency? Say so explicitly.
L387-389: The wetted perimeter was not examined or even discussed thus far in this study, so how is the reader supposed to know that an increased DR accompanied an increased wetted perimeter for this lahar? Where is this data? The wetted perimeter is not shown on Figure 6.
L394-395: Here is one instance where a reason for a pattern is given, this is good, but is incomplete because it doesn’t say why lateral excitation relates to the DR.
L397: I would disagree that this figure shows a “good” correlation. It looks very scattered. Did you compute the Pearson’s correlation coefficient?
L398: What timeframe? Also, Figure 2f looks like laminar fluid flow from what I can see, I don’t see plug-like flow. To me, plug-like flow involves a very high sediment concentration and lower pore pressures and is like a big plug of sediment that is being kind of bulldozed along by a wetter flow behind it. Plug-like flow seems like it would have a lot of particle collisions, but this sentence implies it has fewer.
L403-404: Same comment as for L397.
L414: I don’t know what you mean by “traditional” here and elsewhere. Be more specific. Do you mean it reached phase 3 of the 4-phase model you are invoking? Do you know that these descriptions are the case at that time from video or other data or is this based just on the seismic interpretation?
L443-444: The only actual measurable properties that were estimated and presented in this study came from non-seismic instruments (velocity and sediment concentration), is that what you are referring to here?
L446-448: Wasn’t this shown by Procter et al. and other references already?
Figure 9: Should the x axis be labeled as time since this is showing the lahar passing by fixed points? Also, why is PSF not also L for the bow wave in panel C as it is for the other two? Also, say what the values ranges of PSF and DR are for D, I, H, L, and M.
L472: what is considered “when the lahar arrives”? The lahar head? SLF/HF? TF?
L477: If new flows are going to be used in this paper, you need to provide more background information about them.
Figure 10: Show when each lahar has passed/ended because in the text you refer to times when the lahar has passed and talk about things that happen for the “entirety of the event” but do not say what time that includes.
L485: What observations? No references were given for the statement this is referring to.
L493: Can you be more explicit about what you mean by little evidence? I do see elevated seismic amplitudes at these times, they just aren’t as high as at later times.
L505-507: The amplitude differences weren’t really examined in this paper, and certainly weren’t compared to any data on bedload. This is also stated as fact and I’m not sure it’s established scientific fact, even if there is a reference.
L511-515: These sentences are all way too vague.
L519: I don’t see the logic behind why a flow overtopping the channel would reduce the amplitudes.
L522: I don’t recall hearing about the flow breaking out of the channel at COLL previously, are we supposed to know about this already?
L525: Awkward description, please describe more succinctly.
L539: RTMT and COLL do have distinct H/V peaks, this is not an accurate statement. The smoothing level you used in your H/V analysis also controls how narrow the peaks can be.
L560-561: This sentence implies you know how the flow evolved with distance only from the seismic sensors, but you had several previously published papers on these flows and other data as well. This is misleading.
L568-569: I didn’t see anything in the paper that explained how a DR could be used in a warning system explicitly. Some discussion is needed for what exactly they would be used for in the warning system and how, and what additional analysis is needed to make that possible.
L695: Information is missing from this reference.
Supplement:
Figures S1-S3 Label times of different arrivals on these plots.
Kommo should be Konno.
What was the bandwidth of the Konno and Ohmachi smoothing?
In H/V analysis section, say how much data was used. Was it the entire day?
Frequency is misspelled in Figure S4 labels.
Figure S4 says the H/V results are for the lahar signal, but the text says ambient noise from the previous day was used. |
