Dynamics of the Askja caldera July 2014 landslide, Iceland, from seismic signal analysis: precursor, motion and aftermath

Schöpa, Anne; Chao, Wei-An; Lipovsky, Bradley P.; Hovius, Niels; White, Robert S.; Green, Robert G.; Turowski, Jens M.

doi:https://doi.org/10.5194/esurf-6-467-2018

Articles | Volume 6, issue 2

https://doi.org/10.5194/esurf-6-467-2018

© Author(s) 2018. This work is distributed under
the Creative Commons Attribution 4.0 License.

Special issue:

From process to signal – advancing environmental...

https://doi.org/10.5194/esurf-6-467-2018

© Author(s) 2018. This work is distributed under
the Creative Commons Attribution 4.0 License.

Articles | Volume 6, issue 2

Research article

|

14 Jun 2018

Research article |

| 14 Jun 2018

Dynamics of the Askja caldera July 2014 landslide, Iceland, from seismic signal analysis: precursor, motion and aftermath

Anne Schöpa, Wei-An Chao, Bradley P. Lipovsky, Niels Hovius, Robert S. White, Robert G. Green, and Jens M. Turowski

Download

Final revised paper (published on 14 Jun 2018)
Supplement to the final revised paper
Preprint (discussion started on 07 Dec 2017)

Interactive discussion

Status: closed

AC: Author comment | RC: Referee comment | SC: Short comment | EC: Editor comment

- Printer-friendly version

- Supplement

RC1: 'Review', Jacqueline Caplan-Auerbach, 05 Jan 2018
RC2: 'comments from reviewer', Anonymous Referee #2, 09 Jan 2018
AC1: '"Final Response"', Anne Schöpa, 16 Feb 2018
EC1: 'Associate Editor comments', Kate Allstadt, 16 Feb 2018
- AC2: 'Reply to associate editor comments', Anne Schöpa, 15 Mar 2018

Peer-review completion

AR: Author's response | RR: Referee report | ED: Editor decision

AR by Anne Schöpa on behalf of the Authors (16 Feb 2018) Author's response

ED: Referee Nomination & Report Request started (03 Apr 2018) by Kate Allstadt

RR by Anonymous Referee #2 (09 Apr 2018)

Suggestions for revision or reasons for rejection

Thank you very much for the modification. I have two major comments for your revised manuscript.

Section 3.2.2
I think it would be easier to understand if you can write in the following manner (which is how you computed):
1) present inverted force(Assuming a block model with a constant landslide mass over time, inverted force can be expressed as the product of the mass and time-series acceleration.)
2) explain how to obtain mass mass (Then, we find the most likely mass by…)
3) show the most probable mass value
4) velocity and displacement obtained from the force and estimated mass

P14 L18-22
“Although individual events are roughly discernible in seismograms, the effects of attenuation and superposition of multiple sources makes time domain analysis difficult (Fig. S7). Following previous studies, we therefore prefer spectral analysis 20 over the time domain (Dmitrieva et al., 2013; Hotovec at al., 2013; Winberry et al., 2013; Lipovsky and Dunham, 2016). We note that the tremor observed by Yamada et al. (2016) was observed at a much shorter source-to-station distance <1 km, whereas our closest station is 3.5 km from the landslide source area.”

I think detectability of discrete events and the transmission of energy in the long-distance are different. For example, the seismic station in Yamada et al. (2016) was <1km, but the stick-slip event was not recorded at any other stations (R~20km), whereas the signal was detectable even R~30km for the Askja event. This implies that the stronger attenuation or smaller amplitude in the example of Yamada et al. (2016), compared to this article. In the above sentences, it seems that the reason why the individual event was not detectable was due to the source-to-station distance. However, I think it is not true, since Poli (2017) showed clear individual events for R=30km, but your data did not even R~3.5km.

Hide

RR by Jacqueline Caplan-Auerbach (23 Apr 2018)

Suggestions for revision or reasons for rejection

This review is my second of this paper, and I appreciate the response that the authors gave to my first review. Many of my original questions have been well addressed in this version of the paper. I still have some concerns as well, and these (as well as areas in which my questions have been satisfied) are discussed below.
Previous concerns:
1. I wondered if the frequencies were actually harmonic: In this draft the authors describe specific frequencies (2.5, 5, 7.5 Hz), but I still find this very difficult to see. The frequency bands shown in the spectrogram are very narrow and it’s hard to see if this is truly the frequency of the signal, or if it results from the length of the Fourier transform. Either way, I don’t see that there are clear signals at 5 or 7.5 Hz. It is also very difficult for me to see how the red/blue lines drawn on the spectrogram correspond to the actual signal…I don’t see that the lines correlate with actual frequencies in the spectrogram. Finally, the authors model upgliding and downgliding, but they do not comment on the feature that sweeps from 12 Hz to 14, and back down again. I don’t think that all signals necessarily need to be modeled, but this one stands out in all of the data but goes unnoticed or unexamined.

2. My previous review asked the authors to clarify when the landslide begins—this is now clear.

3. In my original review I expressed doubt that the slide could happen to have two asperities, both of which failed with sufficient regularity to generate identically repeating events, but one of which showed acceleration while the other showed deceleration. In this version the authors retain that model for one patch but assume the other changes by growing in size. While that is also plausible, it still feels like too great an assumption. I note that reviewer #2 in the first draft was concerned that assumptions are quickly taken to be truth, and I feel that this is a good description of this model—it still feels implausible.

4. While the authors have done a good job making the force history more clear (I appreciate the figures showing these forces), I agree with the associate editor that it’s concerning that the force history still has the landslide source at the top of the slide rather than the center of mass.

5. The authors added some important references, for which I thank them.

6. In their response to the associate editor’s comments, the authors state that the supplementary Figure 7, “shows that individual events are in fact discernable, but that they are rather jumbled in appearance and difficult to interpret”. This puzzles me, because I don’t see any discrete events within the seismogram. I’m not sure what they are referring to as individual events, but if they feels such events are existent, they should at least be pointed out. I also agree with one of the other commenters (either reviewer #2 or the associate editor) that it doesn’t make sense to have discrete events that are too small to be seen individually, yet that merge together to be recorded across the network (an argument made in the revision). It is far more compelling to consider that there never were discrete events before they were recorded as tremor.

I do think it’s reasonable to interpret tremor as being overlapping events—that’s certainly a commonly invoked mechanism. But I’m not sure there is sufficient evidence here to suggest that this is the best explanation for this particular tremor.

In summary, some things were well corrected, and I very much appreciate the care that the authors took in responding to reviewer concerns. However, I’m still unconvinced that the tremor signal comprises discrete events. I do find the signal fascinating and worthy of commentary, however.

Hide

ED: Publish subject to minor revisions (review by editor) (23 Apr 2018) by Kate Allstadt

AR by Anne Schöpa on behalf of the Authors (03 May 2018) Author's response Manuscript

ED: Publish subject to technical corrections (10 May 2018) by Kate Allstadt

The authors have done a thorough job of addressing the reviewers comments and I believe the manuscript is ready for publication after a few minor corrections:
-On page 8, the authors use the word "arrest" improperly, the word that should be used for a landslide coming to a stop is "rest" -e.g., "...before coming to rest."
-Also on page 8, the authors make a contradictory statement that the material going into the lake was much smaller than the total mass as justification for not considering the trajectory as continuing into the water as far as it was observed, but then in the next paragraph give a reported number for the volume of the submerged mass that was as high as a third of the total mass - that is a substantial portion of the mass. This contradiction needs to be addressed.
-Also on page 8, the authors describe the process for choosing the correct mass that gives the most-likely trajectory, but do not give a description of the criteria that was used to determine which trajectory of they many they computed was "best".
-on page 9, line 19, "as it is the case" should be changed to "as is the case" and on line 20, "which that" should be changed to "which"
-page 12, line 19 - do the authors mean "high frequency" here? And also, please give a reason why a force history inversion cannot be done for high frequency signals.
-Section 6.2 - this section is a great addition to the paper, but the authors discuss just tremor in general. Please be clear here and throughout that gliding tremor/gliding spectral lines are also commonly observed in volcanic settings and in relation to some of the models discussed. And also, on pg 15 line 22, consider whether the change in recurrence interval of discrete events is the cause of gliding spectral lines in all possible tremor models, not just those related to stick-slip.
-page 15 line 10, change "montion have" to "motion has".
-Section 6.3 - this discussion about a seismically-based early warning system for landslides, which requires identification of a pre-existing potential landslide source, will make readers question why then not use a warning system that provide a more direct observation of the landslide deformation, such as high-sample rate GPS on the slide mass. Be sure to stress what advantage the proposed approach might have.
-page 17 line 20 - why do you propose slow deformation "should produce a seismic signal" - don't the conditions have to be just right for stick-slip sliding to occur, as discussed in great detail in earlier sections of the paper? There are many landslides for which repeating precursory events were not observed despite relatively close seismic stations.

Hide

ED: Publish subject to technical corrections (11 May 2018) by Tom Coulthard (Editor)

AR by Anne Schöpa on behalf of the Authors (17 May 2018) Author's response Manuscript

Download

Article (18706 KB)
Full-text XML

Short summary

On 21 July 2014, a voluminous landslide entered the caldera lake at Askja, Iceland, and created tsunami waves inundating famous tourist spots. The high hazard potential of the site motivated our study in which we analysed seismic data and found a precursory tremor signal intensifying in the 30 min before the landslide. Our paper shows the potential of seismic monitoring techniques to detect precursory activity before a big landslide that could be used for an early-warning system.