Modelling seismic ground motion and its uncertainty in different tectonic contexts: challenges and application to the 2020 European Seismic Hazard Model (ESHM20)

Graeme Weatherill, Sreeram Reddy Kotha, Laurentiu Danciu, Susana Vilanova, and Fabrice Cotton

Download

Final revised paper (published on 23 May 2024)
Preprint (discussion started on 04 Sep 2023)

Interactive discussion

Status: closed

RC1:
'Comment on nhess-2023-124', Anonymous Referee #1, 16 Oct 2023
This paper presents elements of the 2020 European Seismic Hazard model. In particular, the paper covers a discussion of special regions where additional data and refinements could be made, it discusses additional elements for regionalization of attenuation, and the adaptation of GMM for subduction and deep seismicity. The paper is very informative and well written. I do not have any major technical comment. Most of my comments are provided below to increase clarity or to correct some editorial mistakes. Congratulations to the authors on a very interesting and well written paper.

Lines 55-60. This entire paragraph is poorly worded.
It is not clear what is meant by saying that “Probabilistic seismic hazard curves yield a wide array of products…, including … uniform hazard spectra that describe the shape of a spectrum of constant probability models of building number, typology and value.”

I believe that the authors are trying to describe the process of integrating hazard with fragility and vulnerability to obtain risk. This is not at all easy to gather from the description.

Figure 3. I recommend adding to the caption an indication of how many models are plotted in each subplot (i.e., how many logic-tree branches there are). I initially reverted back to Figure 1 to try to infer that number; only upon further reading I realized that the reference should be Figure 4.

Figure 4. I believe that some of the signs in front of the epsilon values should be negative.

Line 387. Not clear what is mean by “the parametric craton”

With regards to the fuzzy inference system to classify subduction events: has this system been verified or validated elsewhere? Some explanation on what is a fuzzy classification system may be warranted. For example, most readers will not understand how a classification is defuzzified (Line 917).

Editorial:
Line 111: “feedbacks” should be in singular. A similar use of feedback as plural also occurs in line 522.
Line 118: “its rationale” instead of “is rationale”
Line 191: “principal” should be “principle”
Line 312: A closing parenthesis is missing
Line 501: a closing parenthesis is missing
Line 781: a word seems to be missing (i.e., “treat Iceland separately from …”?)
Line 853-854: a verb seems to be missing (should it be “that are inconsistently reported”?)
Line 997: something seems to be missing. Not clear what is “produced lower than the original”
Line 1417: minority is misspelled.
Citation: https://doi.org/10.5194/nhess-2023-124-RC1
- AC1:
  'Reply on RC1', Graeme Weatherill, 02 Jan 2024
  Reply to Referee Comment #1
  We thank the reviewer for their positive comments on the paper and for their thorough review of the manuscript. A revised manuscript with be submitted shortly in which we address their main comments and have made all the editorial changes needed to fix the small errors and typos that the reviewer has identified.
  Our responses to the main comments provided by the reviewer are as follows:
  “Lines 55-60. This entire paragraph is poorly worded.
  It is not clear what is meant by “Probabilistic seismic hazard curves yield a wide array of products…, including … uniform hazard spectra that describe the shape of a spectrum of constant probability models of building number, typology and value.”
  
  I believe that the authors are trying to describe the process of integrating hazard with fragility and vulnerability to obtain risk. This is not at all easy to gather from the description.”
  
  REPLY: We thank the reviewer for their feedback on this paragraph and acknowledge that the wording was not optimal in some places. This paragraph is re-worded in the revision to the manuscript.
  “Figure 3. I recommend adding to the caption an indication of how many models are plotted in each subplot (i.e., how many logic-tree branches there are). I initially reverted back to Figure 1 to try to infer that number; only upon further reading I realized that the reference should be Figure 4.”
  
  REPLY: Clarification of the number of branches being plotted in the figures has now been added to the caption.
  “Figure 4. I believe that some of the signs in front of the epsilon values should be negative.”
  
  REPLY: We thank the reviewer for spotting this error. The negative epsilon values have been corrected.
  “Line 387. Not clear what is mean by “the parametric craton””
  
  REPLY: This has now been fixed to read “the parametric GMM”, which is correct for the context of the sentence.
  “With regards to the fuzzy inference system to classify subduction events: has this system been verified or validated elsewhere? Some explanation on what is a fuzzy classification system may be warranted. For example, most readers will not understand how a classification is defuzzified (Line 917).”
  
  REPLY: The reviewer’s suggestion for further clarification regarding the fuzzy inference system is well taken and additional sentences of explanation have now been added. We have applied this system in other regions of the world and with each application have always inspected either all of the event classifications or a sample of the classifications (depending on the catalogue size) visually to verify their classification based on their location within the subduction system. Validation of the fuzzy classification system against independently labelled data sets such as those of the NGA Subduction ground motion database, or processed ground motion databases for other countries such as those for Japan or New Zealand, is more challenging owing to a variety of different factors. Among these are 1) usage of different classifications for tectonic regions within the subduction system (not all databases include outer rise or upper mantle wedge), 2) different definitions of the subduction interface not all of which are publicly available (we use Slab 2.0 for this purpose, but other interface models may give different results). We have added a short note the question of validation but acknowledge that further validation studies would be beneficial if opting to apply this approach (or similar) in the future.
  
  All other editorial comments have been corrected in the revised manuscript, to be uploaded in due course.
  
  Citation: https://doi.org/10.5194/nhess-2023-124-AC1
RC2:
'Comment on nhess-2023-124', Laura Peruzza, 27 Nov 2023

The paper by Weatherill and coauthors introduces new concepts and practices for identifying the appropriate ground motion model in PSHA, by using a single backbone model to which regionalized adjustments or scaling factors are applied. The calibration is applied to the 2020 European Seismic Hazard Model, by resorting to some database of observations. My congrats to the author for the relevant work, that I hope will drive the future hazard assessments toward a more careful use of the existing GMMs.
I have only a few minor comments, and some typos/editorial corrections to suggest.

The first comment refers to the definition of shallow crustal seismicity described in chapter 2. I did not find in the text a clear definition of what shallow crustal seismicity means; probably one simple sentence can clarify it for the reader, even without resorting to the figure 13, that I agree with the authors is well located where it is now in the text. I also suggest simplifying the titles and subtitles of chapter 2, for example in “2. The regionalized backbone ground motion model for shallow crustal seismicity” “2.1 Background and motivation” “2.2 Development of GMM logic tree for shallow crustal seismicity (non-craton)”.
The second comment is the graphical representation of Fig. 4 and Fig. 18, where it is not easy to grasp the differences in the two branch sets, consider to put the references out of the box, as they are in common, and enter the red label in the box, as they refer to given model parameters that differ in each branch set.
Finally, I suggest to mention in the development for the future chapter, around line 1401, the existence of other data processing, like for example in the NESS2-eBASCO flatfile, that avoiding to filter the low frequency content, may result in relevant changes (increase of expected values) on the derived GMM.

Here the other small annotations for typos/corrections.
Lines 55-57: sentence not clear, consider rephrasing
L94: ESRM20 instead of the second ESHM20
L138, 152 and throughout the text: check the spelling of Pezeschk, Kuehn
L191: his instead of their
L204: the branches are 27, correct?
L222, on : Mh is not defined, check the uniform use of sigma and tau symbols to indicate the standard deviation, in this and in the following subchapter
L245: c3 in caption is dc3,r? the same in the legend of the figure
L272, L502: space missing in regionR, ofKotha
L308: in the caption magnitude 7.5 instead of 7.0
L358 and throughout the text: check the date or missing ref for Goulet et al. (2021?)
L412, L501: close parenthesis after the symbol
L452: missing verb
L456: avoid the acronym SRTM , as never used later
L459: dimension for slope not correct
L466: Wald & Allen reference missing in the list
L486, L829: defined instead of defines
L531: Grunthal et al 2012 can be cited here for the EMEC catalogue
L546, L669, L809: add comma in the reference
L651: set instead of sot
L717: check the date or missing reference
L839: missing subject after Cypriot?
L849: add the label “subduction interface” in the figure
L854: are instead of a
L866: add c to tectoni
L872: in the caption, add RIF, RIS meaning
L873: is not the magnitude always given in the flatfile?
L928: figure 15 instead of 14
L942: applies and comparing?
L944-5 and throughout the text: check the dates in Hassani (2021?) and Si (2022?)
L1008: figure 15 instead of 14
L1025-6, 1033: check the closure of parenthesis, and missing spaces
L1034, 1046: figure 17 instead of 16
L1060: add dot, commas in the references
L1062 and throughout the text: check the date in Manea (2022?)
L1075: not clear sentence
L1101: in caption figure 7 instead of 6
L1106: in caption 200 km instead of 2000
L1187: missing comma, b,c in Zhao
L1269: Kotha & Traversa not in reference list
L1293: is ESHM13 instead of 20?
L1389: Campbell 2017 not in reference list
L1506: wrong spelling of INGV
Duplicated references at lines 1617-1619; 1689-1692
Not cited references at lines 1628, 1674, 1678

Citation: https://doi.org/10.5194/nhess-2023-124-RC2
- AC2:
  'Reply on RC2', Graeme Weatherill, 02 Jan 2024
  We thank Dr Peruzza for her positive feedback and thorough review of the manuscript. A revised manuscript with be submitted shortly in which we address their main comments and have made all the editorial changes needed to fix the small errors and typos that Dr Peruzza has identified.
  Our responses to the main comments are provided as follows:
  “The first comment refers to the definition of shallow crustal seismicity described in chapter 2. I did not find in the text a clear definition of what shallow crustal seismicity means; probably one simple sentence can clarify it for the reader, even without resorting to the figure 13, that I agree with the authors is well located where it is now in the text. I also suggest simplifying the titles and subtitles of chapter 2, for example in “2. The regionalized backbone ground motion model for shallow crustal seismicity” “2.1 Background and motivation” “2.2 Development of GMM logic tree for shallow crustal seismicity (non-craton)”.”
  
  REPLY: We thank the reviewer for indicating that we had omitted the definition of shallow crustal seismicity, and for their suggestions for the sub-section titles of chapter 2. A clear definition of shallow seismicity is now added to the beginning of section 2.2, and the sub-sections titles have been shortened accordingly.
  “The second comment is the graphical representation of Fig. 4 and Fig. 18, where it is not easy to grasp the differences in the two branch sets, consider to put the references out of the box, as they are in common, and enter the red label in the box, as they refer to given model parameters that differ in each branch set.”
  
  REPLY: Figures 4 and 18 have now been revised in line with the suggestions from the reviewer.
  “Finally, I suggest to mention in the development for the future chapter, around line 1401, the existence of other data processing, like for example in the NESS2-eBASCO flatfile, that avoiding to filter the low frequency content, may result in relevant changes (increase of expected values) on the derived GMM.”
  
  REPLY: We thank the reviewer for highlighting the NESS2-eBASCO data set, which will likely prove to be useful in refining near-fault and long-period motion in future European GMMs. Reference to this has now been added in the “Future Developments” in Section 5.
  All other editorial issues identified by the reviewer have been resolved in the revised manuscript, which will be submitted shortly.
  
  Citation: https://doi.org/10.5194/nhess-2023-124-AC2

Peer review completion

AR: Author's response | RR: Referee report | ED: Editor decision | EF: Editorial file upload

ED: Publish as is (21 Feb 2024) by Pierre-Yves Bard

ED: Publish subject to minor revisions (review by editor) (26 Feb 2024) by Paolo Tarolli (Executive editor)

AR by Graeme Weatherill on behalf of the Authors (27 Feb 2024) Author's response Author's tracked changes Manuscript

ED: Publish as is (03 Mar 2024) by Paolo Tarolli

ED: Publish as is (03 Mar 2024) by Paolo Tarolli (Executive editor)

AR by Graeme Weatherill on behalf of the Authors (05 Mar 2024) Manuscript

Short summary

The ground motion models (GMMs) selected for the 2020 European Seismic Hazard Model (ESHM20) and their uncertainties require adaptation to different tectonic environments. Using insights from new data, local experts and developments in the scientific literature, we further calibrate the ESHM20 GMM logic tree to capture previously unmodelled regional variation. We also propose a new scaled-backbone logic tree for application to Europe's subduction zones and the Vrancea deep seismic source.