Reviewer comment

Reply

Basili et al. (the manuscript) report on the construction of EFSM20, a fault-source model of continental scale that was used as a basis for the European Seismic Hazard Map 2020 and is available to users who desire to develop seismic hazard assessments, geodynamic models or post-earthquake analyses. The manuscript fulfills to some extent the role of “user manual” for a pre-existing product and is therefore “frozen” by its features. Since the product in question is of great utility for the community and has unquestionable quality, the publication of the manuscript is warranted, and the review will focus mostly on the quality of the communication.

We thank Joao Fonseca for his comments and for positively considering our manuscript's particular (or possibly unusual) character.

The authors state that EFSM20 is a compilation of published data which were blended, harmonized and augmented with recent results. Blending and augmenting are self-explanatory concepts, but harmonization can mean different things and since it implies changes to the original data all the steps and criteria taken to harmonize different datasets must be made explicit and described in detail. The manuscript addresses harmonization in section 2 (Methods). It describes data prioritization criteria (line 139), mentions the harmonization of overlaps (line 147) and re-sampling (line 150), but the semantic dimension of the data harmonization was not described in a way that I could identify. Then in section 3 (Results) what is presented as harmonization is the inference of variability to single-valued data points using the properties of the data collection (see next paragraph). This can hardly be described as harmonization in my view.

We understand the reviewer’s concerns. There is no commonly used definition of harmonization in scientific writing. We indeed borrowed the term from other domains. In particular, we adopted this term with the meaning of “working on complementary compartments to have them play together to achieve an overall strategic objective”. The original datasets were not specifically created for our intended use; therefore, we needed to perform several uncodified operations. For the benefit of the readers, we can introduce this term the first time it is mentioned and then rearrange some of the statements where the term could be confused for something specific.

The treatment of data variability was obscure and seemed non-standard. The authors assign a “dip variation ratio” and a “slip-rate variation ratio” to individual faults dividing the range (maximum value minus minimum value) by the average value of the fault collection (line 185 and line 194), but it is not clear what this ratio means. More worryingly, the authors inform that when only a single value of dip is available for an individual fault this ratio is used to “extrapolate [infer?] the dip uncertainty range” around the single-value dip angle. In line 362 the authors state that “the dip harmonization procedure used 660 dip-slip faults to determine an average dip variation ratio of 0.21” and this value “was then applied to incorporate [augment?] the range of dip variability in the remaining 195 dip-slip faults”. It is hard to understand why the dip (or slip-rate) uncertainty of an individual fault should reflect an average value of the collection to which it belongs, unless very heavy and hard-to-justify assumptions are made to model these variables as ergodic stochastic processes. No references are given for the statistical treatment of these parameters, so it can be concluded that the approach was “home-made”. References or a more robust justification are required.

The procedure adopted at this stage is definitely non-standard because we faced a non-standard problem. However, common knowledge and experience from many geological works carried out over the years can justify the assumptions here.

The total uncertainty on dip angle and slip rate depends on two main factors: the natural variability (aleatory), which is partly specific to individual faults and partly common to the tectonic process - especially the faulting type, and the uncertainty due to the level of knowledge (epistemic). 

We rely on the following elements: 1) the reported single value represents an average estimated by the dataset’s authors; 2) a large number of faults in different tectonic settings with valid ranges acquired by a multitude of scientists adequately captures the collective epistemic uncertainty in the estimates; 3) the internal variability is generally limited because the amount of variation within a fault system is one of the criteria most geologists use to separate the fault system in different portions. 

Although these criteria may seem to provide a coarse evaluation, the opposite, i.e., relying on the single value, would imply that neither internal variation nor uncertainty is present, which would be a preposterous assumption and would be reflected in artificially smaller uncertainty in the derived parameters such as moment rate, maximum magnitude.

Besides this concern with the way the data variability is treated, the remaining remarks (listed below) have to do with the quality of the text, which is often obscure. It should be expected that a manuscript with 38 co-authors would reach the review process in better shape.

The corresponding author takes care of the manuscript presentation. In this case, RB takes full responsibility for the final quality of the manuscript. 

For the benefit of the readers, we will submit a revised version carefully considering all the suggestions by both reviewers.

Figures 3 and 8 could benefit from larger fonts for improved readability.

The font readability depends on the final dimension of the figure. We accept the suggestion and will make sure with the editorial office that the lettering reads clearly.

João Fonseca

25/10/2023

Additional comments/suggestions:

Line

1 to 10 – author name format is not correct, remove extra commas.

Suggestion accepted.

45 (and elsewhere) – the acronym is OCT, not OTC

Thank you for spotting this typo. We will correct it throughout.

50 – add link to OGC

Suggestion accepted.

55 – the definition of fault source reads more like that of an active fault, and doesn´t derive from the previous comment as implied by the use of “Since”.

To avoid confusion about using the conjunction “since” at the beginning of a sentence, we will modify the text to read “Considering that ...”.

62 – “The combination … increases”: unclear, rephrase.

Suggestion accepted. We will replace the text of the entire sentence with “The combination of the spatial scale in fault mapping and the temporal scale of their recurrent behavior in generating earthquakes makes the geologic fault data more important, relative to other datasets, in forecasting larger magnitude earthquakes.”

72 – “… to gain insights into the good continuity of plate boundaries”:  unclear, rephrase.

We actually wrote: “… to gain insights from the good continuity of plate boundaries”. The term “good continuity” or “good continuation” is the common name of the Gestalt Principle of Perception. This refers to the common practice of exploiting the capability of large-scale features (such as a plate boundary) to aid the identification of gaps and inconsistencies of smaller-scale features (such as our crustal fault sources). Therefore, we do not see the need for changes.

90 – “… providing different map feature realizations”: unclear, rephrase.

We will replace “realizations” with its synonym, in this context, “implementations”.

136 – “… to completing…”: use “to complete…” instead.

Suggestion accepted.

141 – “… was publicly…”: use “were publicly…” instead

Suggestion accepted.

156 – “…values are then …”: use “…values were then …” instead, to harmonize with previous sentence

Suggestion accepted.

168 – “… we removed or modified …”: explain with what criteria

The criteria are introduced in the preceding sentence. We took action when we identified down-dip intersections leading to unrealistic structural configurations. We assume that the average reader has basic knowledge of realistic structural configurations as explained in structural geology textbooks.

176, 177, 200  – “... made by (Johnston, 1994)”: use “made by Johnston (1994)” instead

Thank you for spotting these typos related to a misuse of the citation software.

186 and 193 – A variable that is the ratio of two dips should not be called Dd , which implies dimensions of dip. A better choice would be rd or equivalent. Likewise for slip-rates.

Suggestion accepted.

198 – “… of the crustal faults …”: use “… of a crustal fault …” instead

Suggestion accepted.

199 – “… that a fault can host …”: use “ … that the fault can host …” instead

Suggestion accepted.

210-213 – “In the second step, … and removed”: unclear, explain better and justify.

Suggestion accepted.

We followed a common procedure to identify and remove outliers based on percentiles. Very large Δ𝑀𝑤 can arise from occasionally extreme geometric fault parameters that do not necessarily reflect the characteristics of an entire fault but only a part of it or that produce values outside the range of applicability of the scaling relations.

We modified the text of this sentence as follows. “In the second step, we calculate the distribution of magnitude deviations (Δ𝑀𝑤−=𝑎𝑣𝑔𝑀𝑤−𝑚𝑖𝑛𝑀𝑤) and (Δ𝑀𝑤+=𝑚𝑎𝑥𝑀𝑤−𝑎𝑣𝑔𝑀𝑤) from the average for all faults. A preliminary inspection of these deviations showed that the Δ𝑀𝑤 distributions are strongly skewed; therefore, we considered values above the 95th percentile (corresponding to Δ𝑀𝑤>|0.5|) to be outliers and removed them.” See also the reply to Reviewer #2 on the same issue.

213 – “Then we obtained the distributions ranking”: obscure, explain better.

Here, we refer to the procedure for calculating percentiles. We will merge this statement with the successive one to make it clearer. The new sentence will read as: “In the third step, we obtained the updated distributions of percentile ranks and extracted...”

221-226 – The discussion on seismic efficiency may be replaced by a short statement that a value of one was adopted, and why. Later it may be discussed under possible future improvements.

In reality, we did not adopt a seismic efficiency value equal to 1. In lines 224-266, we stated that “.. the seismic efficiency was not assigned...”. The value of one is an implicit consequence of not assigning that value in the moment rate calculation. For this reason, we then remark that “...it is thus left to the user to choose a value to apply in applications.” (lines 226 and 266). To clarify this statement, we propose replacing the last sentence in this passage with the following: “The information on slip rate values was adopted in bulk from large datasets of regional compilations, and it is therefore obtained using very heterogeneous estimation approaches. In some data sets, slip rate values correspond to the total (seismic and aseismic) slip, whereas fault slip rates of other data sets correspond to only their seismic part. Therefore, the seismic efficiency was not assigned to individual faults in this dataset. This implicitly means that for the moment rate calculations, the seismic efficiency is as if it was equal to 1, and it is thus left to the users to choose a value to apply in their applications.”

245 – “… constant in the rest of the subducted slab…”: use “… constant above the rest of the subducting slab…” instead

We respectfully disagree with this suggestion and are afraid it would lead to a mistake if adopted. “Above” the rest of the subducting slab there is only the upper plate.

249 – “… the 100-150ºC isotherm … “: why the broad range?

This is a common range for the thermal constraints of the seismogenic portion of the subduction interface. Similar values were found in various subduction zones from around the world (e.g., Oleskevich, D. A., Hyndman, R. D., and Wang, K. (1999). The up-dip and down-dip limits to great subduction earthquakes: Thermal and structural models of Cascadia, south Alaska, SW Japan, and Chile. J. Geophys. Res. 104, 14965–14991. doi: 10.1029/1999JB900060).

316 – “… GEM Global Active Faults Database …”: it would be instructive to hear how this database interacts with EFSM13, 20, etc

We cannot add more detail than those explained in the cited paper.

373 – “… the faults largest potential is rare“: obscure, rephrase.

Suggestion accepted. We can replace the text of the entire sentence with “The average maximum magnitude distribution (Figure 3, lower-left panel) is somehow left-skewed, indicating that the crustal faults hosting the largest magnitude potential are very rare.” Alternatively, we could delete it following the suggestion of Reviewer #2.

383 – “The obtained values …”: use “The obtained values of maximum magnitude …” instead.

Suggestion accepted. We replaced the text with “The obtained maximum magnitude values ...”.

392 – “…were not …”: use “…was not …” instead

Suggestion accepted.

410 (Figure 4 caption) – what is the meaning of the colours of the two bands?

We modified the last part of the caption as “...including the uncertainty of the upper and lower seismogenic depths represented by color bands as indicated in Figure 2.”

429  - “… geometric extrusion …”: should it be “… geometric protrusion …” instead?

Geometric extrusion is a canonical term referring to creating a surface from a line. It is also widely used to name the tools provided by software for 3D modeling such as AutoCAD (https://help.autodesk.com/view/ACD/2023/ENU/?guid=GUID-85FD1106-8F10-4EE8-B0FB-99F1E3AEE405) or MOVE (https://www.petex.com/media/2724/move201821_release_notes.pdf) that perform the operation we made.

472 – “… to help to refine …”: use “… to help refine …” instead

Suggestion accepted.

489-491 – this basic definition of slip rate is not needed, and if needed it should be at the first occurrence of the concept.

We agree that a basic definition of slip rate is optional. However, slip rate is a key parameter in several applications, especially earthquake hazard estimates. Therefore, in this passage, we want to briefly recall it and some common facets of its practical determination. We then mention the cases we encountered to introduce the readers to how those components that entered the dataset affect the parameters derived from it, namely, the moment rates. We can slightly modify the paragraphs to reflect this reasoning.

498-518 – The emphasis on seismic efficiency is not justified since it is taken as 1 If anything that choice should be justified. Passages around “Users are cautioned…” and “The slab geometric reconstructions …” are obscure.

We are afraid we have not made clearly enough that we are not taking seismic efficiency equal to 1. As a matter of fact, there is no such parameter in any of the EFSM20 attribute tables (see Appendix A).

In line 224, we stated that “...seismic efficiency was not assigned to individual faults...” which means that the value of 1 is implicitly present. We also added, "..it is left to the user to choose a value to apply in applications.”  We rephrased this statement to make it clearer as indicated to reply for comment on L221-226. The same statement is then repeated for subduction systems. Then, in the discussion, we want to give the readers more context about the implications of such an unavoidable choice.

As regards the sentence about the slab geometric reconstructions, we can modify it to help readers who are less familiar with subduction processes. The new sentence will also include a couple of literature references.

Reviewer comment

Reply

Dear authors,

So sorry this review is so late! Most of my suggestions are minor/local and are in the attached PDF. My only two larger-scale suggestions are:

1. I would combine sections 2 and 3 (data sources, then the whole story about how crustal faults are prepared, then the whole story about how subduction models are prepared).

2. I would emphasize a bit more that the seismic efficiency parameter is conservatively assumed to be 1 for the purposes of this work, especially in the subduction sections.

Great work and hope for publication soon.

-Chris Rollins

We thank Chris Rollins for his comments.

We followed the recommendation of combining sections 2 and 3. We thank the reviewer for this suggestion, which makes the text easier to follow.

We prefer to remind the readers that a value of 1 means no assumption regarding seismic efficiency. We added it in Eq. 5 because it is a quantity that should be addressed, and to emphasize that, we leave it to the user to adopt whichever value best fits their application/usage. To make this statement clearer, we propose to replace the sentence about seismic efficiency with the following: “This implicitly means that for the moment rate calculations, the seismic efficiency is as if it was equal to 1, and it is thus left to the users to choose a value to apply in their applications.”

We also considered all the minor suggestions for text amendments, which significantly improved the manuscript's readability.

Below, we reply to some of the questions/comments provided in the pdf file. The remaining suggestions were all accepted.

L35: There are some who think that spatially continuous source models are more effective! :)

We can replace “the most effective” with “one of the most effective”.

L72: good continuity

We prefer to keep it because “good continuity” or “good continuation” is the common name of the Gestalt Principle of Perception.

L109: I suggest combining this and section 3 into a single "Data and Methods" section, that first lists the data sources (3.1), then discusses the preparation in full (bring the crustal part of 3.2 into 2.1 and the subduction part of 3.2 into 2.2), then 3.3.

We worked on this rearrangement and are ready to submit a revised manuscript that reflects this suggestion.

L129: > 50%

Not necessarily; we considered that a fault could straddle more than two countries, and in that case, the largest portion is considered.

L150: Out of curiosity, was this using linear interpolation or splines or?

We used a 5-km-long moving stick iteratively to redistribute the residue while honoring the two endpoints.

L156: Do you mean to be in order along the fault?

Yes. This is especially useful when using fault traces in software like OpenQuake or tools that assume the dip direction from the node sequence.

L164: Out of curiosity, are all of the fault depths with respect to sea level, so that the subduction zones start at a few km depth?

They are with respect to sea level.

L181: for each fault, you mean?

Yes. We should rephrase the statement and replace it with: “Since some of the original datasets only reported a single value of the dip angle for several faults.”

L212: What was the reasoning for this? What effect does it have?

Very large Δ𝑀𝑤 can arise from occasionally extreme geometric fault parameters that do not necessarily reflect the characteristics of an entire fault but only a part of it or that produce values outside the range of applicability of the scaling relations. We thus followed a common procedure to identify and remove outliers based on percentiles. To seek clarity, we would modify the text of this sentence as follows. “In the second step, we calculate the distribution of magnitude deviations (Δ𝑀𝑤−=𝑎𝑣𝑔𝑀𝑤−𝑚𝑖𝑛𝑀𝑤) and (Δ𝑀𝑤+=𝑚𝑎𝑥𝑀𝑤−𝑎𝑣𝑔𝑀𝑤) from the average for all faults. A preliminary inspection of these deviations showed that the Δ𝑀𝑤 distributions are strongly skewed; therefore, we considered values above the 95th percentile (corresponding to Δ𝑀𝑤>|0.5|) to be outliers and removed them.” See also the reply to Reviewer #1 on the same issue.

Regarding the effects, we recall that among the measures of central tendency, the mean is more sensitive to the existence of outliers than the median or mode. In our case, removing outliers pulled the mean of the entire dataset toward slightly higher values and reduced its standard deviation.

L225: This should be emphasized also in the subduction section.

We propose to repeat the same sentence proposed for the crustal faults.

L361 (Table 1): Would it be possible to add something like this for subduction sources below? e.g. total areas of the four subduction zones, maybe moment rates? If not, I think this table could be put in supplementary or appendices.

Suggestion accepted. We add the number and length of the four subduction sources. In addition, with the suggested reorganization of the text, we move Figure 8 in the “Results” closer to this table, which facilitates the reader to gain an idea of the maximum magnitudes and moment rates of all fault sources (crustal and subduction) in the various tectonic settings. 

L372: 

Suggestion to remove the statement to be considered in light of reviewer #1 comment on the same sentence.

L377: This overlaps with description in section 2; I think it could be moved up to there and combined.

Suggestion accepted.

L387: I think this could be moved up to section 2 and combined with the discussion of this there.

Suggestion accepted. For consistency, we propose moving up the subsequent sentence about rigidity.

L412 (Figure 5):It might be good to put a few thin gray lines that convey that there were three options for every option at every stage. From this image, it looks a bit like e.g. Upper Seismogenic Depth Moused.del 2 was chosen out of the three models, rather than that all three models were

Suggestion accepted.

Line 418 (Figure 6): This and Figure 6 could be combined into a two-panel figure.

Suggestion accepted.

Line 441 (Table 2): I think this could be supplementary or in an appendix.

Suggestion accepted.

L453: remove “(except for Overseas Countries and Territories, OTCs)”?

Suggestion accepted. The statement is needed only once.

L458-9. Remove??

We prefer to keep the statement because it points out some peculiar aspects of Mediterranean subduction that many readers may not be familiar with. 

Line 460 (Table 3): I think this could be supplementary or in an appendix.

Suggestion accepted.