Pseudo-Probabilistic Design for High-Resolution Tsunami Simulations in the Southwestern Spanish Coast

Round 1

Reviewer 1 Report

This study generates many input parameters for tsunami simulation considering probability density distributions and performs simulation based on them. Actually, the study seems to be conducted in a reasonably scientific manner, but I think that the results are far from the main goal written in abstract: "Understanding the potentially devastating effects leads to the development of security and resilience measures such as evacuation plans or economic investment necessary to rapidly soften its consequences." The authors clearly state that such examination is a future work. Nevertheless, I think that at least the authors must summarize the features of the immense simulation results to show how the results vary according to the variation of the input models that are carefully considered in this study. Without that, the readers cannot recognize the effectiveness of your approach to generating the input models. Just showing the two specific results is meaningless. Even if the authors consider that this work is limited to elaboration of the simulation technique, it is important to show what difference is produced by the proposed approach.

In addition, the authors should be careful that the simulations of the neighboring subregions are not necessary consistent in a nested tsunami simulation. For example, in Figure 5, the maximum wave height has discontinuity at the boundary of two regions between the at about -6deg55min and -6deg50min. In the landed region, the wave height distribution linearly disappears. I know such discontinuity of the simulation results between different calculation regions is common but the authors must make figures more carefully to avoid readers' confusion or properly explain the existence of such discontinuity.

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 2 Report

The authors use a pseudo-probabilistic sampling approach to produce 896 deterministic tsunami simulations to study tsunami hazard at a high resolution in the southwest Spanish coast. I propose this paper for publication with the following corrections/clarifications.

Comments:

The title of the manuscript is a bit misleading as one would expect probabilistic outputs but figures 5-10 show the outputs of the deterministic runs. I recommend adding the word “sampling” or “design” in the title to clarify better the scope of the study: “Pseudo-probabilistic sampling approach to high-resolution … coast” or “Pseudo-probabilistic design for high-resolution..” etc.

Section 1.2 L46-53 Emulation techniques are also worth mentioning here as a means to reduce the number of computational runs for uncertainty quantification (see for example Gopinathan et al. 2021).

L89 : “the” discussion section. Some other minor corrections in the language should be done throughout the text.

Section 1.3 or Section 2.1: It is not clear how this work builds upon previous work. For example, the authors can expand on whether the tsunamigenic potential of these faults has been studied in the past and relevant results.

Section 2.3 It is not clear why the rake and strike are selected to vary as opposed to other fault parameters such as for example the length of the fault which can have a critical impact on inundation. The rake and strike parameters largely depend on the geometry of the fault so the variance in these cases is to a large extent predetermined. Please elaborate more and also provide with the parameter ranges.  

Section 2.3 Variance in the fault segmentation can also play a critical role for inundation. It is not clear if this is considered in this study.

Line 203-204: Rather than the gap or cluster of input values an issue with LHS and similar methods is the waste of runs in unneeded regions. Adaptive sequential methods are usually explored to account for that (see for example Salmanidou et al., 2021).

Lines 216-222: It is also worth mentioning the literature in relation to tsunami modelling as for example the work by Snelling et al. 2020.

Line 276: Please also add the timings of the simulations.

Tables 3-7 can also be moved to the Appendix.

Lines 331-332 (hence a total...computed):  this sentence reads ambiguous I would remove it or rephrase.

Figures 5-10: It would help to associate the domains of these figures with figure 4 to understand the approximate location of the sub-grids e.g. by naming the sub-grids in all figures.

Section 3.4 Figures 5-10: Please elaborate more on the results. Do the figures show the results of one deterministic run or the maximum outputs of all the runs? If it is the former, why are those specific faults selected? Are they the worst-case scenarios?

Final comment: This is only a suggestion, but I believe it would elevate the manuscript to have a figure that somehow associates the sampled data with the end results of all the selected runs for a specific fault/location.

Literature:

Gopinathan Devaraj, Heidarzadeh Mohammad and Guillas Serge 2021 Probabilistic quantification of tsunami current hazard using statistical emulationProc. R. Soc. A.4772021018020210180 http://doi.org/10.1098/rspa.2021.0180

Salmanidou, D. M., Beck, J., Pazak, P., and Guillas, S.: Probabilistic, high-resolution tsunami predictions in northern Cascadia by exploiting sequential design for efficient emulation, Nat. Hazards Earth Syst. Sci., 21, 3789–3807, https://doi.org/10.5194/nhess-21-3789-2021, 2021

Snelling, B.; Neethling, S.; Horsburgh, K.; Collins, G.; Piggott, M. Uncertainty Quantification of Landslide Generated Waves Using Gaussian Process Emulation and Variance-Based Sensitivity Analysis. Water 2020, 12, 416. https://doi.org/10.3390/w12020416

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 3 Report

Overall this work is excellent. My main concern is in the introduction part. The authors need to describe in more detail the previous studies of tsunami simulations and PTHA in the Southwest Spanish coast, especially around the Andalusian Atlantic coast. The possible previous studies of such pseudo-probabilistic method must also be described in detail. In the other part of the manuscript, such as in the method, results, discussion, the authors already described and discussed the work in detail.

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 4 Report

Congratulation for your job. 

Only two comments:

a) add figures from wave propagation to the shoreline and inside (around the buildings) to show example from your methodology.

b) I consider that its necessary to add a paragraph to the conclusions in which the clearly state (in points) how your method affect or correct the result and the final product (for the insurance company).

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Thank you for considering my comments on the previous review. The manuscript and figures are favorably improved.