Agent-Based Modeling of Tsunami Evacuation at Figueirinha Beach, Setubal, Portugal

Round 1

Reviewer 1 Report

This paper aims at simulating, using agent-based modelling, the evacuation in the situation of a tsunami in Figueirinha, in Portugal. The range of methods used in the study is fairly comprehensive, and organizing the paper with sub-section headings that correlate to the research questions should render easier to understand how the research tasks were accomplished. I do, however, have some major suggestions for improving the quality of the paper. First, authors should provide more information about the Agent-based model, putting the objectives, directives and details of sub-models, if ABM is clearly used. Sometimes I have the feeling that GIS models are used, while ABM and GIS are clearly not the same.
Second, no definition of concepts (evacuation) and the lack of the state of knowledge (in this area or in the research field) is missing in the introduction.
Third, the survey seem to use speed pathways and distance, but calibrated with only three persons. How can we sure that these data are sufficient, and if I think this could impede run-out of the models, test and sensibility to the initial conditions are never discussed.
Four, usefulness for the authorities and local stakeholders remain unclear.

 

 

Clear

Author Response

Thank you very much for your comments and suggestions.

 

Reply to the First comment: As indicated in the text, “the objectives of the paper are to use ABM to simulate the evacuation of beach users at Figueirinha Beach (Figure 1c) in order to understand whether it is possible to evacuate all beach users in less than 35 minutes [8], which corresponds to the time of the first tsunami wave’s arrival.”

Regarding the model development, it is an ABM programmed in Python, developed using Mesa, an open-source Python library for agent-based modeling. We revised the introduction of section 2.4 and provided two further references.

 

Reply to the Second comment: We revised the introduction by adding some references regarding the evacuation strategies.

 

 

Reply to the Third comment: Indeed, we used a small group of people to provide basic field survey data, and its results were validated by other published papers. We are aware that an evacuation exercise could provide results with a larger number of people, but it depends on the collaboration of civil protection. As mentioned in the introduction, evacuation exercises and drills are expensive and difficult to organize. However, we do not exclude conducting, in future work, a comparison of the model's results with empirical data obtained from evacuation exercises, as well as focusing on an analysis component regarding the computational model's robustness to variations in initial conditions.

 

Reply to the Fourth comment: As mentioned in the text, the present population data was provided by the Civil Protection. We had several meetings with them to discuss the data, and the model development, and our results will be directly used to update the evacuation plan of the beach.

The beach has several hazard problems, the most urgent in 2023 the rock-fall. As indicated in Figure 3a, the Left side of the road was closed because there were some large rock blocks that were in danger of falling. On the other hand, our findings show that the Left Route is the one that provides a quicker evacuation.

Reviewer 2 Report

The study conducted by Santos et al. utilizes an Agent-Based Model (ABM) to simulate the evacuation of beachgoers in the event of a tsunami. The findings of the study indicate that, among the 10 scenarios examined, evacuating all beach users in less than 35 minutes is not feasible. Additionally, the study identifies only two available evacuation routes. These insights hold valuable implications for coastal planning and disaster management. It is suitable for publication in Geosciences.

My main concern is regarding the 1755 scenario. As the authors stated, ABM allows the modeling of complex social and environmental systems by representing individuals or agents with their own decision-making rules and interactions. However, I wonder if conducting the worst-case scenario might lead to an overestimation of potential risks.

 

Other comments

1）Line 26: The first paragraph of the beginning should be used to review the historical tsunami disasters in Portugal. Besides the 1755 tsunami, what other notable events were there? I suggest not including excessive information about administrative institutions, as this is a scientific article.

2）Line 126: What magnitude of earthquakes specifically leads to inundation? Please specify.

3）Figure 2: The ground satellite images are sourced from where? Please provide a reference.

4）Table 1: A field survey is a relatively rough measurement, so there is no need to write to the decimal point.

5）Line 229: This model is the core of this paper, but the author lacks detailed description of it. Please provide a specific introduction to the mechanism of this model and its previous application cases and discuss its limitations in the Discussion section.

6）Line 338: There are significant differences in the results across different time periods. Besides schools, what other factors could possibly influence the results, such as tourists or weather conditions?

7）Line 450: As part of the discussion, the author should address corresponding mitigation measures. I suggest briefly mentioning the role of a tsunami early warning system, which can enhance the comprehensiveness of the paper.

Reference:

https://doi.org/10.1007/s00024-021-02910-z

https://doi.org/10.1029/2020JC016275

Minor editing of English language required.

Author Response

Thank you very much for your comments.

As indicated in the list below, (Other Comments, 1)), the review of the historical tsunamis was added to the text. Setubal was hit by only two tsunami events: in 1755 and in 1969. The scenarios calculated by Santos et al. 2017 [8] showed that the 1969 tsunami only inundated the coastline of the beach, while the 1755 event inundated the entire beach, including the parking lot up to about 8 m above mean sea level.

Following the 2004 Indian Ocean Tsunami and the 2011 Tohoku Tsunami, extreme events are considered the worst-case scenarios and must be taken into account for warning systems and emergency planning. Furthermore, there are many other worldwide examples of natural and technological disasters that lead to a change and update of legislation, rules, and procedures. Since 9/11, we also have learned to deal with terrorist and cyber attacks which are other examples of worst-case scenario situations.

 

Reply to comment 1): Thank you very much for your comments. Santos et al., 2017 [8] compiled the historical and instrumental earthquakes that triggered tsunamis in Portugal. Setubal was hit by the 1969 earthquake, with magnitude Ms 8.0, and the 1755 earthquake, with magnitude Mw 8.7. In Figuerinha beach no hazard has occurred yet, due to prevention measures carried out by the municipality stakeholders regarding rock-fall and forest fires. We added the list of previous tsunamis in the text and reduced the information about administrative institutions.

 

Reply to comment 2): The 1755 Earthquake had an estimated magnitude of Mw 8.7; the text was revised.

 

Reply to comment 3): The source is indicated in the plot itself, and is provided directly from the ESRI ArcMap software (watermark):

Source: Esri, DigitalGlobe, GeoEye, Earthstar Geographics, CNES/Airbus DS, USDA, USGS, AeroGRID, IGN, and the GIS User Community.

 

Reply to comment 4): Thank you very much for your comments. We removed the decimal point in Table 1.

Reply to comment 5): Thank you for your comments. We modified the introduction of section 2.4 and provided two further references.

Reply to the comment 6): Thank you very much for the comments. We did not consider other conditions in the model. The input parameters in the model were the number of children and adults; in the scenarios of June and July Morning, there are also schools. Therefore, in this paper, the main difference in the results is related to the presence or not of the schools at the beach, as mentioned in the text. We understand that other factors can be analyzed, and we do not exclude extending the work and address other factors in the model.

 

 

Reply to comment 7): Thank you very much for providing the two papers that have been included in the text. We also modified the discussion, accordingly.

Reviewer 3 Report

Geosciences-2552689   REVIEW 

 

This paper uses agent-based modeling to study the best way to evacuate beach goers in the event of a tsunami at a famous beach in Portugal.  The study is well done, and the paper is well written with only a few minor English errors.  I would recommend publication after perhaps a few minor changes. 

 

MINOR COMMENTS AND QUESTIONS: 

 

[1] Line 56, 61: You are using the tsunami generated by the 1755 Lisbon earthquake for your modeling.  Is it possible that an earthquake with an epicenter closer to the Portuguese coast could occur, resulting in a shorter tsunami arrival time?  Will your results apply in that case? 

 

[2] Line 78: For unfamiliar readers, a general brief description, with a bit more detail, of agent-based models might be useful. 

 

[3] Line 136: Do you mean that Meeting Point Left should be used in addition to Meeting Point Right? 

 

[4] Line 141: Do you mean “Route 1”? (is this a spelling issue?) 

 

[5] Line 174: Would it be useful to include km/hr (units familiar from auto driving)?  For example,  “… 1.0 m/s (or 3.6 km/hr) …”.

 

[6] Line 368-370: This is not a full sentence.  It is a sentence fragment.  One way of making it grammatically correct would be “… a limit state, for example, a function …”. 

 

[7] Line 402: The word “fewer” is more grammatically correct than “lesser”. 

 

[8] Line 440: Here is a possibly better solution: build a hospital at Meeting Point Left !! 

Good.  See comments to authors.

Author Response

Thank you very much for the comments and we and follow the suggestions below.

 

Reply to the comment [1]: As presented and discussed by Santos et al., 2017 [8], the seismic activity in Portugal is considered moderate. The authors carried out the tsunami numerical modeling of the 1969 and 1755 events that hit Setubal. The numerical model results show the 1969 tsunami reached Figueirinha beach at 44 minutes after the earthquake, and the 1755 event at 35 minutes. For this reason, the 1755 event is the worst-case scenario related to travel time and inundation area. We altered the text, by adding the 1969 event.

 

Reply to the comment [2]: We included a citation of a handbook [16] which has 30 chapters and can be useful for a student, stakeholder, or unfamiliar readers.

 

Reply to the comment [3]: Yes. Meeting Point Right already exists, and we propose a second area (Meeting Point Left) that should be used in addition to the Meeting Point Right. We altered the text.

 

Reply to the comment [4]: Yes. Thank you very much for the correction. We modified the text.

 

Reply to the comment [5]: Thank you very much for the suggestion. We modified the text.

 

Reply to the comment [6]: Thank you very much for the correction. We modified the text.

 

Reply to the comment [7]: Thank you very much for the correction. We modified the text.

 

Reply to the comment [8]: Thank you for the suggestion. However relocating the hospital is not that simple, and in the near future, it is not an option. Instead, we added some suggestions in the text: Continue to conduct preventive measures to contain the rock–fall; increase the topographic level of the road; installation of more walkways on the sand beach and increase the width of the existing ones.