the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 12 Mar 2025
Assessing atoll island future habitability in the context of climate change using Bayesian networks
Abstract. Atoll islands are threatened by multiple climate change impacts, such as sea-level rise and extreme sea-level events, and ocean warming and acidification. A recent approach to assess climate change risk to these islands is to use multi-criteria expert judgment methods. These approaches can serve as a basis to the development of Bayesian Networks (BNs) integrating expert knowledge and uncertainties to perform climate risk assessments. Here, we use the model structure and expert knowledge of (Duvat et al., 2021), who assessed future risk to habitability for four Indian and Pacific Oceans’ atoll islands, in order to discuss the advantages and limitations of the BN model. Advantages of the approach include the explicit treatment of uncertainties and the possibility to query expert knowledge in a non-trivial manner. For example, expert knowledge can be used to assess risks to habitability and future uncertainties and to explore inverse problems such as which drivers can exceed specific risk thresholds. Our work suggests that BN, though requiring a certain level of implementation expertise, could be used to assess climate change risk and support climate adaptation.
-
Notice on discussion status
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
-
Preprint
(2375 KB)
-
Supplement
(2358 KB)
-
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(2375 KB) - Metadata XML
-
Supplement
(2358 KB) - BibTeX
- EndNote
- Final revised paper
Journal article(s) based on this preprint
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2024-3884', Anonymous Referee #1, 26 Mar 2025
-
AC1: 'Reply on RC1', Mirna Badillo Interiano, 10 Jun 2025
We thank Reviewer #1 for their positive comment.
Citation: https://doi.org/10.5194/egusphere-2024-3884-AC1
-
AC1: 'Reply on RC1', Mirna Badillo Interiano, 10 Jun 2025
-
RC2: 'Comment on egusphere-2024-3884', Anonymous Referee #2, 26 Mar 2025
Summary of manuscript:
This study evaluates whether Bayesian Networks (BNs) based on expert judgments can effectively assess the integrated risk of climate change, and demonstrates a specific application to the habitability of atoll islands. A key methodological contribution is the translation of scores and confidence levels from Duvat et al. (2021) into probabilistic values, alongside the development of a Bayesian Network model. The manuscript presents a well-structured and clearly written methodology, with precise descriptions and strong visualizations. Strengths and limitations are discussed comprehensively. The study's scope aligns well with the focus of this special issue.
Minor comments:
-Lines 8-9: Reduce the use of "and" for improved readability.
-Line 11: Specify the model type used by Duvat et al. to clarify whether it is a Bayesian Network.
-Lines 12, 51, 69 etc.: Remove brackets from citations of Duvat et al. -Line 24: AR6 citation is missing.
-Lines 35ff: Land loss is not the only factor impacting food and water supply or economic activities— suggestion to include other relevant processes such as groundwater salinization.
-Lines 45ff: Clarify what is meant by “each driver of risk” and how it connects to Step 1 of the methodology.
-Table 1 Caption: The second and third sentences of the caption seem misplaced — consider integrating them into the main text or table itself. I would change "consideration of climate change" to "affected by climate change."
-Figure 2 (and others): Improve resolution.
-Line 118: What is the meaning of "and stores the prior probability."?
-Figure 3: Specify the unit—does it represent meters of erosion or a 1–5 scale from Duvat et al.?
-Equation 1: Define p(x_i) and I suggest to note in the text that conditional probabilities are used.
-Line 145: "Malé" instead of "Male’."
-Table 2: Consider adding flooding from precipitation as a hazard in Malé, given high urbanization, coastal engineering, and inadequate drainage systems. Also, one could mention freshwater salinization as a key hazard in the Maldives.
-Figure 6: The x-axis label ("risk level") is the same as the plot titles—clarify the distinction. Replace "CL" with "Confidence Level" for clarity. Instead of using two dotted lines to indicate intervals, consider a transparent shaded area for better visualization.
-Lines 224ff: Clearly state that the model is populated with these input data, rather than just used for validation. If validation means that aggregated risk to island habitability aligns with Duvat et al., be explicit about this definition to avoid ambiguity.
-Table 4: Ensure consistency in conditional probability notation (e.g., specify all dependencies or none, such as RCP vs. "Risk to island habitability = High/Very High").
Major comments:
-Clearly define the literature gap. Line 40: Provide a clearer definition of integrated risk assessment. Is the literature gap identified in the caption of Table 1, i.e. no Bayesian Networks for small islands include integrated risk assessments? Or is it that integrated risk assessments for small islands are generally lacking? Does the statement in Line 67 contradicts this by suggesting that some BN studies with integrated climate risk assessments do exist — please clarify.
-Terminology Precision (Lines 80ff): Ensure consistent and precise use of "risk" and "probability." It would be helpful to define key terms explicitly. For example: What constitutes risk? Are risk factors treated as variables, probabilities, or both? Does Research Question 1 focus on estimating the probability of island inhabitability? How is risk level determined? What are the drivers of risk? Later in the manuscript, risk criteria are introduced with a 1–5 scoring system. While the methodology clarifies the model and its variables, the Introduction would benefit from clearer definitions and framing to enhance coherence and reader understanding.
-Section 4.4: Effectiveness of Adaptation Measures. The study does not apply concrete adaptation measures but rather evaluates risk reduction scenarios. Consider renaming Figure 10 to “Risk levels under different reduction scenarios” or “… under different risk reduction measures” (as already phrased elsewhere in the manuscript). Additionally, in Line 386, stating that the study "considers examples of adaptation measures" is misleading—reword to reflect that it assesses potential risk reductions that could be achieved through adaptation.
Citation: https://doi.org/10.5194/egusphere-2024-3884-RC2 - AC2: 'Reply on RC2', Mirna Badillo Interiano, 10 Jun 2025
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2024-3884', Anonymous Referee #1, 26 Mar 2025
-
AC1: 'Reply on RC1', Mirna Badillo Interiano, 10 Jun 2025
We thank Reviewer #1 for their positive comment.
Citation: https://doi.org/10.5194/egusphere-2024-3884-AC1
-
AC1: 'Reply on RC1', Mirna Badillo Interiano, 10 Jun 2025
-
RC2: 'Comment on egusphere-2024-3884', Anonymous Referee #2, 26 Mar 2025
Summary of manuscript:
This study evaluates whether Bayesian Networks (BNs) based on expert judgments can effectively assess the integrated risk of climate change, and demonstrates a specific application to the habitability of atoll islands. A key methodological contribution is the translation of scores and confidence levels from Duvat et al. (2021) into probabilistic values, alongside the development of a Bayesian Network model. The manuscript presents a well-structured and clearly written methodology, with precise descriptions and strong visualizations. Strengths and limitations are discussed comprehensively. The study's scope aligns well with the focus of this special issue.
Minor comments:
-Lines 8-9: Reduce the use of "and" for improved readability.
-Line 11: Specify the model type used by Duvat et al. to clarify whether it is a Bayesian Network.
-Lines 12, 51, 69 etc.: Remove brackets from citations of Duvat et al. -Line 24: AR6 citation is missing.
-Lines 35ff: Land loss is not the only factor impacting food and water supply or economic activities— suggestion to include other relevant processes such as groundwater salinization.
-Lines 45ff: Clarify what is meant by “each driver of risk” and how it connects to Step 1 of the methodology.
-Table 1 Caption: The second and third sentences of the caption seem misplaced — consider integrating them into the main text or table itself. I would change "consideration of climate change" to "affected by climate change."
-Figure 2 (and others): Improve resolution.
-Line 118: What is the meaning of "and stores the prior probability."?
-Figure 3: Specify the unit—does it represent meters of erosion or a 1–5 scale from Duvat et al.?
-Equation 1: Define p(x_i) and I suggest to note in the text that conditional probabilities are used.
-Line 145: "Malé" instead of "Male’."
-Table 2: Consider adding flooding from precipitation as a hazard in Malé, given high urbanization, coastal engineering, and inadequate drainage systems. Also, one could mention freshwater salinization as a key hazard in the Maldives.
-Figure 6: The x-axis label ("risk level") is the same as the plot titles—clarify the distinction. Replace "CL" with "Confidence Level" for clarity. Instead of using two dotted lines to indicate intervals, consider a transparent shaded area for better visualization.
-Lines 224ff: Clearly state that the model is populated with these input data, rather than just used for validation. If validation means that aggregated risk to island habitability aligns with Duvat et al., be explicit about this definition to avoid ambiguity.
-Table 4: Ensure consistency in conditional probability notation (e.g., specify all dependencies or none, such as RCP vs. "Risk to island habitability = High/Very High").
Major comments:
-Clearly define the literature gap. Line 40: Provide a clearer definition of integrated risk assessment. Is the literature gap identified in the caption of Table 1, i.e. no Bayesian Networks for small islands include integrated risk assessments? Or is it that integrated risk assessments for small islands are generally lacking? Does the statement in Line 67 contradicts this by suggesting that some BN studies with integrated climate risk assessments do exist — please clarify.
-Terminology Precision (Lines 80ff): Ensure consistent and precise use of "risk" and "probability." It would be helpful to define key terms explicitly. For example: What constitutes risk? Are risk factors treated as variables, probabilities, or both? Does Research Question 1 focus on estimating the probability of island inhabitability? How is risk level determined? What are the drivers of risk? Later in the manuscript, risk criteria are introduced with a 1–5 scoring system. While the methodology clarifies the model and its variables, the Introduction would benefit from clearer definitions and framing to enhance coherence and reader understanding.
-Section 4.4: Effectiveness of Adaptation Measures. The study does not apply concrete adaptation measures but rather evaluates risk reduction scenarios. Consider renaming Figure 10 to “Risk levels under different reduction scenarios” or “… under different risk reduction measures” (as already phrased elsewhere in the manuscript). Additionally, in Line 386, stating that the study "considers examples of adaptation measures" is misleading—reword to reflect that it assesses potential risk reductions that could be achieved through adaptation.
Citation: https://doi.org/10.5194/egusphere-2024-3884-RC2 - AC2: 'Reply on RC2', Mirna Badillo Interiano, 10 Jun 2025
Peer review completion
Journal article(s) based on this preprint
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 914 | 132 | 25 | 1,071 | 59 | 22 | 38 |
- HTML: 914
- PDF: 132
- XML: 25
- Total: 1,071
- Supplement: 59
- BibTeX: 22
- EndNote: 38
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Mirna Badillo-Interiano
Jérémy Rohmer
Gonéri Le Cozannet
Virginie Duvat
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(2375 KB) - Metadata XML
-
Supplement
(2358 KB) - BibTeX
- EndNote
- Final revised paper
Accepted as it is