Bow-Tie-Based Risk Assessment of Fishing Vessel Marine Accidents in the Open Sea Using IMO GISIS Data

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The topic is relevant, and applying a structured RA framework to fishing vessel safety is a recommendable effort. However, the overall research objective is unclear. It remains ambiguous whether the study aims to model accident risk, analyze causation, or support preventive measures. Following is my comments:.

1. In abstract the opening phrase “This used…” is grammatically incorrect. It should read “This study used” to ensure proper academic phrasing.
2. In the abstract the phrase “67 open-sea shipping vessel accidents” is unclear and inconsistent with the paper’s focus on fishing vessels.
3. The abstract lacks a clear problem statement or motivation. It immediately lists methods and accident types without explaining the research gap or why open-sea fishing vessels represent a distinct problem.
4. In the introdusction, the firts paragraph about global seafood consumption is irrelevant to the paper’s technical focus on fishing vessel safety and risk assessment.
5. Introduction, lines 47-54: Several claims about fragmented jurisdiction, lack of legislative coverage, and limited rescue systems are presented without supporting references.
6. In the introdusction (lines 56-63) the claim of “extensive research associated with commercial shipping” is exaggerated, and reference [10] relates to aviation, not commercial shipping or maritime safety. Citations should be revised to accurately reflect relevant studies.
7. Further (more focused) literature review is factual but overly descriptive. Many cited works are summarized without synthesis or critical linkage to the present study.
8. In the introduction (lines 56-63) Authors mix the problem identification with methods and lacks references supporting the claim that open-sea studies are limited.
9. The paper does not reference recognized issues such as high traffic density, informal operations, or weak coastal regulation, which are well-documented in global and Chinese fisheries reports.
10. The data source includes only officially reported cases, excluding many coastal and nearshore fishing accidents, particularly in Chinese and regional waters with limited reporting. This represents a significant limitation of the study’s data representativeness and overall validity.
11. The methodological framework is technically sound, but the empirical basis is weak. The dataset includes only 67 cases, which are limited to officially documented open-sea incidents and exclude most coastal fishing accidents.
12. The study does not distinguish which party initiated the regulatory violation or operational failure, limiting the applicability of findings for preventive or corrective measures.
13. The inclusion of “pump failure/insufficient capacity” as a causal factor is unclear (lines 276-277). Bilge pump capacity is unlikely to be a determining cause in collision- or capsize-related sinkings unless prolonged flooding scenarios are explicitly analyzed. The causal linkage should be clarified or reconsidered.
14. The bow-tie approach is applied mechanically without defining the top event, barriers, or intended application. It is unclear how the integrated FTA–ETA–QRA results translate into actionable risk management or prevention measures.
15. The discussion merely restates the statistical results without deeper interpretation or connection to real-world safety management. The conclusions lack analytical depth and do not clarify how the bow-tie based findings can inform preventive or operational strategies.

Comments on the Quality of English Language

Some sentences are not smooth or grammatically off, and technical terms are used inconsistently. 

Author Response

Dear Reviewer,

C1. In abstract the opening phrase “This used…” is grammatically incorrect. It should read “This study used” to ensure proper academic phrasing

C2.  In the abstract the phrase “67 open-sea shipping vessel accidents” is unclear and inconsistent with the paper’s focus on fishing vessels.

C3.  The abstract lacks a clear problem statement or motivation. It immediately lists methods and accident types without explaining the research gap or why open-sea fishing vessels represent a distinct problem.

A1-3. The abstract has been fully revised (14-27).

C4. In the introduction, the firts paragraph about global seafood consumption is irrelevant to the paper’s technical focus on fishing vessel safety and risk assessment.

C5. Introduction, lines 47-54: Several claims about fragmented jurisdiction, lack of legislative coverage, and limited rescue systems are presented without supporting references.

C6. In the introdusction (lines 56-63) the claim of “extensive research associated with commercial shipping” is exaggerated, and reference [10] relates to aviation, not commercial shipping or maritime safety. Citations should be revised to accurately reflect relevant studies.

C7. Further (more focused) literature review is factual but overly descriptive. Many cited works are summarized without synthesis or critical linkage to the present study.

C8. In the introduction (lines 56-63) Authors mix the problem identification with methods and lacks references supporting the claim that open-sea studies are limited.

A4-8. The introduction has been fully revised (31-142).

C9. The paper does not reference recognized issues such as high traffic density, informal operations, or weak coastal regulation, which are well-documented in global and Chinese fisheries reports

A9. The revised manuscript now acknowledges the absence of issues like high traffic density, informal operations, and weak coastal regulation due to dataset limitations, specifically in the Discussion section (762-770).

C10. The data source includes only officially reported cases, excluding many coastal and nearshore fishing accidents, particularly in Chinese and regional waters with limited reporting. This represents a significant limitation of the study’s data representativeness and overall validity.

A10. We sincerely appreciate your observation regarding data representativeness. The revised Discussion and Conclusion now clearly state that our analysis is limited to officially reported cases, which exclude many coastal and nearshore accidents—particularly in regions with limited reporting. We recognize this as a significant limitation and have advised interpreting our findings with caution (762-770).

C11. The methodological framework is technically sound, but the empirical basis is weak. The dataset includes only 67 cases, which are limited to officially documented open-sea incidents and exclude most coastal fishing accidents.

A11. To ensure objectivity and reliability, the study was based on IMO GISIS data, and the limitations of this approach are clearly acknowledged in the revised manuscript. We have stated that our findings are preliminary and have recommended further research with broader datasets (135-142, 762-770, 846-852).

C12. The study does not distinguish which party initiated the regulatory violation or operational failure, limiting the applicability of findings for preventive or corrective measures.

A12. The revised manuscript addresses this point by providing a detailed analysis of regulatory violations in documented collision accidents. As stated in the Results section, a review of 35 collision cases identified a total of 78 breaches of the COLREGs, revealing that multiple violations commonly occurred in individual incidents. The most frequently infringed rules and the average of 2.2 breaches per accident are reported, demonstrating that collisions result from several interacting procedural failures, not a single lapse. This addition enhances the practical value of the findings for preventive and corrective strategies (436-445).

C13. The inclusion of “pump failure/insufficient capacity” as a causal factor is unclear (lines 276-277). Bilge pump capacity is unlikely to be a determining cause in collision- or capsize-related sinkings unless prolonged flooding scenarios are explicitly analyzed. The causal linkage should be clarified or reconsidered

A13. The manuscript clarifies that pump failure or insufficient pump capacity is a critical causal factor specifically for sinking events involving progressive flooding, rather than for collision or capsizing accidents. This is supported by FTA analysis and risk ratio values, which demonstrate that inadequate bilge pump capacity, particularly on older vessels, can lead to uncontrolled water ingress and ultimate vessel loss. The revised content elaborates on the cascade mechanism and type-specific causation, addressing the reviewer’s concern (529-538).

C14. The bow-tie approach is applied mechanically without defining the top event, barriers, or intended application. It is unclear how the integrated FTA–ETA–QRA results translate into actionable risk management or prevention measures.

A14. The revised manuscript now explicitly defines the top events, barriers, and intended analytical scope of the bow-tie framework. We clarified that the study focuses on causal pathways and severity outcomes based on officially documented accident factors, while not evaluating existing safety barriers (240-244).

Additionally, we provided accident-type-specific preventive measures—covering collision, capsizing, and sinking prevention—to illustrate how the integrated FTA–ETA–QRA results translate into actionable safety strategies (798-808).

C15. The discussion merely restates the statistical results without deeper interpretation or connection to real-world safety management. The conclusions lack analytical depth and do not clarify how the bow-tie based findings can inform preventive or operational strategies.

A15. The revised discussion section provides deeper interpretation by connecting the statistical findings to practical safety management. It analyzes how specific risk factors identified through the bow-tie, FTA, ETA, and QRA methods translate into preventive strategies—such as targeted COLREG training, maintenance of technical equipment, and operational controls for different accident types (798-808).

The conclusion further clarifies how these risk structures inform differentiated interventions and policy development, thereby addressing analytical depth and real-world applicability (853-856).

 

 

We sincerely appreciate your thoughtful review. We recognize that the available data are limited and that there remain many aspects of the study that can be further improved. Your comments offer valuable direction, and we will keep them in mind as we continue to advance the quality and depth of our research. Thank you very much.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

The authors present a comprehensive risk assessment framework of open-sea fishing vessel accidents, which combines the bow-tie model, FTA, Firth logistic regression, ETA and QRA. The manuscript is well structured and clearly written and the findings are significant and represent a novel contribution to the field. I recommend that this manuscript could be accepted after minor revisions have been made to address the points below.

1. The Introduction section reviews the related works on maritime accidents and claims that the fishing vessel accidents in the open seas have been relatively well studied, as well as highlighting other limitations. However, this literature review is inadequate as it does not present the latest research achievements and the research gap in fishing vessel accidents. The claimed research gaps may not sufficiently justify the contributions and novelty of this study. It’s therefore advised to improve the part of literature review.
2. data limitations. The limitation of sample size of fishing vessel accidents is transparently acknowledged by the authors. The small sample size of fishing vessel accidents (N=67) is still insufficient for this study although the Firth regression is applied to respond to this challenge. Consequently, the results and findings should be interpreted with caution in view of this limitation. It’s recommended that the discussion and conclusion should more emphasize that the findings are preliminary and should be validated using a larger, more robust dataset.
3. section 2.2 methods. This section expounds the bow-tie model, FTA, Firth logistic regression, ETA and QRA that were applied to the methodology. However, the presentation is unclear and illogical. Therefore, a logical and knowledgeable framework/chart flow of the study is advised for a clearer presentation.
4. Equations (1) and (2) are very simple and not presented in the proper way. It would be readily more readable if more information about these equations and the elements of Tables 6 and beyond, which are derived by those methods (FTA, ETA and QRA) is provided.
5. line 37 page 1. ‘…forecast to increase 4.2% 36 from 20.4 kg in 2017 to 21.3 kg in 2018…’ is improper. This description should be updated to the year 2024 or 2025.

Author Response

Dear Reviewer,

C1. The Introduction section reviews the related works on maritime accidents and claims that the fishing vessel accidents in the open seas have been relatively well studied, as well as highlighting other limitations. However, this literature review is inadequate as it does not present the latest research achievements and the research gap in fishing vessel accidents. The claimed research gaps may not sufficiently justify the contributions and novelty of this study. It’s therefore advised to improve the part of literature review.

A1. We have strengthened the literature review by incorporating recent research achievements and by more clearly articulating the remaining research gaps in open-sea fishing vessel accidents. (71-82).

C2. data limitations. The limitation of sample size of fishing vessel accidents is transparently acknowledged by the authors. The small sample size of fishing vessel accidents (N=67) is still insufficient for this study although the Firth regression is applied to respond to this challenge. Consequently, the results and findings should be interpreted with caution in view of this limitation. It’s recommended that the discussion and conclusion should more emphasize that the findings are preliminary and should be validated using a larger, more robust dataset.

A2. To ensure objectivity and reliability, the study was based on IMO GISIS data, and the limitations of this approach are clearly acknowledged in the revised manuscript. We have stated that our findings are preliminary and have recommended further research with broader datasets (135-142, 762-770, 846-852)

C3.  section 2.2 methods. This section expounds the bow-tie model, FTA, Firth logistic regression, ETA and QRA that were applied to the methodology. However, the presentation is unclear and illogical. Therefore, a logical and knowledgeable framework/chart flow of the study is advised for a clearer presentation.

A3. A methodological flow chart has been added to Figure 2 to provide a clearer and more logical presentation of the analytical framework (212).

C4. Equations (1) and (2) are very simple and not presented in the proper way. It would be readily more readable if more information about these equations and the elements of Tables 6 and beyond, which are derived by those methods (FTA, ETA and QRA) is provided.

A4. We have added additional explanatory information and further enhanced readability by expanding Sections 2.2.2 (FTA) and 2.2.4 (ETA) to provide clearer methodological descriptions. (122-134, 203-263, 264-275, 302-306, 307-325).

C5. line 37 page 1. ‘…forecast to increase 4.2% 36 from 20.4 kg in 2017 to 21.3 kg in 2018…’ is improper. This description should be updated to the year 2024 or 2025.

A5. The entire Introduction section has been revised to improve clarity, structure, and alignment with the reviewer’s suggestions.

 

We sincerely appreciate your thoughtful review. We recognize that the available data are limited and that there remain many aspects of the study that can be further improved. Your comments offer valuable direction, and we will keep them in mind as we continue to advance the quality and depth of our research. Thank you very much.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

This study proposes a comprehensive risk assessment framework that identifies the dominant factors of various accidents, addressing the shortcomings of existing research in systematically analyzing accidents involving fishing vessels on the high seas. However, the current study has the following limitations:

1. The sample size used in this study is too small, particularly with only 6 capsizing incidents, which may compromise statistical significance and generalizability. Please add more samples.
2. What is the basis for the logic gate settings in the Bow-Tie diagram? Please supplement with sensitivity analysis.
3. Provide more specific cases or case studies demonstrating that violations of COLREG are the primary cause of collisions.
4. Given the correlation between environmental factors like weather, explain the rationale for assuming event probabilities are independent in the QRA model.
5. Include a comparative analysis with previous maritime safety risk assessment models.
6. Provide a comparison between the Bow-Tie-based model and other common assessment methods (e.g., Monte Carlo simulation or Bayesian networks).
7. Describe how uncertainties in event probabilities (e.g., weather conditions) are addressed.

Author Response

Dear Reviewer,

C1. The sample size used in this study is too small, particularly with only 6 capsizing incidents, which may compromise statistical significance and generalizability. Please add more samples.

A1. We fully agree that the limited sample size—particularly the small number of capsizing cases (n = 6)—restricts statistical significance and generalizability. However, to ensure objectivity and reliability, this study relies exclusively on officially reported IMO GISIS data, which inherently reflect the current limitations of fishing-vessel accident reporting systems. Accordingly, the revised manuscript explicitly states that the findings should be considered preliminary and require further validation as more comprehensive international datasets become available (135-142)

To address rare-event and small-sample bias, we applied bias-reduced (Firth) logistic regression, supplemented by Fisher’s exact test and the Haldane–Anscombe correction to confirm analytical consistency. These methods were selected specifically to improve robustness under severe sample limitations (296-301).

We also clarified several sources of uncertainty, including limited sample size, voluntary reporting bias in IMO GISIS submissions, and the assumption of conditional independence among environmental variables, which may influence probability estimates under extreme weather co-occurrence (671-678).

C2.  What is the basis for the logic gate settings in the Bow-Tie diagram? Please supplement with sensitivity analysis.

A2 We clarified the basis for the logic gate settings in the revised manuscript. The AND/OR gates in the bow-tie diagram were assigned directly from causal relationships explicitly described in official accident investigation reports. AND gates were used when multiple factors had to occur simultaneously to trigger the top event, whereas OR gates were applied when a single factor alone could initiate the accident. This explanation has been added along with examples for collision, capsize, and sinking cases (225-235).

To validate these configurations, we conducted a sensitivity analysis by altering the AND/OR structures of the dominant causal blocks for each accident type. The results showed that collision risks were relatively insensitive to gate changes, confirming the structural robustness of the human-factor block. Capsize risks exhibited strong asymmetry between overload-only and overload-plus-environment scenarios, reflecting the multifactor nature of capsizing events. For sinking, both pump–hull OR and AND configurations yielded risks lower than the baseline, suggesting the presence of additional unreported causal pathways (368-393; 710-753).

C3.  Provide more specific cases or case studies demonstrating that violations of COLREG are the primary cause of collisions.

A3. We added specific evidence showing that COLREG violations are the primary cause of collisions, including the fact that 35 collision cases contained 78 rule breaches, with multiple violations occurring in most incidents (436-445).

C4. Given the correlation between environmental factors like weather, explain the rationale for assuming event probabilities are independent in the QRA model.

A4. We clarified that the QRA model assumes conditional independence among event probabilities. This assumption is justified because (1) ETA decomposes consequence branches for computational clarity, (2) environmental factors no longer directly influence post–top-event consequence transitions, and (3) this approach is consistent with standard ETA/QRA practices widely used in maritime risk assessment and IMO FSA methodology (318-325).

 

C5. Include a comparative analysis with previous maritime safety risk assessment models.

C6. Provide a comparison between the Bow-Tie-based model and other common assessment methods (e.g., Monte Carlo simulation or Bayesian networks).

A5-6. We addressed the reviewer’s comment by adding a comparative explanation between the bow-tie-based framework and previous maritime safety risk assessment models, particularly Bayesian Networks (BN). The revised manuscript clarifies that although BN frameworks are theoretically capable of modeling accident causation and consequences, data-driven BN structures face fundamental limitations due to the lack of contextual information in marine casualty datasets (e.g., latent factors, failed barriers). As a result, BN models tend to capture correlations rather than true causal mechanisms (250-263).

In contrast, the bow-tie–FTA–ETA–QRA framework explicitly represents causal pathways and accident-type-specific structures, allowing systematic integration of causation, consequence development, and severity outcomes. These revisions clarify the methodological rationale for selecting the bow-tie approach and provide the comparative analysis requested by the reviewer (112-134).

C7. Describe how uncertainties in event probabilities (e.g., weather conditions) are addressed.

A1-3. We have explicitly stated this point by clarifying in the manuscript that the ETA model assumes conditional independence at each branching point, consistent with standard ETA methodology (318-325).

 

We sincerely appreciate your thoughtful review. We recognize that the available data are limited and that there remain many aspects of the study that can be further improved. Your comments offer valuable direction, and we will keep them in mind as we continue to advance the quality and depth of our research. Thank you very much.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

Thanks for the revisions and for carefully addressing the reviewer comments. The manuscript has been improved, and most of the previously identified issues have been resolved. The abstract, introduction and methodological explanations are clearer, and the limitations of the dataset are now more appropriately acknowledged. The additions regarding COLREG violations, pump related sinking mechanisms and the clarification of top events in the bow tie model are constructive and strengthen the paper.

A few points still require minor clarification, particularly the need for a more concise articulation of the research gap and clearer justification of the bow tie application given the absence of barrier analysis. These can be addressed with minor edits.

Overall, the manuscript is now close to publishable quality. I recommend acceptance after minor revisions.

Comments on the Quality of English Language

A few sentences could benefit from minor stylistic polishing to improve conciseness and avoid overly long constructions.

Author Response

Dear Reviewer,

C1. More concise articulation of the research gap.

A1. The section has been revised to provide a more concise articulation of the research gap (104-109).

C2. Clearer justification of the bow tie application given the absence of barrier analysis.

A2. The section has been revised to provide a clearer justification for applying the bow-tie method despite the absence of barrier analysis (261-290).

C3.  Improve conciseness and avoid overly long constructions.

A3. The sentence has been made more concise and the overly long construction has been revised as follows.

 

Original 1: Despite increasing acknowledgment of its economic significance, fishing vessel operations consistently exhibit markedly higher accident rates than those of commercial merchant shipping.

 

Revised 1: Despite their economic importance, fishing vessels have markedly higher accident rates than commercial merchant ships (34-35).

 

 

Original 2: Fishing vessel accidents occurring in open-sea waters present challenges that differ fundamentally from those in coastal or territorial areas. These incidents are characterized by fragmented jurisdiction and dispersed responsibility, as well as reporting and investigation systems that are considerably weaker than those established for merchant ships.

 

Revised 2: Open-sea fishing vessel accidents differ fundamentally from coastal incidents due to fragmented jurisdiction, dispersed responsibility, and weaker reporting and investigation systems (41-42).

 

 

Original 3: Because open-sea accident data rely heavily on limited and voluntary reporting, the findings of this study should be regarded as preliminary and subject to further validation as more comprehensive datasets become available through improved international reporting schemes.

 

Revised 3: Given limited voluntary reporting, findings are preliminary and require validation through improved international reporting systems (133-134).

 

 

Original 4: These differentiated causation structures highlight the need for accident-type-specific management strategies: enhancing regulatory compliance and watchkeeping to prevent collisions, improving stability management and load control to prevent capsizes, and strengthening hull integrity and pump system reliability to mitigate sinking events.

 

Revised 4: Prevention strategies must be accident-type-specific: strengthen compliance and watchkeeping for collisions, enforce loading controls for capsizes, and improve hull and pump reliability for sinkings (603-605).

 

 

Original 5: Beyond the quantitative modeling, this analysis links statistical findings to operational risk management practices. For collision prevention, priority measures include intensified COLREG and watchkeeping training, enhanced bridge resource management (BRM), fatigue mitigation protocols, and the adoption of navigation-support technologies such as AIS, radar, and ECDIS.

 

Revised 5: Statistical findings translate to specific operational controls: collision prevention re-quires intensified COLREG training, BRM, fatigue protocols, and navigation-support technologies (AIS, radar, ECDIS) (823-825).

 

 

Original 6: As datasets broaden and models evolve, the causal and sensitivity-based analytical structure presented in this study provides a scalable platform for identifying and managing the distinct safety risks inherent to open-sea fishing operations.

 

Revised 6: This framework provides a scalable platform for managing accident-type-specific safety risks as reporting systems improve (843-845).

 

 

We sincerely appreciate your thoughtful and constructive review. Your careful assessment and detailed insights have contributed meaningfully to the further refinement of this manuscript. Your comments provided valuable guidance, and we have carefully considered them while improving the quality and clarity of our study. Thank you once again for your time and expertise.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

I have no further comments.

Author Response

We sincerely appreciate your thoughtful and constructive review. Your careful assessment and detailed insights have contributed meaningfully to the further refinement of this manuscript. Your comments provided valuable guidance, and we have carefully considered them while improving the quality and clarity of our study. Thank you once again for your time and expertise.