Dear author/s,

Thank you very much for your efforts. I have included my notes and suggestions below.

Abstract

The abstract says that multi-hazard approaches are not widely used in real-world situations, but it should explain what that means. It is suggested to make clear that older models often get risk wrong because they treat hazards like liquefaction or tsunamis as if they happen separately. The effect seen in the Napier case study should be measured, for example, by showing which hazard caused the most damage, or by noting that the model used data from 30,000 homes to produce its detailed results. Use stronger action words like “enable high-resolution risk pricing” or "permit evidence-based land-use planning." This will make the research more useful for people outside of academic circles.

Introduction

Please clearly define what you mean by internal consistency in this study. Does it mean consistency over time, across locations, or from different sources? Please point out the main problem with how samples are chosen. Say if the framework uses any methods to reduce differences in results or not.

Methodology

It is important to explain how the relationship between different intensity measures (IM) is managed. Giving a clear description of how these IM are combined would help.

The current method (shown in Figure 2) calculates damage separately and then combines it. In real life, shaking can weaken a building and make it less able to withstand a tsunami. The paper should discuss this complex damage process, or clearly state whether it uses a simple "max-damage" or "union" rule and what its limits are. The paper says more simulations are needed for results to settle than in single-hazard models. Please give a number or measure for this. In a multi-hazard model, the average damage might settle faster than the highest possible loss.

The tsunami model uses a set of 330 pre-made scenarios based on simple slip patterns. There might be a difference between the detailed slip used for shaking and the simpler slip used for the tsunami scenarios. To keep things clear, the paper should discuss how the seafloor changes in the specific simulation compared to those in the scenario library.

The landslide hazard is shown as just one movement value. Landslides in this setting can occur immediately during an earthquake or later due to rain on weakened slopes. The method should make clear whether it models only landslides that occur during the earthquake.

The model shows liquefaction and lateral spreading as separate parts in the diagram, but these often happen together. It should be clear whether the model considers them together or treats them as separate ways buildings can be damaged.

Results

The results show total damage numbers but often do not compare the Average Annual Loss for each hazard, either on its own or when combined.

Please add a Peril Contribution Plot. This would let readers see when hazards like tsunamis start to cause more damage than ground shaking.

The results are based on the max-damage rule. Please test how sensitive the results are to this rule. If the model used a different way to combine damage, like adding up the chances of damage instead of just taking the worst case, how much would the total loss change?

Discussion

To keep the calculations manageable, the tsunami model library was limited to 330 scenarios. During the simulation, the model picks the closest scenario from this set. The discussion should clearly measure the error caused by this choice. It is also important to discuss whether this approach might miss the most extreme risks, especially for events that do not fit neatly into pre-made scenarios.

The framework uses a max-damage rule to combine damage from different hazards. This is a cautious simplification. The discussion should talk about how damage can depend on the order of events, for example, how shaking can make buildings more likely to collapse in a tsunami. Saying that future versions could include more detailed damage rules would give a clear plan for future research.

The paper says that more simulations are needed for results to settle in multi-hazard models than in single-hazard ones. The discussion would be stronger with more detail on the trade-off between how quickly the model runs and how stable the results are.

Please discuss the potential issues with using US-based HAZUS damage estimates for New Zealand buildings. The discussion should also suggest using local damage data in the future to improve the average damage estimates.

The current model does not include other hazards that can occur at the same time (such as tides or storm surges) or long-term changes (such as sea-level rise). Talking about how rising sea levels from climate change would affect tsunami flooding in this model would make it more useful for long-term city planning, not just for earthquake risk.

Conclusion

The paper says that risk increases as more hazards are included. Please give a simple scientific summary about how much the risk increases. For example, adding more hazards raised the average yearly loss compared to a model with only shaking. This gives other researchers a number to compare with different locations or building types.

The results show that shaking and liquefaction are the main hazards in Napier. Please clearly say when this order of importance is true.

In the conclusion, talk about how much consistency was reached. While the hazards come from the same source, the ways buildings are damaged mostly stay separate.

Thank you again for your work,

Best regards.

General comments

The manuscript presents a Monte Carlo-based probabilistic framework that couples ground shaking, tsunami, liquefaction, lateral spreading, and landslides for cascading earthquake risk assessment, demonstrated on residential buildings in Napier, New Zealand, using a 100,000-year Hikurangi Subduction Zone catalogue. However, although the effort is highly appreciated, the work does not provide a meaningful innovation or a significant improvement over existing multi‑hazard methodology.

Specific comments

1. The tsunami modelling relies on only 33 rupture scenarios, with 10 slip variations each, explicitly due to computational constraints. This small scenario set may under‑represent the full epistemic and aleatory uncertainty in subduction tsunami generation.
2. Liquefaction and lateral spreading are modelled conditionally on shaking, yet the paper does not clearly state whether co-occurrence effects (e.g., lateral spreading exacerbating liquefaction damage) are captured or treated independently.
3. The landslide damage model uses a rule-based approach with 3 damage states. Please comment more on the choice and on the damage model. No limitation are presented.
4. A justification for using HAZUS fragility functions in New Zealand should be given.
5. Residential building exposure is simply mentioned in the paper without a discussion of the structural typologies, and without any sensitivity or completeness analysis.
6. The paper does not assess scalability or feasibility of the proposed methodology which is computationally cumbersome and may limit real-world applicability.
7. The combination of damage states from multiple hazards into a single harmonized DS lacks a rigorous and thorough approach.
8. Why is only the Hikurangi Subduction Zone (HSZ) taken into account? A seismic hazard disaggregation should be presented to justify the choice.
9. How are the epistemic uncertainties from the fragility functions or hazard estimations taken into account?
10. Extensive geotechnical investigations are mentioned in the paper, yet no clear presentation is shown.
11. The study stops at damage states and ratios without translating them into annualized losses, repair costs, downtime, or social impacts.
12. The authors should explicitly address non-structural damage due to ground shaking.
13. Since less than 20% are exposed to lateral spreading or landslide, it is obvious that the most important hazards are ground shaking and liquefaction.
14. Landslide hazard modelling consists only on open-slope dry rock and debris avalanches. Is this appropriate for Napier? For other sites, can the modelling be extended?
15. What does “Internally consistent framework” actually mean? Could the authors extend the definition?

In my opinion, before the work can be considered for publication, the authors should substantially strengthen the methodological justification, expand the modelling components, clarify assumptions, and demonstrate the added value and scalability of the proposed framework. Addressing the specific points listed above would greatly improve the scientific contribution and reliability of the study.

The overall quality of this manuscript is very low and it fails to meet the minimum standards required for publication. The study lacks a clearly articulated and scientifically meaningful research question, and the motivation and claimed contributions are weak and unconvincing. The work does not demonstrate substantive novelty relative to existing studies, and much of the manuscript relies on methodological description without delivering new scientific insight. The data, assumptions, and methodological choices are insufficiently justified, and key sources of uncertainty and limitation are not rigorously addressed, which undermines the robustness and credibility of the results. The analysis remains superficial, with a clear lack of validation, benchmarking, or critical comparison against established approaches. Most importantly, the discussion is inadequate: it does not engage deeply with relevant literature, does not critically interpret the results, and fails to clearly articulate the scientific implications, applicability, or limitations of the work. Overall, the manuscript lacks scientific rigor, originality, and clear added value, and is therefore not suitable for publication in its current form.