This study investigates the potential tsunami hazard from landslide scenarios in the Gulf of Pozzuoli using a sequence of numerical models. The authors present four scenarios (three submarine and one subaerial) and simulate the associated tsunamis using both shallow water (SW) and non-hydrostatic (NH) models. The study addresses important local hazard concerns in a densely populated coastal area. While the paper is generally well-structured, several critical aspects require clarification and improvement before the manuscript can be considered for publication.

Major Comments:

Landslide Scenario Definition and Assumptions



The tsunami waveforms generated by landslides are sensitive to the initial conditions, including the location, volume, geometry, and material properties of the sliding mass. The authors mention that the four scenarios were constructed based on a "worst-case credible" approach. However, the paper lacks sufficient detail on how the initial conditions, especially volume and geometry, were determined. The cited reference (Zaniboni and Armigliato, 2025) is listed as a work in progress and is not yet available, making it difficult to assess the robustness of the scenarios.  

Moreover, the assumption that geotechnical and geomorphic properties are similar across all four scenarios is not justified or even explicitly stated. This assumption should be clarified, as differences in material properties can significantly influence landslide dynamics and tsunami generation. For instance, variations in density, cohesion, yield strength and/or internal friction angle can lead to different failure mechanisms and velocities, thereby affecting the characteristics of the generated tsunami waves.



The authors should provide more detailed explanations for each scenario's setup, including volume estimates, slope angle, and material assumptions. If geotechnical properties are assumed identical across cases, this simplification should be clearly stated and discussed, along with a justification for why this simplification is reasonable in this context.

Tsunami Generation Mechanism and Modeling Approach

The modeling approach uses a one-way coupling scheme, where landslide motion is simulated independently and used as input for tsunami generation and propagation. This approach, while computationally efficient, may be insufficient to capture certain physical mechanisms, particularly for subaerial landslides like scenario 4, where the interaction with the water column is highly dynamic and nonlinear. In reality, the water displaced by the landslide can, in turn, influence the landslide's motion.

The subaerial case involves a mass plunging from above sea level into the water with high velocity, which contrasts with the more gradual submarine slope failures of the other scenarios. Given these differences in physical processes, it is unclear whether the same numerical treatment is equally valid for both types of landslide.



The authors should justify the use of the same modeling framework across all scenarios and clearly discuss the limitations of their approach, especially regarding the subaerial case. They should acknowledge the potential limitations of the one-way coupling and discuss how this might affect the accuracy of their results.

Role of Froude Number and Energy Transfer Efficiency

A crucial omission in the discussion of tsunami generation is the role of the Froude number Fr=U/gH, which characterizes the relationship between the landslide velocity (U) and the shallow water wave speed (gH). When the Froude number approaches 1, energy transfer from the landslide to the water is most efficient due to resonance-like effects. This can lead to significant amplification of the generated waves.  

While the authors analyze landslide velocities and acknowledge the influence of slide speed and dispersion, they do not discuss the possible amplification effects that occur when the slide velocity approaches the wave celerity. This is particularly relevant for Scenario 4.



The authors should discuss whether any of their scenarios reach near-critical Froude conditions and, if so, whether their model can appropriately represent the associated amplification. Even a simplified estimation of the Froude number for each case would enhance the paper. This analysis would provide a more complete understanding of the tsunami generation process and its potential impact.

Minor Comments:

L9 "consisting into" -> “consisting of”

L38 “In the specific” -> “Specifically”

L43 “assessing” -> “by assessing”

L128 "with no back-interactions considered" -> "and back-interactions are not considered"

L134 “a finite time” ->unclear since an earthquake occurs in a finite time. Consider rephrasing to something like "a non-instantaneous generation process" or "a generation process that evolves over time".

Table 2 - I do not think Table 2 is necessary. We may replace it by a paragraph. Also Table 2 was wrongly referred to at L284 and L308. They should be Table 3 instead.

Figure 1&2 - Two figures can be combined.

Figure 3 is almost identical to Figure 2&4 of Aiello et al. (2012). I am concerned about the permission to use these figures. The authors should provide confirmation that they have obtained the necessary permissions to reproduce or adapt these figures. 

The manuscript titled "Tsunamigenic potential of unstable masses in the Gulf of Pozzuoli, Campi Flegrei, Italy" provides a numerical study of four submarine (three) and subaerial (one) landslide scenarios and their corresponding tsunamigenic impact in the Gulf of Pozzuoli. The work is looking at past evidence of collapse to reproduce modes of failure and tsunamigenesis with the largest potential attributed to subaerial collapse. The codes UBO-BLOCK and UBO-TSUIMP/JAGURS are used for simulating mass failure and tsunami respectively. The dispersive potential of the wave characteristics is also investigated in the study. The authors conclude that only some of the scenarios examined have an impact on the adjacent coastlines. This is a comprehensive investigative work that can strengthen the existing knowledge of tsunamis in the region. However, some key points need to be addressed.

General comments

1)

P.2 L45-46, P8 L196-198, P17 L355 and elsewhere

Although it is noted that the masses have been selected based on previous deposits and present morphology, it is not clear why these are the worst possible scenarios that could happen in the region. Can the authors affirm this with certainty?  This needs to be carefully addressed as it can be misleading for future policy and hazard mitigation efforts as the current work indicates that there is not a significant risk from submarine but only subaerial landslide tsunamis in the region. Future failures may not replicate past events and an increase in volcanic activity, variance in the location of failure, and higher collapse volumes may increase tsunamigenic potential and the impact may be larger, even more so if the possibility of such events is ruled out.

2)

S2.2

Subaerial landslide tsunamis generally have a greater tsunamigenic potential than submarine ones and have more complexity due to the interactions between the solid mass, water and air.  Modelling the high impact, and complex slide kinematics often requires 3D Navier-Stokes solvers, SPH, CFD or VOF models. It is not clear whether this complexity is captured with the study's numerical codes, an issue which could underestimate the hazard. It is not clarified in the manuscript how the authors distinguish between the modelling of the submarine and subaerial failure, besides the different solvers used in the propagation modelling.

Specific comments:

P.2 L45-46

The worst-case referenced approach is not yet published and therefore hard to understand and verify. The authors should expand more on the methodology.

P.2 L127, 128-130: Air Entrainment is of importance when focusing on subaerial landslides, the generation involves a triple phase interaction.

P3 L139-141. What are the underpinning equations in UBO-BLOCK? The model seems to have been primarily used in the modelling of submarine landslide tsunamis

P6 L167 As JAGURS is nested, it is worth clarifying that the nested approach is omitted in that case.

P6 L167 Please also give more details on whether the simulations were run locally or in a cluster, CPU time and time of the event.

P10 L231-233 This statement should be substantiated by references.

P10 L240-254 Although the authors mention the locations of the deposits and the corresponding volumes it is not clear how these volumes were estimated based on the observed deposits. A few sentences explaining the approach would help here.

P11 L263-265 If any, what kind of rheology is assumed for the sliding mass?

P17 L355 This statement standing alone reads quite strong and I think it cannot be backed up by only assessing 3 scenarios of collapse, it should be better rephrased to the specific case studies as in P19 L408.

Figure 10 I think xlabel is mistyped and '(m)' is wrong?

P20 L427-429 I agree with the authors, and I think this is not clear throughout the manuscript so far.

P22 L472-L479 I believe an important addition to the discussion would be how parameterisation and probabilistic approaches, as for example surrogate models, can help when it comes to future hazard assessments in the region and the capability of assessing multiple scenarios of collapse with large variances in location, volume and mode of failure.