This manuscript presents a valuable and timely contribution to tsunami hazard assessment for the understudied Negros-Sulu megathrust in the Philippines. Given the scarcity of geodetic constraints and the complete absence of this subduction zone from global slab models like Slab 2.0, the authors adopt a pragmatic and well-reasoned approach. By combining structural segmentation, multiple dip angle scenarios (10°, 20°, 30°), and rupture-scaling relationships (Allen & Hayes, 2017), they generate 18 plausible rupture scenarios and model the resulting tsunami propagation using the JAGURS code. The study identifies highly exposed coastal regions (e.g., Sindangan-Labason, Sipalay-Bayawan) and provides critical insights into the roles of wave directivity, coastal morphology, and nearshore bathymetry in controlling tsunami intensity. The findings have direct implications for disaster risk reduction (DRR), including evacuation planning and the targeting of paleotsunami studies. The manuscript is well-organized, the figures are clear and informative, and the discussion of limitations (uniform slip, lack of splay faulting) is honest and constructive. I recommend publication in NHESS after minor revisions. Major Scientific Comments: Uniform Slip Assumption: The authors correctly acknowledge the limitation of using uniform slip (Section 5.2). The reference to misfit analyses (An et al., 2018; Nakata et al., 2019) justifies its use for identifying "highly exposed areas," which is the stated goal. This defense is sufficient. Specific Suggestions for Revision Line 50-55 (Introduction): The mention of the washover deposit in Zamboanga del Sur (Claro et al., 2021) is relevant. Consider briefly stating that this could be a paleotsunami candidate related to the Cotabato or Sulu Trench, tying it back to the need for the exposure maps presented later. Line 165-170 (Methodology):The merging of IFSAR DTM (5m) with GEBCO (15 arcsec) is a good practice. Please specify the resolution of the final merged product used in the simulations. Section 4.1 (Megathrust Source Parameters):The segmentation into NT1-1, NT1-2, NT2, ST, ST1, and ST2 is clearly explained. However, a small schematic cross-section (in addition to Figure 3b) showing the relative positions of these segments along strike (e.g., a simple 2D profile) could help readers visualize the segmentation more intuitively. Section 5.2 (Implications for Tsunami Hazard Potential): The statement regarding vertical evacuation is crucial. Please consider rephrasing slightly to emphasize that while the study identifies where rapid evacuation is needed, site-specific inundation modeling is required to determine if vertical evacuation is feasible and where safe structures should be located. Figure 14 (Exposure Map):This is an excellent summary figure. Consider adding a small inset map showing the 16 coastal regions (I-XVI) for reference, as they are discussed extensively in the text and other figures. Technical Corrections - Line 240 (Table 2): The table formatting in the provided PDF appears slightly misaligned (e.g., "NT1-110° v 20°"). Please ensure the final version is typeset correctly so that the segment names and comparison rows are clearly separated. - Line 425 (References): The reference for GEBCO (GEBCO Compilation Group, 2024) has a placeholder DOI. Please update with the correct DOI. - Figure 5 and 6:The wavefront contours in Figure 6 are very clear. In Figure 5, consider adding a color bar title (e.g., "Wave Height (m)") for absolute clarity, although it is present in the caption.

#

# General points

#

The manuscript "Seismic and Tsunami Hazard Potential of the Negros–Sulu Megathrust, Philippines" develops a suite of highly simplified tsunami scenarios to help us better understand tsunami hazards in the region. This is an important topic and of interest for readers of EGUSphere. 

My main criticism of the manuscript is that the 18 modelled tsunami scenarios are very simplified, and it appears that they all produce waves that are smaller than some historically observed tsunamis produced by smaller magnitude earthquakes. This suggests that the simplicity of the source models is failing to characterise hazards in the region. This is an early study of the region, and I do not expect the models to be perfect. But currently the limitations are not clear enough - readers with an interest in the hazard may be misled. To strengthen the manuscript, I strongly encourage the authors to consider other scenarios with more compact high-slip rupture, which may better represent past events (and thus be more appropriate for hazard assessment). There should also be a clearer discussion of how the results compare with history, to help readers interpret the model uncertainty, which will guide both future work as well as appropriate use of results for hazard assessment.  

Beyond this core issue, there are other issues that will be easier to address:

- Many of the figures use colourschemes that make them hard to see clearly. Please use colourschemes with more variation, which better resolve small values. This includes Figs 5, 9, 10, 11 in the main text, and many figures in the supplementary.

- Although the manuscript is generally well written, I found some parts difficult to understand. 

See the detailed comments for more information.

In my judgement addressing these issues will require major revisions before the manuscript is suitable for publication. But I think this is certainly achievable.

#

# Detailed comments

#

L30: "..that underscore accurate nearshore bathymetry..." -- consider replacing with "that underscore the need for accurate nearshore bathymetry ..."

L31 -- instead of "exposure maps" consider just "maps", since usually 'exposure' implies an analysis of how the built environment could be affected..

L34: In recommending the use of modelling for policy, it will be very important to consider limitations and uncertainties [is the 6-8m peak really enough, given the historical observations?]

L44: Notice these historical earthquakes, with magnitudes in the low 8s, had peak wave heights that exceed the largest modelled scenarios in this study. That suggests that the modelled scenarios are not accounting for the diversity of real earthquake generated tsunamis. Furthermore, globally, it is common for earthquakes with magnitudes in the mid 8s to produce waves much larger than modelled here (e.g. Cheung et al. 2022).

L50-55: More care is needed when discussing interpretations of seismic gaps, coupling and slip deficits. It is not as straightforward as implied. At the global scale, there are sites where geodetic evidence suggests weak coupling even though there is significant evidence for tsunamis (e.g. Witter et al 2016, Debaeker et al 2026). It has been suggested that the neglect of viscoelastic effects in geodetic studies can lead to mistaken conclusions about the readiness of faults to rupture (Wang et al., 2021). Also, the recent Kamchatka 2025 earthquake may have slipped more than the deficit accumulated since the last large event (Liu et al. 2026). Thus, these concepts should not be blindly accepted.

L63: insert "and hydrodynamic complexities such as " before "nonlinear effects"

Figure 1: 

- In panel 'b' it is hard to see the coastline or otherwise locate the pixels. Consider adding the trench locations to make this easier. 

- In panel 'a', what are the velocity vectors measured relative to?

- In panel A, the 'max wave height' results do not all appear to be on the coast. Are they reported at the earthquake location? Make sure the reader knows how to interpret them. 

L77: "where < 1.0  -- suggest to replace with "where smaller values ". Also consider replacing "are associated with a higher proportion of large earthquakes and stress accumulation" with "indicate that large earthquakes contribute more to the long-term seismic moment rate". 

L78: "the evaluated number of declustered seismicity" -- suggest to replace with "the number of earthquakes in the declustered catalogue"

L80-82: Note there will be substantial statistical uncertainty in the GR parameters when using small numbers of earthquakes. There will also be uncertainties about the appropriateness of the model itself (even with many earthquakes). 

Equations 1 - 3: 

- Eq 1 is using the 'unlimited width' Allen and Hayes (2017) regression for rupture area, while also limiting the width to 196 km. Allen and Hayes (2017) provide another regression relation for the case with a capped width, which is their preferred model. Why not use the 'capped width' regression relation, given that widths are capped in this study?

- Equation 2/3 provide both the average slip & the max slip. How is the max slip used to make the earthquake sources? This isn't clear. If both max and average slip are used, there must be some spatial non-uniformity of the hypothetical slip models. That needs to be clearly described. 

- I don't like the use of superscripts as footnotes to describe the standard deviation, because the meaning is not standardised and readers will easily get confused. If the uncertainty in the predictors is used (a very good idea) then why not include terms such as (for sigma = 0.266) " + 0.266 * eps" and state that "eps" has a standard normal distribution. If Mw is predicted from the dimensions (instead of the converse) then provide that equation separately with its own predictive uncertainty. Alternatively, use a table like in Allen and Hayes (2017) to treat both cases. 

General comment on Section 3.1: From the description that has been provided I cannot tell what the source parameters are (length, width, magnitude, mean slip), what the magnitude is, or how the maximum slip is used. Please be precise about *exactly* what was done to construct the scenarios, and remove any extraneous equations. It will help to move Table 1 up to this part of the text. Remove the maximum slip if it isn't actually being used (although I would prefer if the scenarios accounted for possible localised high slip). 

Another general comment on Section 3.1: Please carefully consider whether these source models are realistic, noting historical observations of earthquakes closer to Mw 8 with larger waves that most modelled scenarios (despite the latter having larger magnitudes). This could partly be due to the coarse resolution of the tsunami model at the coast. But I think it's very likely to reflect that uniform slip tsunami sources with average scaling-relation slip sometimes do a poor job of representing real tsunamis. In such cases compact uniform slip models or heterogeneous slip models often perform better (e.g. An et al., 2018; Davies, 2019, 2025). One reason is that often localised patches of high slip ("asperities") contribute disproportionately to the tsunami. 

L150-155: The tsunami model is using low-quality elevation data, and its 15 arcsecond resolution corresponds to cell sizes of roughly 450 m. I think this will be OK to model offshore waves. But 450m resolution is insufficient for mapping inundation, and may miss sites with more localised coastal amplification. This should be mentioned.

L190 -- Both average and maximum slip are presented, but I cannot see how the maximum slip is being used to represent the sources. If it's not used, it shouldn't be mentioned. That said, I would strongly encourage consideration of alternative, more compact scenarios that have higher slip, which may better describe the large waves seen historically for earthquake magnitudes in the low 8s.

Table 2 isn't easy to interpret. For instance, it's hard to know whether the max/min differences reflect a significant part of the model, or just isolated differences. Also the standard deviations I think it would be better to provide figures in the supplementary material for each case, using a colour scheme with more variability near zero, so we can well see variations at both small and large scales.

L318-328: This paragraph mentions how the simple tsunami source representations might cause over-or-under estimation of the tsunami size. I think it needs to be edited to mention how the maximum runups observed in historical events with magnitudes in the low 8s are sometimes larger than the modelled scenarios, clarifying that underestimation is very plausible. 

L329-330: "The use of single-fault models with uniform slip distributions in this study can serve as baseline information in light of the scarcity of geodetic data for finite-fault slip models." -- You don't need geodetic data to construct non-uniform slip models -- there is a huge literature on stochastic approaches -- see e.g. Geist (2002), Geist and Oglesby (2014). I think this statement should be removed.

L330: Here I think the text is misrepresenting the results of An et al (2018). The latter study shows that uniform slip models can work well if they are unusually compact (something also found in Davies (2019, 2025)). They don't provide support to more naive uniform slip models as used in this manuscript, which often underestimate tsunami wave heights (according to the latter studies). 

Figure 15: Please add any historical observations of runup to the plot, along with a label showing the earthquake magnitude. This will help make the model limitations clear, and can be integrated into the limitations discussion. 

#

# References

#

Allen, T. & Hayes, G. P. Alternative Rupture-Scaling Relationships for Subduction Interface and Other Offshore Environments Bulletin of the Seismological Society of America, 2017, 107, 1240-1253

An, C.; Liu, H.; Ren, Z. & Yuan, Y. Prediction of Tsunami Waves by Uniform Slip Models Journal of Geophysical Research: Oceans, American Geophysical Union (AGU), 2018, 123

Cheung, K. F.; Lay, T.; Sun, L. & Yamazaki, Y. Tsunami size variability with rupture depth Nature Geoscience, Springer Science and Business Media LLC, 2022, 15, 33-36

Davies, G. Tsunami variability from uncalibrated stochastic earthquake models: tests against deep ocean observations 2006-2016 Geophysical Journal International, 2019, 218, 1939-1960 

Davies, G. Tsunami Variability From Stochastic Earthquake Models: Tests Against Fourteen Tsunamis at Australian Tide Gauges Journal of Geophysical Research: Solid Earth, 2025, 130, e2025JB031949

Debaecker, S.; Feuillet, N.; Satake, K.; Sowa, K.; Yamada, M.; Sato, T.; Nakamura, M.; Watanabe, A.; Saiki, A.; Saurel, J.-M.; Occhipinti, G.; Yu, T.-L. & Shen, C.-C. Evidence of megathrust earthquakes and seismic supercycles in subtropical Japan from millennia-old coral microatolls Nature Communications, Springer Science and Business Media LLC, 2026, 17

Geist, E. Complex earthquake rupture and local tsunamis. Journal of Geophysical Research, 2002, 107

Geist, E. L. & Oglesby, D. D. Tsunamis: Stochastic Models of Occurrence and Generation Mechanisms Encyclopedia of Complexity and Systems Science, Springer New York, 2014, 1-29

Liu, C.; Bai, Y.; Lay, T.; He, P.; Wen, Y.; Xiong, X. & Taymaz, T. Simple unilateral rupture of the great Mw 8.8 2025 Kamchatka earthquake Science, American Association for the Advancement of Science (AAAS), 2026, 391, 812–817

Wang, K.; Zhu, Y.; Nissen, E. & Shen, Z.-K. On the Relevance of Geodetic Deformation Rates to Earthquake Potential Geophysical Research Letters, American Geophysical Union (AGU), 2021, 48

Witter, R. C.; Carver, G. A.; Briggs, R. W.; Gelfenbaum, G.; Koehler, R. D.; La Selle, S.; Bender, A. M.; Engelhart, S. E.; Hemphill-Haley, E. & Hill, T. D. Unusually large tsunamis frequent a currently creeping part of the Aleutian megathrust grl, 2016, 43, 76-84