the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 25 Nov 2025
Equatorial storm surge risks revealed by the 2001 tropical cyclone Vamei
Abstract. Tropical Cyclone Vamei, which emerged at 1.5° N on December 27, 2001, challenged the prevailing idea that near-equatorial areas are safe from storm surges, as it resulted in localised flooding in Singapore and Malaysia, revealing a rare yet critical regional hazard. Thus, we investigated storm surge risk assessment in Singapore based on numerical simulations using Delft3D in multiple scenarios. We first validated the accuracy of our simulation results by comparing them with nine observation points around Singapore Island. We then conducted the simulations as a suite of alternative scenarios created by moving the known track of Tropical Cyclone Vamei, modelling more intense storms corresponding to a 1-in-1000-year scenario and considering future sea level rise induced by global warming. When a 1000-year probability of occurrence was assumed, the maximum storm surge height around Singapore increased to 0.595 m. For a 1000-year cyclone with its path shifted 0.8° southward, sea level rise scenarios of +0.7 m and +2.0 m resulted in inundation areas of 34.5 km² and 90.7 km², respectively. While the calculated storm surge height remained largely unchanged despite future sea level rise, the inundation area in Singapore expanded significantly. This indicates that sea level rise is a primary contributor to this expansion, highlighting the importance of considering future sea levels in inundation assessments. Further research is necessary to assess potential changes in the frequency and intensity of tropical cyclones impacting Singapore under future climate scenarios.
- Preprint
(13339 KB) - Metadata XML
- BibTeX
- EndNote
Status: open (until 16 Feb 2026)
- RC1: 'Comment on egusphere-2025-5703', Anonymous Referee #1, 22 Dec 2025 reply
-
RC2: 'Comment on egusphere-2025-5703', Anonymous Referee #2, 20 Jan 2026
reply
General comments
Typhoons passing near the equator and the damage they cause are rare, and the cases examined in this study are therefore interesting. However, as a scientific paper, there are major problems as outlined below, which may prevent readers from achieving a sufficient understanding of the study.
This study aims to evaluate storm surge risk with a 1,000-year return period; however, what is actually assessed is storm surge generated by a typhoon whose intensity corresponds to a 1,000-year return period for low-latitude typhoons. The method used to estimate this intensity is not sufficiently explained. Storm surge is determined not only by typhoon intensity but also by factors such as track, size, and translation speed, the effects of which cannot be ignored. Therefore, these two concepts (1,000-year storm surge and 1,000-year typhoon) are not equivalent. A proper evaluation of this distinction is required, and factors that are not accounted for should be clearly stated.
Although tides and waves are included in the analysis, it is questionable whether their interactions need to be considered. In the water-level changes used for validation, the majority of the temporal variation is dominated by tides, making it difficult to adequately verify the reproducibility of storm surge–induced sea level anomalies. If interaction with tides is essential, comparisons with and without tidal forcing should be conducted. As it stands, figures such as Figures 4, 5, and 8 make it difficult to distinguish the reproducibility and influence of storm surge components. Wave analysis results are shown in Figure 5, but similarly, their contribution is not discussed. There is insufficient explanation regarding how much high waves contribute to water level rise or whether overtopping analysis was conducted in the inundation simulations. Consequently, the reliability of the hindcast simulations cannot be properly evaluated.
Although some relationship can be observed between the radius of maximum wind and the central pressure (or maximum wind speed) of a typhoon, in practice there is considerable variability, and accurate estimation should be difficult. This issue is not addressed, and the validation is concluded by attributing inaccuracies to uncertainty in the typhoon track without presenting verification of temporal changes in wind speed, wind direction, or pressure. This approach is not appropriate.
Overall, the explanations of the methodology and results are insufficient, making the content difficult to understand. Furthermore, much of what is discussed in the Discussion section does not correspond to the results of the present simulations and is therefore inappropriate.
Specific comments
- (L44) There is no explanation of the specific approach for the following methods. ‘Downward counterfactual analysis’. Please provide a concise explanation and significance, not just a list of references.
- (L105) It is reported that significant flooding and mudslides have occurred, but supplementary explanation is needed regarding whether this was caused by rainfall or strong winds (storm surge), what kind of rainfall characteristics did the typhoon have? and where the damage occurred.
- (L112) Empirical rules of the radius of maximum winds (Quiring et al., 2011) are likely to exhibit considerable variation. While acceptable for simulations under hypothetical conditions, I cannot endorse their use for evaluating the accuracy of follow-up calculations. It is also necessary to verify the reproducibility of the wind field time series.
- (Table 1) If the definition of wind speed differs, it should be noted in the table's footnotes or converted to a comparable value.
- (L125) Generally, remnants of typhoons can sometimes affect storm surges. Particularly in cases like this one where the wind-driven effect is relatively small, the swinging back effect could potentially cause the second wave to become larger. An explanation is needed as to whether the current conditions are sufficient.
- (L156-L161) It is unclear what this paragraph is explaining. What is assumed the maximum wind speed 45 knots? If it refers to Table 1, it might be Typhoon Vamei, but why is it appropriate to adopt Vamei as an example? Please outline your reasoning process. How many typhoon cases are ultimately used for GPD parameter estimation?
- (L168) What exactly are large-scale parameters?
- (L163-L172) This method is insufficient to determine the target typhoons. Storm surges depend not only on a typhoon's intensity but also on its path, size, and translation speed. Crucially, whether the typhoon of certain intensity passes over the target area is also important. This problem should be explicitly stated.
- (L174-L179) I do not understand the meaning of this explanation. While an explanation of how Chang et al. made their estimation is also necessary, even considering that, it remains unclear why the recurrence interval for Typhoon Vamei can be determined as 250 years despite the availability of observational data, and how this aligns with creating a cumulative distribution from past typhoon data.
- While it is possible to estimate worst-case scenarios, I believe risk assessments incorporating probability are insufficient. I request explanation specifying what assumptions are required.
- (L201-202) To consider diverse storm surges, we must also account for the effects of their progression speed and scale. While typhoon intensity and scale are assumed one-to-one correspondences in this case, but actually there is considerable variation.
- (Table 2, Maximum storm surge height around Singapore) If it is the maximum value in region D3, the location where it occurs differs in each case. Shouldn't we use a comparable metric?
- (Table 3) Although the unit is not specified, a tidal deviation of around 10 cm would be considered quite small. Even considering only the static suction effect from the central pressure, an increase of about 5 cm would be expected. That is, even accounting for estimation errors in the central pressure, the effect of wind-driven surges appears to be quite small. This characteristic is not reflected in the discussion of the results. Interpreting everything as merely a matter of typhoon trajectory data is far too simplistic.
- (L225) ‘Thus, high storm surges were produced by higher wind velocity and lower pressure.’ Supporting data is required for comparisons of wind speed and air pressure, as well as the relationship between the closest approach distance and the radius of maximum wind speed.
- (L228-229) ‘Thus, the simulation accuracy depended on the accuracy of observation of the tropical cyclone’s track, minimum central pressure, and maximum wind speed.’ More specifically, a qualitative explanation is needed as to why the underestimation and overestimation occurred. First, regarding the wind field, the reproducibility has not been demonstrated.
- (Figure 4) Since Figure 4 should compare the reproducibility of sea level anomaly, please compare the results after applying a high-pass filter to both.
- (L234-236) If tidal calculations fall outside the scope of this research, there is no need to perform analyses incorporating tides, and the research direction becomes unclear. Even if the analysis includes tides, verification of the tidal component and verification of tidal level deviation must be conducted separately.
- (L253) ‘The maximum wind speed is 25 m/s in domain 1.’ How does this value correspond to the one shown in Table 1?
- (L254-256) ‘The maximum significant wave height is 5.86 m in Domain 1, while it was 4.3 m in Domain 3. The maximum significant wave height in domain 1 was calculated in southern Singapore because waves penetrated from the southeastward of Singapore, which faces the Pacific Ocean.’ I do not understand the necessity of presenting analysis results for waves. Was the contribution of wave-induced water level rise significant in this analysis? Are you analyzing wave overtopping and inundation? Without any explanation, I consider it unnecessary within the paper's structure.
- (Figure 5) Without understanding the temporal variations in the wind field, including wind direction, it is difficult to comprehend the phenomenon. Water levels also contain tidal components, making it nearly impossible to discern what to observe. The results for c and f do not appear to correspond.
- (L269) ‘The typhoon's maximum wind speed of 1000-year probability of occurrence is 84 knots’ How do you read the 1000-year probability value from this diagram?
- (L273) ‘Based on the relation proposed by Quiring et al. (2011), the radius of maximum winds of Typhoon Vamei was 76.4 km.’ There is no guarantee that this is the actual maximum wind radius of Typhoon Vamei.
- (L274) ‘The radius of maximum winds of a 1000-year typhoon is 54.7 km’ There is no guarantee that this assumption is correct. If the radius of maximum wind speed changes, the results will also change significantly.
- (L305-321) What is the purpose of this section? Its content is unrelated to the research findings.
- (L328-330) ‘Therefore, a sea level anomaly of the same size as Tropical Cyclone Vamei occurred in Singapore, aside from tropical cyclones. However, in this study, even when sea level rises were assumed, maximum water levels around Singapore remained the same (Table 2).’ I don't understand what you mean or which results you're referring to. Please explain more specifically.
- (L332-335) ‘In the case of our simulation, the intensity of storm surges around Singapore did not significantly change, so even if sea level rise were generated in the future, the intensity of storm surges would not significantly increase around Singapore.’ What exactly is meant by storm surge intensity? I don't understand what we're discussing. Please explain the hypothesis that storm surge intensity changes due to sea level rise.
- (L338-341) The scope of the discussion appears overly broad and unrelated. Please focus on topics directly relevant to the findings of this research.
- (L369-371) ‘When the path of Vamei is moved southward, the maximum storm surge height around Singapore increases from 0.127 m to 0.183 m, and inundation increases from 8.06 km2 to 8.83 km2.’ There is no explanation of the mechanism. It merely states the numerical results.
Technical corrections
- (Figure 1) The letters SS and BK are hard to read in Figure 1. Is the station RL not included in D3? Some additional explanation is needed.
- (L112 and L450) Typo concerning author name of reference, Fujii and Mitsuta (1986).
- (Table 1) There is an error in the table number.
- (L130 )An explanation of what the abbreviation stands for is required. ‘MPA’
- (L139-L141) The similar expressions appear repeatedly, which I believe constitutes redundant phrasing. ‘We achieved this by considering the return period using the generalised Pareto distribution (GPD). The generalised Pareto distribution (GPD) (e.g., Coles 2001; McInnes et al. 2016) was employed for the typhoon intensity.’
- (L183-L187) The same content is stated at the beginning of the paper, and I believe this constitutes redundant expression.
- (L197) An explanation of what the abbreviation stands for is required. ‘CCRS’
- (L221, the maximum water level rise) The term “maximum tidal deviation” is more appropriate.
- (Table 3) The unit of water level is not shown.
- (L258) ‘… Tropical Cyclone Vamei (Fig. 1)’ Is the referenced figure correct?
- (L261) ‘… especially west of Tekong Island (Fig. 5f).’ It is difficult to interpret the results from the figure.
Citation: https://doi.org/10.5194/egusphere-2025-5703-RC2 -
RC3: 'Comment on egusphere-2025-5703', Giovanni Scardino, 20 Jan 2026
reply
This work is very interesting and considers an important topic for coastal hazard. Here, a modelling approach was followed to simulate the impact of cyclone Vamei in Singapore, also in a framework of future sea-level rise. The methodology and results are well explained and support the increase of coastal flooding related to the joint action between storm surge and sea-level rise. I have just few comments that can be considered to improve the manuscript.
Minor comments
Line 19: In this sentence “For a 1000-year cyclone with its path shifted 0.8° southward, sea level rise scenarios of +0.7 m and +2.0 m resulted in inundation areas of 34.5 km²….” it is not clear why the cyclone track is shifted. Maybe a brief sentence could help the reader to better understand the dynamic of southward shift.
Line 90: rephrase like this: We used domain decomposition to simulate the area around Singapore using a fine computational grid.
Line 91: It is not clear if you applied a nesting for all the spatial domains (D1, D2, D3). Please highlight this aspect.
Lines 105-108: This sentence is almost a discussion. Maybe it could be better to move into the Section 5.
Lines 123-124: When you are reporting the tropical cyclone speed, do you mean the maximum wind speed?
Lines 182-183: This sentence represents a discussion.
Line 188: When modifying storm variables such as size and intensity, are you accounting for associated changes in thermodynamic fields (e.g., air temperature, sea surface temperature, specific humidity)? If so, please include a brief explanation of how your modifications are consistent with these thermodynamic changes.
Line 197: Are you sure that CCRS projections are undated?
Line 198-199: Consider to correct the sea-level rise with the value of vertical land movements of Singapore area (for example, see 10.1109/JSTARS.2025.3573299).
Table 2: How did you assess the inundated areas, report a description in the methodology. Then I suggest to insert also the intensity variables that you have changed, such us central pressure and maximum wind speed.
Fig. 5: Please also insert some toponyms in the maps, because you cited the island names and cities in the text.
Section 5 Discussion needs to be improved. Although tsunami contributions to the study areas have been documented in the literature, this study is limited to the simulation of tropical cyclones, a phenomenon that differs fundamentally in its mechanism of water level change. So I suggest to avoid a discussion with tsunami flooding compared to your cyclone simulation.
Citation: https://doi.org/10.5194/egusphere-2025-5703-RC3 -
CC1: 'Comment on egusphere-2025-5703', Xingkun XU, 22 Jan 2026
reply
Publisher’s note: this comment is a copy of RC4 and its content was therefore removed on 22 January 2026.
Citation: https://doi.org/10.5194/egusphere-2025-5703-CC1 -
RC4: 'Comment on egusphere-2025-5703', XINGKUN XU, 22 Jan 2026
reply
This manuscript assesses storm surge and coastal inundation hazards for Singapore using Tropical Cyclone Vamei (27 Dec 2001) as a baseline and a “downward counterfactual” framework to explore hypothetical scenarios (southward track shifts, intensification to a 1-in-1000-year cyclone, and sea-level rise). Delft3D-FLOW coupled with SWAN is employed to simulate water level, currents, and wave effects through radiation-stress feedback (wave setup). Modelled water levels are compared against tide-gauge observations at nine locations, and scenario impacts are summarized using peak storm surge heights and inundation areas. The paper is potentially valuable, but the reliability of the headline results depends strongly on assumptions that are currently under-documented (e.g., pressure–wind conversion, extreme-value extrapolation and its uncertainty, wave contribution to water level, and inundation derivation).
- The manuscript establishes a relationship between maximum wind speed and minimum central pressure using the collocated JMA dataset over 0–15°N and 100–125°E (Fig. 7). Because the subsequent “1000-year cyclone” construction and wind-field forcing depend strongly on both and , the results can be highly sensitive to the chosen conversion (and to scatter in Fig. 7). I strongly recommend a sensitivity test or uncertainty propagation。
- If I understood correctly, the model is run with two-way Delft3D-FLOW–SWAN coupling, so waves can affect water level through wave setup. Since the surge magnitudes at some tide gauges are relatively small (~0.1–0.2 m), even a modest wave-setup contribution could matter. Could the authors quantify this more directly? A clean way is to run a sensitivity test with SWAN coupling switched off (FLOW-only) and compare the tide-gauge water levels (peak, timing, RMSE/bias). This would clarify whether waves are a second-order detail here, or whether they noticeably affect the conclusions.
- The paper says surge is extracted using a high-pass filter, but the filter settings aren’t described. With small surge signals, different filter choices can shift the peak or change its magnitude. Please report the filter design clearly. Ideally, also show a quick robustness check (e.g., two reasonable cutoff choices) to demonstrate that the peak surge is not an artefact of the specific filter settings.
- Inundation area (km²) is a key outcome, yet the manuscript does not clearly document whether inundation is computed dynamically (wetting/drying in FLOW) or via static post-processing against a DEM/DTM, nor how vertical datums and coastal defenses/connectedness are handled.
- 5 is hard to read because the colour scale looks too compressed. I can’t easily tell spatial differences or infer values. I suggest adjusting the colour limits / tick marks (or using discrete levels) so the spatial gradients become more interpretable.
- Sections 5.1 and 5.3 read more like Introduction material than discussion of this study’s results. I suggest tightening them and keeping the Discussion closer to what is actually shown/diagnosed in the paper. 5.1: the equatorial cyclone genesis discussion is interesting, but unless the paper presents results directly related to genesis, it may fit better in the Introduction or as brief context. 5.3: similarly, if cold surges / Borneo vortex are not directly diagnosed in the analysis (no supporting diagnostics shown), they should be framed more cautiously as plausible mechanisms or limitations rather than as confirmed drivers.
- Table 2 and line 330 suggest that peak water levels do not change under the SLR scenarios. Could the authors explain why?
Minor
L125–127: Simulation period says “in January” although the event is Dec 27, 2001; likely should be December
L110–113 + Fig.2 caption: Text cites Holland (1980) parametric model, while the caption states Holland (2010). Please align and specify which formulation is used.
L95–99: Manning’s is set uniformly to 0.025 across beaches, seafloor, and paved surfaces. This is a strong simplification; please justify or clarify whether any spatially varying roughness was tested.
L145–170: Please clarify whether y is per-storm maximum wind, and how independence is ensured for POT sampling.
L255–260: Statement that waves penetrated from “southeastward of Singapore, which faces the Pacific Ocean” is geographically incorrect and should be revised (Singapore is sheltered by surrounding seas/straits; “Pacific Ocean” is misleading here).
Fig.7: y-axis says “Minumum central pressure (Pa)” but plotted values (~880–1020) indicate hPa; also spelling “Minumum” → “Minimum”.
In Fig. 4(e), the JTWC-forced simulation (blue) shows a small double-peaked feature during the first ~2–3 h that is not present in the observations and is much weaker in the JMA-forced run. Any reasons?
Line 155: Could the authors please check that the definition of “maximum wind” is consistent between the model forcing and the JMA best-track data (e.g., averaging period, reference height, and whether it is sustained wind vs gust)?
Citation: https://doi.org/10.5194/egusphere-2025-5703-RC4
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 4,158 | 112 | 34 | 4,304 | 23 | 15 |
- HTML: 4,158
- PDF: 112
- XML: 34
- Total: 4,304
- BibTeX: 23
- EndNote: 15
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Watanabe et al have presented the results from Delft-3D simulations of the 2001 cyclone Vamei and additional scenarios based on the cyclone path and parameters. Using the scenarios, they have estimated inundation areas in Singapore. This research illustrates the importance of modeling extreme events in order to better prepare urban centers from expected damages.
The manuscript is generally clear and appropriately succinct. I have listed some comments below for further improvement and clarifications regarding the methodology and conclusions: