the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 21 Nov 2025
A new framework for the assessment of potential future disasters caused by typhoons using multi-model ensemble experiments
Abstract. Assessing future changes in typhoon-related hazards is essential for climate adaptation. We develop an event-based storyline framework coupling high-resolution meteorological simulations with river discharge and storm-surge models to quantify future risks under specified warming levels. Using Super Typhoon Hagibis (2019) as a case study, we conduct multi-model, multi-initial-condition ensemble experiments to account for structural model uncertainty and internal variability. Three meteorological models are run with 27-member initial-condition ensembles to drive three river models and two storm-surge models. Experiments cover the present climate and two warming scenarios (+2 K, +4 K). Across ensembles, Hagibis intensifies under warming: precipitation and near-surface winds strengthen, along with lower central pressure relative to present-climate runs. Consistent with these changes, many river basins in eastern Japan exhibit increased discharge, amplifying flood risk, and coastal models indicate larger storm surges. For Typhoon Hagibis, we quantify uncertainties in peak discharge and maximum storm-surge level. The multi-model combination widens uncertainties by sampling structural model differences, thereby spanning a wider range of plausible outcomes. Thus, our multi-model, multi-initial-condition framework provides a more comprehensive assessment of future typhoon-related risks than single-model experiments.
- Preprint
(2084 KB) - Metadata XML
-
Supplement
(1293 KB) - BibTeX
- EndNote
Status: final response (author comments only)
- RC1: 'Comment on egusphere-2025-4772', Anonymous Referee #1, 29 Dec 2025
-
RC2: 'Comment on egusphere-2025-4772', Anonymous Referee #2, 05 Jan 2026
The objective of this paper is to develop a quantitative framework for assessing future typhoon-related hazards under climate change. The study uses Super Typhoon Hagibis (2019) as a case study and adopts an event-based storyline approach, integrating high-resolution atmospheric, hydrological, and storm surge models. It employs multi-model, multi-initial-condition ensemble experiments to represent both structural and internal uncertainties. Future hazard changes are evaluated under +2 K and +4 K warming scenarios, aiming to provide a comprehensive risk assessment.
Despite the richness of the data and the ensemble framework employed, the manuscript lacks clarity, focus, and sufficient scholarly rigor to warrant publication in its current form. Significant revisions are required to address the key structural, logical, and academic issues outlined below.
Major Comments
- Lack of logical coherence
While the manuscript introduces a robust and ambitious modeling framework, the logical flow connecting the research objectives, methodology, and conclusions is often unclear. Transitions between sections are weak, and key modeling decisions (e.g., the selection of the case study, ensemble member pruning) are insufficiently explained, making it difficult for the reader to follow the scientific reasoning. - Excessively lengthy and redundant descriptions
The manuscript contains lengthy and repetitive descriptions that obscure the core message. For example, Sections 1 and 2 contain overlapping discussions on the benefits of high-resolution modeling and ensemble approaches. The text would benefit greatly from streamlining to improve readability and clarity. - Insufficient review of prior literature
Although a few relevant studies are cited (e.g., Kanada et al., 2021; Takayabu et al., 2015), the literature review is shallow and fails to clearly position the current work within the broader research context. Notably, several recent studies that use storyline or ensemble methods for typhoon risk assessment are omitted or only briefly mentioned.
Minor Comments
- Undefined abbreviations
Abbreviations such as NHRCM and GSM appear without proper definition (e.g., in Sections 2.1–2.2). These should be clearly defined upon first use for reader comprehension. - Lack of quantitative definitions
Key terms like "high resolution" and "super typhoon" are not quantitatively defined. The absence of such definitions reduces scientific precision. For example:
- What grid spacing constitutes “high resolution” (e.g., ≤ 60km)?
- What is the threshold wind speed or classification for a “super typhoon”?
- Citation formatting errors
Some references, such as in paragraph 40, do not follow standard citation formats. These should be carefully reviewed and corrected according to journal guidelines.- Misleading section title: “Discussion and Conclusion”
In paragraph 260, the section is titled “Discussion and Conclusion,” yet no substantial discussion is presented. A proper Discussion section is expected to interpret findings, compare them with previous research, and examine implications or limitations. Currently, the section functions as a conclusion only.- Inaccurate use of the term “Multi-model”
In paragraph 265, the authors refer to their use of meteorological, hydrological, and coastal models as a “multi-model” approach. However, since these models serve different functions, the more appropriate term would be “coupled modeling framework.” The term “multi-model” generally applies to the use of multiple models serving the same purpose to capture structural uncertainty.Citation: https://doi.org/10.5194/egusphere-2025-4772-RC2 - Lack of logical coherence
-
RC3: 'Comment on egusphere-2025-4772', Anonymous Referee #3, 05 Jan 2026
General comments
This study proposes a framework for assessing the potential risks of climate change induced by river discharge and storm surges using multiple numerical models and the Pseudo Global Warming (PGW) method. The topic is timely, as the results can be applied to evaluate compound floodings near coastal areas and provide insights into disaster mitigation. However, the manuscript has several drawbacks that the authors should address. Please refer to the following comments for further revision. Also, I recommend using a professional editorial service to improve the writing quality, as there are several grammatical errors and typos.Specific comments
Line 106: The computational settings across the three models appear to differ significantly. For instance, the NHRCM model utilizes only the cumulus scheme, whereas the CReSS model utilizes a slab ocean model without applying spectral nudging. These differences may over-emphasize the variability of the CPMs, potentially undermining the robustness of future projections. The authors should explain and justify the rationale behind these settings in the main text.Line 108: It is necessary to describe the criteria for selecting the 5 cases from the 27 ensembles. I also recommend indicating the selected TCs in Fig. 3a to improve clarity.
Line 114: The exact timing used to calculate the differences in SST and air temperature should be described. For instance, what specific period does “future” refer to? In addition, d4PDF does not include results for 2023, which seems inconsistent with the statement in Line 270. Please correct this or add an explanation if necessary.
Section 2.4.2: The bulk formula used to calculate the sea drag coefficient and the treatment of the wind speed threshold should be described. Additionally, a description of the modeled bottom stress is essential for storm surge simulations and should be included.
Line 164: The author should explain whether the lack of data pertains to all selected analysis sites, or if there are other contributing factors. Moreover, given that river discharge is generally measured at multiple stations along the target rivers, it is necessary to clarify whether these locations also experienced data gaps or if the data quality was too poor to be utilized in this study.
Lines 174-175: The spread and Fig. 3a indicate that TC tracks from some ensemble members were not closely aligned with observations near Tokyo Bay. The phrase “model results closely follow” should be revised to accurately reflect the simulated results.
Line 184: Even if the translation speeds of the TCs differ among ensemble members, the authors should verify the accuracy of the time series for precipitations, sea level pressure, wind speed and wind direction, as these directly affect the temporal evolution of river discharges and storm surges. At least, this comparison should be conducted for the five selected TCs.
Figure 4: The simulated central pressures suggest that NHRCM (CReSS) tend to simulate the highest (lowest) wind speeds according to the gradient wind balance; however, the simulated maximum wind speeds show the opposite trend. This is counter-intuitive. The authors should elaborate on the possible reason for this gap.
Lines 196-200: There appear to be no results in Fig. 4 that correspond to the statistics (e.g., 10.9 hPa in line 196) mentioned in the text. Please revise the figure or the text to address this inconsistency.
Figure 7-1: I recommend showing the storm surge waveform with the highest accuracy (i.e., the lowest RMSE) among the ensemble members rather than the mean values. Averaging storm surges may obscure local peaks due to phase differences. Indeed, the mean lines do not capture the seiche observed around 10-12-19, which is typical for Tokyo Bay.
Figure 7-2: The mean lines indicate that the storm surge models underestimated the observed tidal residuals despite the tendency to overestimate TC intensities in the CPMs. This is also a counter-intuitive result. Please investigate and elaborate on the reason for this.
Figure 8: The y-axis should show the ratio rather than the maximum surge levels, as the caption indicates that the same definition used in Fig. 6 applies to both the bars and error bars.
Technical comments
Line 38: Avoid using the word “like,” as it makes the research targets (e.g., typhoons and heavy precipitation) sound ambiguous.Line 40: Please provide a more specific description of “inner variability.” Does this refer to uncertainty in the meso-scale system arising from natural variability rather than model differences?
Lines 43-44: The phrase “SOLA” in the references is unnecessary and should be deleted.
Line 53: Rephrasing “integration” to “forecasting” would improve the clarity of the paper.
Line 70: The phrase “natural variability” is inappropriate for weekly ensemble simulations, as it typically refers to interannual to decadal scales.
Line 71: The meaning of “Both trials” is unclear. Please clarify.
Line 78: The phrase “wave height” usually refers to the amplitude of wind waves or swells, not storm surges. Please correct the terminology accordingly.
Line 102: I recommend revising the description of the DDS procedure for the GCM. The current text does not clearly explain the sequential downscaling steps through domains with different grid sizes.
Line 104: The meaning of “it” is unclear, making the sentence difficult to understand. Please provide a more detailed description of the DDS procedure.
Line 129: It is preferable to state the spatial resolution of CaMa-Flood in the main text.
Line 237: Please add the name of the tide station to the manuscript.
Figure 3: Please add “(a)” immediately after the figure number.
Table S2: Delete the unnecessary question mark in the table. Additionally, please cite the reference related to the J-FlwDir data.
Citation: https://doi.org/10.5194/egusphere-2025-4772-RC3
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 239 | 75 | 19 | 333 | 32 | 13 | 14 |
- HTML: 239
- PDF: 75
- XML: 19
- Total: 333
- Supplement: 32
- BibTeX: 13
- EndNote: 14
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Overall Assessment
The manuscript presents a framework for assessing potential future typhoon‑induced disasters by coupling high‑resolution meteorological simulations with river discharge and storm‑surge models, applied to Super Typhoon Hagibis (2019) as a case study. The framework combines three typhoon models, three river models, and two storm‑surge models with multi–initial‑condition ensembles to represent structural model uncertainty and internal variability. The results suggest intensification of the typhoon under warming, with increased precipitation, stronger near‑surface winds, and lower central pressure, accompanied by higher river discharges in many eastern‑Japan basins and larger storm surges. While the framework is conceptually valuable and the experiment design is ambitious, in its current form the manuscript does not yet meet the requirements for publication. The concerns outlined below, particularly regarding robustness, interpretation, and implementation of the framework, should be addressed.
General Comments
1. The paper primarily focuses on Super Typhoon Hagibis (2019) as a case study, which may limit generalizing the findings to other typhoons or extreme weather events. Further research is needed to validate the framework across multiple events and regions. How well do the results generalize to other typhoons or extreme weather events beyond Super Typhoon Hagibis (2019)? Have you tested the framework on other events, and if so, what were the findings?
2. The framework proposed in this study assumes a linear coupling of the GSM data and the difference component of the climate change of air temperature and sea surface temperature between the +2 or +4 K future and 2023. This assumption may oversimplify the actual climate change impacts and should be further validated. How does this assumption impact the accuracy of the results, and are there alternative methods or models that could be used for a more sophisticated coupling?
3. The storm surge and hydrological models are used separately rather than within a fully coupled framework, which restricts the representation of dynamical interactions within the typhoon-rainfall-runoff-surge system. In reality, storm structure, precipitation, river discharge, and coastal water levels evolve in a tightly linked way, and this coupling is not explicitly represented when the two impact models are run independently. It also remains unclear how the water levels (or flood indicators) produced by the two models are combined; flooding hazards should be assessed using both sources of information with particular care, especially because coastal surge levels are influenced not only by meteorological forcing but also by river inflow at river mouths.
4. When comparing Fig. 4(b) with Fig. S2(b), the maximum wind speed does not increase under warming for the NHRCM and WRF simulations, in contrast to CReSS and to the warming‑induced intensification indicated by changes in central pressure and precipitation. Consequently, the authors’ statement that “all three of these results indicate the strengthening of typhoons due to global warming, suggesting that storm surges and flooding will also become more hazardous accordingly” (lines 200–201) is not fully justified as written. This inconsistency is counter‑intuitive and potentially important, yet it is not explored or discussed in sufficient detail in the current version of the paper.
5. The paper does not extensively discuss the computational requirements and potential limitations of the framework, such as the resource-intensive nature of running multiple models and ensemble experiments. What are the computational requirements for running the framework, and how do they scale with the number of models and ensemble members? Are there strategies to optimize the computational efficiency of the framework?
Specific Comments