Preprints
https://doi.org/10.5194/egusphere-2025-1938
© Author(s) 2025. This work is distributed under
the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
https://doi.org/10.5194/egusphere-2025-1938
© Author(s) 2025. This work is distributed under
the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
12 May 2025
| 12 May 2025
Large discrepancies between event- and response-based compound flood hazard estimates
Abstract. Most flood hazard assessments follow the event-based approach, assuming that the probability of flooding approximates the probability of flood drivers. However, this approach neglects information about the temporal and spatial variability of flood drivers and flood processes such as water propagation inland and its interaction with topography. The response-based approach accounts for these factors by using a large number of flood events that allow the calculation of flood probabilities. Here, we compare differences in flood hazards between the event- and response-based approaches for a case study in Gloucester City (NJ, U.S.). We find that compound events with return periods less than 20 years can produce the 100-year (i.e., 1 % annual exceedance probability) flood depths in large areas of the city. This is caused by the temporal and spatial characteristics of these events, such as prolonged high coastal water levels and rainfall fields with higher rainfall rates over urbanized areas. These event characteristics are not included in extreme value models of the flood drivers and are commonly simplified by using a single design event. However, flood hazards largely depend on them, introducing large discrepancies in resulting flood hazards if neglected. The temporal and spatial variabilities of flood drivers need to be incorporated in flood hazard assessments to produce robust estimates.
How to cite. Santamaria-Aguilar, S., Maduwantha, P., Enriquez, A. R., and Wahl, T.: Large discrepancies between event- and response-based compound flood hazard estimates, EGUsphere [preprint], https://doi.org/10.5194/egusphere-2025-1938, 2025.
Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.
Download & links
- Preprint
(1375 KB) - Metadata XML
-
Supplement
(557 KB) - BibTeX
- EndNote
Status: final response (author comments only)
Comment types: AC – author | RC – referee | CC – community | EC – editor | CEC – chief editor
| : Report abuse
-
RC1: 'Comment on egusphere-2025-1938', Anonymous Referee #1, 26 Jun 2025
-
AC1: 'Reply on RC1', Sara Santamaria, 23 Aug 2025
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2025/egusphere-2025-1938/egusphere-2025-1938-AC1-supplement.pdf
-
AC1: 'Reply on RC1', Sara Santamaria, 23 Aug 2025
-
RC2: 'Comment on egusphere-2025-1938', Anonymous Referee #2, 30 Jun 2025
This paper integrates synthetic storm generation (rainfall and storm surge) with a reduced-complexity hydrodynamic model to assess the impact of between-storm variability in flood hazard estimates. The authors find that using a single nominal design storm can provide a very limited understanding of flood hazard or risk, suggesting important directions for future work.Overall, this paper makes a valuable and substantial contribution to the field of coastal flood risk assessment by highlighting the limitations of traditional design storm approaches and advocating for the use of large ensembles and probabilistic methods. However, the paper needs significant revisions to clarify its contributions and limitations, and to improve its readability.My primary concern is that the paper "buries the lede" by spending a lot of time on the specific test case and model development, which detracts from the main contribution of the work. The key findings about the limitations of design storm approaches and the benefits of large ensembles and probabilistic assessments are somewhat obscured by the detailed discussion of the modeling setup and validation. Fundamentally, I don't think the conclusions of the paper are particularly sensitive to specific details about how well-validated the SFINCS model for this watershed is (although there are of course many questions for which that would be important!) I'd suggest moving some of the testbed results and figures to an appendix or supplementary material.A related concern is about the level of polish of the figures. Improving the clarity and presentation of the figures would greatly enhance the overall readability and impact of the paper. Just to pick the first figure as an example, fig 1 spends a large amount of space on the x and y axis labels, wasting space, is very low-resolution, lacks (a) (b) (c) markings, and the use of satellite imagery in the larger figure could more usefully be topography or land use. Similar critique applies to most of the figures in the paper. For another example, I found the number of different vertical lines with different colors and line styles in fig 3 to be confusing and distracting.A third concern is about the discussion of probabilities in this paper, which I similarly find to be confusing and distracting to the reader. As an example, figure 6 talks about the "most likely 100 year event." This is particularly confusing, as a standard interpretation of the "100 year event" in a bivariate context is that there is a 1% chance of a draw from the bivariate distribution landing "outside the curve". While some definition is provided, it is confusing and I don't think that it necessarily supports the development of the paper. Section 3 says "We investigate differences in flooding between the event- and response-based approaches by simulating flooding from a large number (5,000) of compound events that allow estimating the empirical distribution of flooding and comprise several 1% AEP events." This discussion is tortured. Although I understand what you're saying, I would suggest being more specific and explicit with your language. For example, instead of calling something a 25-year compound event or 4% AEP event, I would take a little bit more space and describe it as "an event whose rainfall or storm surge would be expected to be exceeded once every 25 years," which is a bit of a longer text but will greatly enhance the readability of the final work.My last concern is about the methodology used to generate the synthetic storms in a probabilistic framework. The authors provide a brief overview and largely refer to a previously published paper. It would be helpful to have a bit more detail here. In particular, the ways in which the authors sample from this bivariate distribution, and how it conceptually relates to other related approaches such as JPM-OS, climate model downscaling, synthetic storm generation, etc. are not clear (to be clear; those methods solve problems different from what the authors are doing here, so I am not suggesting that they need to defend and justify their choice so much as explain conceptual links and differences). The caveats and limitations of this approach ought to be more clearly discussed in the discussion section, in particular the discussion of how the assumption of uniform rainfall affects the finding that pluvial flooding is relatively insensitive to between-storm variability, which is at odds with other previously published work.Minor PointsI have a few specific suggestions that illustrate the above points, though they do not provide comprehensive wordsmithing for the paper.- page 1: avoid "hinterland" word choice- page 5: Maduwantha.....characteristics: this seems like a lot of work to go through to avoid generating synthetic, realistic events- page 11: "We find that the floodplain of each of these 1% AEP events is different, resulting in very large differences in both flood extent and depth between some of the events." this is perhaps an even better example of how the discussion of return periods here is confusing. What is an AEP event? I eventually was able to figure this out. A key finding of the paper is that there is not a 1:1 link between the return period of the rainfall rate or the storm surge and the return period of the flooding or the damage, in line with other studies. Calling something a 1% AEP event is not helpful here. I don't think the methods need to change, just the presentation.- I appreciate that the authors have pushed their code to GitHub. However, I visited <https://github.com/CoRE-Lab-UCF/MACH-Compound-Flooding/tree/main/Scripts> and was not able to figure out how to reproduce the results in the paper. It would be helpful for them to include a README file with instructions, a `main` script that runs the analysis, clear instructions for setting up the data, etc. I also was not able to figure out how the SFINCS code was run or where the input files for it are. It might be helpful to have a colleague check for reproducibility, as I would likely not be able to reproduce the results of this work.Citation: https://doi.org/
10.5194/egusphere-2025-1938-RC2 -
AC2: 'Reply on RC2', Sara Santamaria, 23 Aug 2025
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2025/egusphere-2025-1938/egusphere-2025-1938-AC2-supplement.pdf
-
AC2: 'Reply on RC2', Sara Santamaria, 23 Aug 2025
Viewed
Total article views: 595 (including HTML, PDF, and XML)
Cumulative views and downloads
(calculated since 12 May 2025)
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 441 | 133 | 21 | 595 | 37 | 15 | 30 |
- HTML: 441
- PDF: 133
- XML: 21
- Total: 595
- Supplement: 37
- BibTeX: 15
- EndNote: 30
Viewed (geographical distribution)
Total article views: 593 (including HTML, PDF, and XML)
Thereof 593 with geography defined
and 0 with unknown origin.
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
1
Latest update: 11 Sep 2025
Sara Santamaria-Aguilar
CORRESPONDING AUTHOR
Department of Civil, Environmental and Construction Engineering, University of Central Florida, Orlando, FL 32816, USA
National Center for Integrated Coastal Research, University of Central Florida, Orlando, FL 32816, USA
Pravin Maduwantha
Department of Civil, Environmental and Construction Engineering, University of Central Florida, Orlando, FL 32816, USA
National Center for Integrated Coastal Research, University of Central Florida, Orlando, FL 32816, USA
Alejandra R. Enriquez
School of Geosciences, College of Arts & Sciences, University of South Florida, St Petersburg, FL 33701, USA
Thomas Wahl
Department of Civil, Environmental and Construction Engineering, University of Central Florida, Orlando, FL 32816, USA
National Center for Integrated Coastal Research, University of Central Florida, Orlando, FL 32816, USA
Short summary
Traditional flood assessments use an event-based approach, assuming flood risk matches the chance of flood drivers. However, flooding also depends on topography and the spatio-temporal features of events. The response-based approach uses many events to estimate flood hazard directly. In Gloucester City (NJ, U.S.), we find that frequent events can cause rare (1 %) flood levels due to their spatio-temporal characteristics. Including these factors is key for accurate flood hazard estimates.
Traditional flood assessments use an event-based approach, assuming flood risk matches the...