the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 05 Oct 2025
Towards a typology for hybrid compound flood modeling
Abstract. Modeling compound flood events requires sophisticated approaches that can capture complex nonlinear interactions between multiple flood drivers. While combining different data-driven and physics-based modeling approaches has shown promise, the criteria for classifying such combinations and the underlying terminology to describe them remain inconsistent in the literature. To establish classification criteria, we introduce a systematic framework for defining and categorizing hybrid physical-statistical modeling approaches in compound flood modeling. Hybrid compound flood models offer significant advantages in terms of prediction accuracy and computational efficiency over traditional single-model approaches, particularly in coastal regions where multiple flooding mechanisms frequently interact. Here, we introduce a systematic framework for defining hybrid models and establish clear classification criteria based on their structural and functional characteristics. We identify three categories of hybrid models: sequential, feedback, and ensemble. Through illustrative examples, we demonstrate how each category leverages the strengths of its component models while also maintaining their independence. The proposed framework enables a systematic evaluation of different hybrid modeling strategies, enhancing model comparability and supporting the development of more effective compound flood prediction tools.
- Preprint
(1616 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
- RC1: 'Comment on egusphere-2025-4623', Anonymous Referee #1, 30 Oct 2025
-
RC2: 'Comment on egusphere-2025-4623', Anonymous Referee #2, 24 Nov 2025
I would like to compliment the authors for a well-organized, and clearly written manuscript that addresses a timely and underexplored topic: the classification of hybrid compound flood modeling frameworks. This review serves as a valuable entry point for researchers and practitioners interested in advancing compound flood modeling and provides a solid foundation for supporting the development of more effective compound flood prediction tools. Please see the “specific comments” below for strengthening the manuscript.
- As a key component of hybrid frameworks, the statistical modeling aspect appears to be less discussed in the manuscript. While they are included in Section 2, there is very little reference to them in sections 3.1 and 3.2 (and in section 4). In Figures 3 and 4 it’s not clear where the statistical/data-driven component sits that makes it a hybrid model. It becomes a bit clearer further down and in the appendix, but it would help to strengthen the focus on the “hybrid” aspect since that is what the paper is about, otherwise some parts become more like a duplicate of the Santiago-Collazo et al. (2019) paper, explaining how different process-based models can be linked/coupled.
- Related to the previous comment, for the sequential modeling approach, the role and implementation of statistical models may vary depending on the boundary condition requirements of the flood models (e.g., bathtub vs. dynamic). These statistical frameworks can range from generating only the peak boundary conditions to producing a full time series of boundary inputs (e.g., Moftakhari et al., 2019; Maduwantha et al., 2025 ). I encourage the authors to briefly discuss how such approaches can strengthen hybrid modeling frameworks
- The section on ensemble hybrid modeling for compound flooding is conceptually sound and addresses a timely topic. The only example provided relates to the use of ensemble methods for rainfall forecasting, which, while relevant, does not fully reflect the complexity involved in applying such an approach to compound flood processes. There are likely to be practical challenges when implementing such ensemble approaches for compound flooding (e.g., can statistical modeling alone generate compound flood depths that are comparable to physics-based model outputs to get the final weighted prediction?). A further discussion on how such approaches can be practically implemented would strengthen the manuscript
- Feedback hybrid models are recognized as models that enable bidirectional information exchange between coupled models/components. However, it is not clear whether bidirectional exchange between only two domains, such as atmospheric and ocean models, is sufficient to categorize them as feedback-hybrid compound flood models. In regions where tides, storm surge, and river discharge interact, tightly coupled models (e.g., wave and ocean circulation models) can provide a more robust way to simulate flooding. However, would similar advancements be achieved when bidirectional information exchange exists only between atmospheric–ocean or atmospheric–land surface models? Same for the example given in line 251, soil moisture content is mentioned as influencing local weather conditions. While this is true, is that alone sufficient to classify the entire modeling approach as feedback?
- In Figure 3, it seems that it’s just using discharge from a river model as input for a coastal model. According to their own definition, it would only be a hybrid model when one of the models is statistical/data-driven. That is not mentioned in the figure or caption. Same for other figures. I assume they mean that one of the models always has to be a statistical model but it’s not clear. Or is it enough if one of the shown model components is itself linked to a statistical model, like a rainfall generator driving the hydrologic model, which then connects to the coastal model!?
- In line 365, the authors mention that the strength of the statistical component is its ability to be faster. However, in the introduction (line 86), the authors only mention the ability of statistical methods to model multivariate dependence as their main strength, without noting the advantage of being faster compared to physics-based modeling. I suggest adding this point to the introduction for consistency.
- In Figure 6, “structural flexibility” is listed as a key consideration under ensemble modeling. However, it is not clear what is meant by structural flexibility in this context, and how it specifically relates only to ensemble modeling over other approaches.
- Some of the relevant recent studies on compound flood modeling are missing from the review (e.g., Jane et al, 2022; Orton et al., 2018). Including them would improve completeness and strengthen the review’s contribution.
Jane, R., Cadavid, L., Obeysekera, J., and Wahl, T.: Multivariate statistical modelling of the drivers of compound flood events in South Florida, Natural Hazards and Earth System Science, https://doi.org/10.5194/nhess-2020-82, 2020.
Maduwantha, P., Wahl, T., Santamaria-Aguilar, S., Jane, R., Dangendorf, S., Kim, H., and Villarini, G.: Generating Boundary Conditions for Compound Flood Modeling in a Probabilistic Framework, EGUsphere [preprint], https://doi.org/10.5194/egusphere-2025-1557, 2025.
Moftakhari, H., Schubert, J. E., AghaKouchak, A., Matthew, R. A., and Sanders, B. F.: Linking statistical and hydrodynamic modeling for compound flood hazard assessment in tidal channels and estuaries, Adv Water Resour, 128, 28–38, https://doi.org/10.1016/j.advwatres.2019.04.009, 2019.
Orton, P. M., Conticello, F. R., Cioffi, F., Hall, T. M., Georgas, N., Lall, U., Blumberg, A. F., and MacManus, K.: Flood hazard assessment from storm tides, rain and sea level rise for a tidal river estuary, Natural Hazards, 102, 729–757, https://doi.org/10.1007/s11069-018-3251-x, 2020.
Citation: https://doi.org/10.5194/egusphere-2025-4623-RC2
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 302 | 93 | 24 | 419 | 15 | 13 |
- HTML: 302
- PDF: 93
- XML: 24
- Total: 419
- BibTeX: 15
- EndNote: 13
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
I would like to compliment the authors on a very thorough and useful paper. I think there is great value in this very thorough review, and I think output such as Figure 6 can be very helpful. I did find the paper very dense which affects its readability; I have included some feedback below which maybe helps in addressing this.
Review
Some literature suggestions