| 24 Feb 2026
Flood Risk Projection Using a Hybrid Simulation Technique
Abstract. Future climate conditions project an increase in the frequency and severity of flooding in many regions of the world. Evaluation of candidate flood adaptation strategies must consider risk assessment methods that capture scenario-based loss and damage (L&D) for cost-benefit analysis. There is a need to develop tools that improve understanding of a region’s risk exposure while recognizing data and resource limitations available for this purpose. This study aims to address this gap by employing a novel approach that utilizes historic L&D data with an eye towards the current end-tail of extreme flood events as a prognosticator of what the future might hold. A hybrid Monte Carlo simulation technique is deployed to develop flood L&D projections under future climate change scenarios and used to estimate return periods of extreme flood events. Application of this methodology is illustrated in a case study using the Northeast region of the United States. The results show decreases in expected return periods of large flooding events, thereby expanding the geographic area of increased risk. These findings suggest this approach could function as a promising screening tool to help guide local flood adaptation planning, including the possibility of adopting this approach for other extreme weather events.
General comments:
The paper attempts to contribute to the theme of future projections of flood risk in a portion of the US, using a novel hybrid approach. One of the key asset is the good quality of the loss and damage dataset used. Unfortunately, I found the article far from publication standards for scientific articles as it is lacking on several aspects, including:
- Some methods seems to be proposed in this work without adequate proofs of their trustability nor any validation on past data.
- Questionable assumptions, perhaps the most relevant one being that changes in the impact of flooding is assumed dependent on the projected changes in mean annual precipitation.
- Incomplete representation of data and methods, which makes results not reproducible.
- Little practical value of results.
- Use of non standard article structure, mixing literature, methods, results, discussions and conclusions in different sections, making the reading more difficult.
More details in the specific comments below.
Specific comments
The introduction is rather chaotic; its structure should be improved. Please use a more standard structure including topic and statement of the problem, brief review of studies assessing this problem, progressively narrowing down to studies which are the most relevant /similar to the one presented, what is missing to the current understanding and why it is important, how this paper is going to improve the state of the art (here some information on the proposed work). In the current version these points are partly missing and those included are scattered in mixed order.
P3, l1-2: this sentence should be placed in the discussion or conclusions.
P4, l7: The section heading does not reflect its content.
P4, l9-22: This part is particularly confusing. It should be restructured, made clearer, so that the reader can understand the key steps involved in the methodology.
P4, l37: where does this equation come from? Please cite the relevant papers. If it’s the one taken from the work by Doktycz et al (2022) it should be clarified better in the text, including some performance scores on observed data to prove its value.
P5, l31-36: I don’t understand the reasoning behind this type of approach. Is it proposed here for the first time or taken from some literature work? In the latter case it should be cited.
P6, l8-10: This should be moved to the results
P6, l18-23: Change in the cost of flooding dependent on the projected changes in mean annual precipitation is over-simplistic. It does not reveal information on the intra-annual variability and given the large non-linearities involved it is not an acceptable assumption. Exceptionally, it could be used if a dedicated analysis is performed on observed data in the selected region, demonstrating that for the past decades such a trend was observed.
Figure 2 is not very relevant, especially as it focuses on temperatures and the submitted article is about precipitation and flooding.
P9, l15 to P10, l3: All this is about methods, rather than results.
Figure 3: Which scenario does the “shifted distribution” refer to? Also, the log-transformed damage is not much of practical use. Why not keeping original damage values and adapting the x-axis with logarithmic tick marks and labels?
I have quite some concerns on the choice of the scenarios shown.
First of all, choosing only one return period of event magnitude reduces a lot the relevance of the work, given that flood impacts are the result of an integral of different event magnitudes with their respective return periods where, depending on the shape of the loss function, smaller but frequent events can weigh more or less than very rare but catastrophic events.
Second, reading the text it sounds like 100-year return-period scenarios are extracted by several 20-year datasets. So even if the monte carlo approach increase the number of samples, extracting 100-year events from 20-year samples drastically increases the estimation uncertainty. Given that these are synthetic data, why didn’t you choose to work with 100-year long (or more) samples?
Third, the authors list quite a number of combinations, obtained by mixing SSPs, time frames, shifts of the body and the tail of the distributions. However, given all the arbitrary assumptions, the uncertainties in data and methods and the choice of only one return period, it sounds more like a simulation experiment but results have little practical use in terms of relevance for steering climate mitigation or adaptation policies.
Conclusions: first paragraph is more suitable for the Introduction. The second paragraph reads like part of the discussion section.