the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Reduction of the uncertainty of flood hazard analyses under a future climate by integrating multiple SSP-RCP scenarios
Abstract. The uncertainty due to small number of ensemble members is a major source of uncertainty in future climate predictions. While the uncertainty due to the above in general circulation models (GCMs) projections can be mitigated by increasing the number of simulation ensembles, only a limited number of large-ensemble experiments are available in CMIP6 future scenario experiments. Here we propose a method that increases the sample ensemble size in evaluations of future hazard, by integrating multiple SSP-RCPs for a time period corresponding to a specific increase in temperature from the preindustrial level (i.e., X °C warming). The success of the method was assessed by investigating whether the uncertainty due to small number of ensemble members could be reasonably reduced. First, the similarity in the spatial distributions of flood hazard projection at the same warming level was determined for different SSP-RCP scenarios. Under a 2 °C warming, all SSP-RCPs had a similar distribution with respect to the change ratio of the flood magnitude. Additionally, we showed that the uncertainty due to the different SSP-RCPs (5 %–10 %) was smaller than the differences between different warming levels such as between 2 °C and 3 °C (around 20 %–30 %), which suggests that differences among SSP-RCPs as to future flood discharge change are relatively small. These results suggested that integrating SSP-RCPs to increase the ensemble size was a reasonable approach, reducing unbiased variance among GCMs in about 70 % of land grid points comparing to the result using SSP5-RCP8.5 alone.
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2024-1027', Anonymous Referee #1, 03 Aug 2024
The manuscript proposes a statistical method to reduce the uncertainty in flood hazard projections by integrating multiple SSP-RCP scenarios. The authors suggest that this approach can mitigate the limitations posed by a small number of ensemble members in future climate projections, particularly for flood discharge assessments. However, the novelty of the approach is overshadowed by methodological and conceptual issues:
1, The core methodology of integrating multiple SSP-RCP scenarios to increase ensemble size is not convincingly justified. The assumption that different SSP-RCP scenarios can be combined as if they were additional ensemble members is problematic due to the inherent differences between scenarios. SSP-RCPs represent fundamentally different socio-economic pathways and climate forcing trajectories, influencing climate variables in distinct ways. The manuscript does not provide a robust theoretical or empirical basis to support this integration method. Although past studies (referred in this paper) indicate that uncertainty can be reduced by increasing ensemble size, they achieve this by using a wide range of initial conditions and climate model physics, grounded in physical principles rather than statistical manipulation. The findings in this paper may be contingent upon the specific GCM product used.
2, While the manuscript aims to reduce uncertainty, it does not adequately address the propagation of uncertainties from various sources, and the uncertainty reduction is not clearly shown. Integrating different SSP-RCPs might introduce new uncertainties, and the manuscript lacks a comprehensive analysis of how these new uncertainties are quantified and managed.
3, The manuscript claims that the proposed method reduces unbiased variance in 70% of land grid points. However, the validation of these claims is insufficient. There is a lack of independent verification using observed data or alternative high-resolution models. Without robust validation, the reliability and applicability of the proposed method remain questionable. At the very least, historical or present climate states should be used for validation purposes, even if the main objective is uncertainty analysis. The large uncertainties in the complex climate-discharge system might make the results meaningless, highlighting the need for validation.
4, Several key explanations in the manuscript are unclear or insufficiently detailed. For instance, the process of determining the similarity of flood hazard projections among different SSP-RCP scenarios is not described in enough detail to be reproducible. Specific queries include: How are the model boundary conditions and initial conditions determined? Is there consideration of evapotranspiration and infiltration? How is the river conveyance capacity estimated? Is there any downscaling method needed to solve the scale mismatch between the coarse resolution climate simulation and the fine resolution hydraulics needed? How about the bias correction in these climate model simualtions?
5, A typo in line 159, there is an extra ).
Citation: https://doi.org/10.5194/egusphere-2024-1027-RC1 -
RC2: 'Comment on egusphere-2024-1027', Anonymous Referee #2, 11 Nov 2024
The paper proposes a methodology to reduce the uncertainty in flood hazard for future scenarios by integrating existing data to create a larger dataset, as uncertainty due to the small number of ensemble members is typically an issue for similar applications.
The proposed approach includes the identification of scenarios that bring to the same increase of temperature, which are then used together to increase the number of ensemble members. Then, the discharge is calculated with such enlarged dataset, to compare discharge with historical data. Unbiased variance is finally used to compare the results taking into account the different size of the considered ensembles.
Although I believe that the approach is interesting and may be appropriate in general applications, I disagree with the application to flood event analysis. This because 30-years time-series are analyzed together as they represent a 90-years time-series, which is not the case, just thinking about the difference in term of discharge values with 30 and 100 years of return period.
- Merging data that brings to the same warning increase seems reasonable. However, the time series used are 30-years long. I am not sure that by putting 3, or 2, time-series together you gain data for 90-year, or 60-year simulation. The three, or two, series, were derived for 30 years, and 90-, or 60-years dynamics may bring to different data, especially in terms of extreme values.
- I suggest the Authors to anticipate the lines from 152 to 162 before chapter 2.1, as it may help the reader in understanding the entire methodology.
- The effect of topography variation could also be taken into consideration.
- There are many references to the supplementary materials. It would be better to think about some metrics that could summarize the behavior of the different conditions tested and move the text that describes the single figures in the supplementary materials themselves. In case the Authors believe that the fomented figures (e.g., 187-189) are necessary to better describe the results, these figures should be included in the main text rather than in the supplementary materials.
- Lines 191-192: These lines further reinforce the doubts expressed at point (1). I think it is not statistically, hydrologically and hydraulic accurate to consider the coupling of three 30-years time-series as representative for 90-years. This should be highlighted as a limitation: not a larger ensemble is required, but a longer one (compare to lines 329-331).
- The effectiveness of the methodology is evaluated by applying the unbiased variance computation. However, only qualitative results are shown (Figure 5 reports the comparison and the difference of the unbiased variance). Quantitative comparison should be performed, to highlight the efficacy of the proposed methodology.
Minor comments
- Typo at line 26: “Occillation” should be “Oscillation”
- Line 106: it is not clear how the 100-years return period discharge is computed.
- Line 174: it should be “Figure 1” and not “Figure 2”
- Figure 2: “correlation coefficient” of what?
- Line 183: “is larger for 3.0 than for 2.0”. Please write that you refer to 2.0°C and 3.0°C. The same for “1.5 and 2.0” in the subsequent lines.”
Citation: https://doi.org/10.5194/egusphere-2024-1027-RC2
Viewed
Since the preprint corresponding to this journal article was posted outside of Copernicus Publications, the preprint-related metrics are limited to HTML views.
HTML | XML | Total | BibTeX | EndNote | |
---|---|---|---|---|---|
198 | 0 | 0 | 198 | 0 | 0 |
- HTML: 198
- PDF: 0
- XML: 0
- Total: 198
- BibTeX: 0
- EndNote: 0
Viewed (geographical distribution)
Since the preprint corresponding to this journal article was posted outside of Copernicus Publications, the preprint-related metrics are limited to HTML views.
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1