the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Dynamic Projections of Extreme Sea Levels for western Europe based on Ocean and Wind-wave Modelling
Abstract. Extreme sea levels (ESLs) are a major threat for low-lying coastal zones. Climate change induced sea level rise (SLR) will increase the frequency of ESLs. In this study, ocean and wind-wave regional simulations are used to produce dynamic projections of ESLs along the western European coastlines. Through a consistent modelling approach, the different contributions to ESLs such as tides, storm surges, waves, and regionalized mean SLR are included as well as most of their non-linear interactions. This study aims at assessing the impact of dynamically simulating future changes in ESL drivers compared to a static approach that does not consider the impact of climate change on ESL distribution. Projected changes in ESLs are analysed using non-stationary extreme value analyses over the whole 1970–2100 period under the SSP5-8.5 and SSP1-2.6 scenarios. The impact of simulating dynamic changes in extremes is found significant in the Mediterranean Sea with differences in the decennial return level of up to +20 % compared to the static approach. This is attributed to the refined mean SLR simulated by the regional ocean general circulation model. In other parts of our region, we observed compensating projected changes between coastal ESL drivers, along with differences in timing among these drivers. This results in future changes in ESLs being primarily driven by mean SLR from the global climate model used as boundary conditions, with coastal contributions having a second order effect, in line with previous research.
- Preprint
(9844 KB) - Metadata XML
-
Supplement
(3029 KB) - BibTeX
- EndNote
Status: open (extended)
-
RC1: 'Comment on egusphere-2024-1061', Anonymous Referee #1, 24 May 2024
reply
The paper uses a dynamic modelling approach that accounts for the different extreme total water level components (mean sea level, tides, storm surges, waves) and some of the non-linear interactions between them. The analysis is performed using one way coupling of a 3D ocean circulation model which captures MSL, tides, and meteorological surge and a wave model (forced with the same atmospheric data and considering water levels from the ocean circulation model). The analysis is conducted for the period 1970 to 2100 and non-stationary extreme value analysis is applied and compared to results from using the simplified “static” approach (or “MSL offset method"). Given the important of changes in extreme coastal water levels in driving changes in coastal flood risk, the analysis adds valuable new insights in terms of which methods are suitable in which regions. I add some major and several minor comments below which need to be addressed to make the manuscript publishable.
Major:
- One big concern that I have is about the uncertainties in the extreme value analysis, which are crucial since they are used to identify regions where the static and dynamic approaches lead to different results. There is no explanation of how uncertainties in the distribution parameters and resulting return level estimates are derived; is it the Delta Method, or Bootstrapping, or something else? Overall, the uncertainty estimates appear unrealistically small and that would lead to more places being identified where static and dynamic approaches are “different”. Looking for example at Figure S1.1 the uncertainties in the 100-year water levels (derived from ONLY 20 years of data) are only a few centimeters in some of the places and maybe 10-20 cm in others for the static approach. They become a bit larger in the dynamic approach because the shape parameter changes signs, but especially for those unbounded distributions the uncertainties are usually very large, particularly when GPD is fitted to short records (as is done here). I don’t think this can be correct and would mean that all the conclusion regarding “significant” differences between static and dynamic approaches are corrupted. I am not sure if maybe some uncertainties were ignored (like the ones in the shape parameter) or not propagated properly (as in the Delta Method), but something is off. I know the authors say they use the tool that was already published but that doesn’t mean it does the right thing. The reason I give “major revision” is because this is a critical point that may change the results/conclusions.
- Calafat et al. (2022; https://doi.org/10.1038/s41586-022-04426-5) showed that trends in storm surges are comparable to trends in MSL at several coastline stretches in Europe and that this has led to pretty large changes in return periods. This clearly challenges the conclusion which is also drawn here in line 370 that “changes in ESL primarily depend on SLR”. They also showed that small ensembles cannot capture the full picture of ESL changes. I would like to see some discussion about how the results presented here relate to that.
Minor comments:
19 “significant” in a statistical sense? If so which significance level? If not ina statistical sense I suggest changing and not using the term in a paper like this where statistical significance is also a big part
34-37 what about freshwater discharge?
97 return levels
115-120 Would the model resolve changes in (coastal) tides as a result of SLR or is it too coarse?
137 after runoff there is a closing bracket but no opening one
153 types
173-201 Somewhere it should be highlighted that a “direct” method is used which fits the GPD to the still water levels which include (often large) deterministic tide signals as opposed to the more appropriate “indirect” methods such as SSJPM where the stochastic surge/wave part is analyzed with the extreme value models and combined with the tides.
198 see my first major comment, it needs to be explained how those confidence levels are derived exactly
Fig. 2 Is that a real example or just (used as) a sketch? If it’s a real example it would be good to mention whether it uses the static or dynamic approach.
242 The German Bight is equally (if not even more) complex than the Dutch coast.
266 This seems to be quite relevant, what is the reason for not including it in the main manuscript?
Fig. 4 Related to my point about uncertainties it would be interesting here to show the ranges of years including the distribution uncertainties and also see where those ranges overlap between SSPs and where they don’t
278 This related to my comment above about changes in tides and whether they can actually be resolved along the coast
325 lead to
Citation: https://doi.org/10.5194/egusphere-2024-1061-RC1
Viewed
HTML | XML | Total | Supplement | BibTeX | EndNote | |
---|---|---|---|---|---|---|
227 | 59 | 12 | 298 | 23 | 9 | 8 |
- HTML: 227
- PDF: 59
- XML: 12
- Total: 298
- Supplement: 23
- BibTeX: 9
- EndNote: 8
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1