the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 28 Oct 2024
Untangling the Waves: Decomposing Extreme Sea Levels in a non-tidal basin, the Baltic Sea
Abstract. Extreme sea level (ESL) events are typically caused by the combination of various long surface waves, such as storm surges and high tides. In the non-tidal, semi-enclosed Baltic Sea, however, ESL dynamics differ. Key contributors include the Baltic's variable filling state (preconditioning) due to limited water exchange with the North Sea and inertial surface waves, known as seiches, which are triggered by wind, atmospheric pressure, and basin bathymetry. This study decomposes ESL events in the Baltic Sea into three key components: the filling state, seiches, and storm surges. Our results show that storm surges dominate the western Baltic, while the filling state is more influential in the central and northern regions. Using a numerical hydrodynamic model, we further decompose these components based on their driving forces: wind, atmospheric pressure, North Atlantic sea level, baroclinicity, and sea ice. Wind and atmospheric pressure are the primary forces across all components, with the Atlantic sea level contributing up to 10 % to the filling state. These findings offer a deeper understanding of ESL formation in the Baltic Sea, providing critical insights for coastal flood risk assessment in this unique region.
- Preprint
(9727 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on nhess-2024-198', Mika Rantanen, 20 Nov 2024
Review of NHESSD manuscript nhess-2024-198 “Untangling the Waves: Decomposing Extreme Sea Levels in a non-tidal basin, the Baltic Sea” by Lorenz et al.
General comments
This manuscript provides quantitative decomposition of extreme sea level (ESL) events in the Baltic Sea into storm surges, filling and seiches. These components are further decomposed into components from various forcings such as wind and atmospheric pressure. The authors use sea level observations from the entire Baltic Sea coastline and simulations with a numerical model. One of the key results is that storm surges dominate ESL events in the western Baltic Sea, while the filling contribution is more important in the central and northern sea regions.
I found the topic of this manuscript highly relevant and valuable. Here in Finland, it is often stated that surges, filling (preconditioning), and seiches together cause the highest sea levels in the Baltic Sea. However, concrete evidence quantifying their relative contributions has been lacking. The same applies to the specific roles of wind and atmospheric pressure in driving high sea levels. In my opinion, this study fills an important gap by providing detailed quantification of these processes.
I think the overall presentation of the manuscript and its language was very good. The structure was logical, and there was a “red line” in the story which made the reading enjoyable. The background was nicely covered with relevant references, making the impression that the authors do know the topic very well. The results were discussed in a detailed way from various perspectives at the end of the paper. While I am not a marine scientist, I still understood most of the text.
Despite the positive feeling I got from the reading, I still found a few aspects which in my opinion require clarification: these are related to 1) methods and 2) the negative correlation between storm surges and filling. These are explained below. In any case, I can recommend publication after these (minor) comments have been addressed.
Minor comments
- In section 2, you present the observations (2.1), the model simulations (2.2) and the diagnostic decomposition method (2.3). There were some parts which I think were missing:
- In 2.3.1 (L115) you do not explain whether the decomposition is done for observations or model simulations or both. When I first did read these sections I assumed that the decomposition was only done for simulated sea level heights, so it was surprising to see in Fig. 3 that the method is applied to both. There is a brief mention in L92 that the decomposition is also done for observations, but this point should be emphasised later when presenting the decomposition method.
- At Section 2.3.1, you could explain how you derive ηESL. Is that extracted from modelled data or directly from observations?
- At L95, it would be clearer to write that the observational time series for different tide gauges are of different lengths. Or are they? And I assume that the detrending is based on the linear trend of the whole time series and not on a fixed period.
- In Section 2.2, it would make the choice of the model more robust if you could briefly mention whether the model has been used successfully in some previous studies.
- In Section 2.2. It was not clear for me what was the temporal resolution (hourly?) of the simulations, and how long were the simulations? And did you simulate the whole year, including the summer season? Overall the time period which was studied should be written more clearly (I found it from L147 but it could come earlier).
- In Section 2.4 (L158-163), the method of calculating the relative contributions of the forcings for sea level remained a little unclear to me. Could it be demonstrated using a single station example? Like writing down the magnitudes of the relative contributions from a station in Fig. 2d. This came back to me when I tried to interpret the sea ice contribution from Fig 7. You say (probably correctly) that its contribution is negative, but in Fig 7 they all look positive because they are presented as pie charts. Is there a contradiction, or have I misunderstood?
- At Section 3.1.1 (L201-213), I didn’t really understand the reason why filling and surges are negatively correlated, especially because in Fig. 2b they seem to be positively correlated (both are positive at the time of maximum). I read several times the sentence “Since the peak sea level of each event is fixed, a particularly high surge would naturally coincide with a lower filling state relative to the mean of the Gaussian distribution.”, but I still didn’t get the idea.
- From a meteorological perspective, strong cyclones are typically associated with (long-lasting) westerly winds, which would intuitively lead to a positive correlation between storm surges and filling. Given that this result appears to be one of the key findings of the study, and also being in an apparent contradiction with other studies, I suggest clarifying the mechanism in greater detail. Providing additional explanation would help resolve this apparent contradiction and strengthen the manuscript’s conclusions.
Other, specific comments
L12: This phenomenon: does this refer to the rising mean sea level or ESL events? Isn’t the ESL events the main cause of flooding, with a smaller contribution from rising sea level?
L31. By input data you mean weather prediction models or reanalysis? Can you mention them explicitly as I was wondering what input data is specifically meant here.
L62 These three temporal. Would it be better to put the three components together in brackets, for example, so that the reader does not have to go back to the previous page to see what the three were?
Table 1. TSClim: temperature or salinity?
Table 1. IceClim: inter-annual is written twice. And what does it mean by neglecting the inter-annual variability? Do you run the model with climatological sea ice cover?
L109. Does this mean you performed seven 58-year simulations?
L237. .. up to 30 %. This sentence remains a bit incomplete. Where does it contribute and what? Can you rephrase it?
L252 and L254 I think you write two times the residual term contribution? Is the 2nd (40%) for Danish Straits?
L258. As a meteorologist, I thought first that baroclinicity means atmospheric baroclinicity. Could it be rephrased to add seawater here?
Figure 7-9. Related to minor comment 1f. I don’t understand how the negative contributions from e.g. sea ice forcing is presented in these pie charts. For me it looks like all the forcings are contributing positively.
L271 wind systems. Maybe wind climatology is a better term here?
L279 its mean -> the mean contribution of filling
L281 on this time scale. Which time scale?
L288 Do you speak about the potential increase due to seiches here? It could be added to the sentence.
L296: 10 % on average. Was this result shown in some figure? If not, better to add “not shown”.
L303 ... currently very small. Maybe add reference to Figure?
L339. Aren't meteotsunamis more of a summer phenomenon, so that they generally don't occur at the same time as wind-driven extreme sea level events, which tend to occur in the winter season? If this is the case, it could be mentioned here.
Citation: https://doi.org/10.5194/nhess-2024-198-RC1 -
RC2: 'Comment on nhess-2024-198', Anonymous Referee #2, 10 Dec 2024
Overall, this paper leverages a validated model to analyse the components of extreme sea levels along the Baltic Sea, exploring their interactions and relative importance. The study introduces an innovative and engaging approach, and while the main conclusions are not entirely novel, they are well-structured, generalizable, and effectively capture the complexities of the Baltic Sea system. The paper is well-written and easy to follow, making it accessible to a broad audience. I recommend the paper for publication, subject to minor or moderate revisions. Below, I provide some specific comments to further enhance the quality of this already good work. The paper is well written, however, there are some typos here and there. Those that I noticed are mentioned, but I suggest the Authors re-read the document looking for minor typos.
SPECIFIC COMMENTS
Line 35: The claim about second-order effects I believe should be corroborated a bit more?
Table 1: Check typo.
Lines 104-105: Can you explain the reason of the 7% increased wind without only relying on the reference?
Line 118: can you please specify the filter order? Any specific reason why you used Butterworth filter? Can you explain the physics behind 7 days?
Line 120: “…a time window of +- 7 days” means a window of 14 days centred on the peak? Can you please clarify the overall approach that you used? What are the steps? Can you explain the link with the peak sea level?
Eq 2: can you please show if the results of the fitting provide realistic amplitude and phases?
Line 125-126: please clarify and justify your belief about the error. I do think that is negligible in your work, but it would be nice to have some solid ground to say so.
Chapter 2.3.2: how do you justify your decompositions? I am referring in particular to the seiches component.
Line 143: what did you do for detrending the time series? Can you please show both time series and what you removed to get the detrended one?
Line 148: can you please rewrite the following part “….which are shown in Fig. 1, see Fig. 2a for the identified events.” It is not clear.
Line 154: Please clarify the content of the bracket.
Line 156: it is not clear how you used the maximum values within +-24hours, can you clarify?
Line 158: Why do you need to normalise the components? It is not clear this step, please clarify what you did and provide the reasons to do so. Why did you not use the distributions from the original dataset?
Line 173: if you use the countries to explain the figure 3.a, please add the borders in the figure.
Lines 187-188: due to the layout it is not clear the resolution you are referring.
Lines 191-195: explain more clearly the reasoning behind your claims. Moreover, please provide at least some names of the locations on the map too. Moreover, Cuxhaven is not in the Baltic sea, why do you present?
Chapter 3.1.1:
- As you mentioned, the selection method (i.e. POT) and the assumed linear summation together induce at least some of the mentioned negative correlation. Can you please explain why this is less important in the surge/seiches correlation and why you keep having a positive correlation? Your explanation of the positive correlation is ok, but why the induced negative part is here less important?
- Can you please provide the details of the correlation analysis? Which correlation coefficient did you use? Can you please mask the map points having the p-values lower than a reasonable threshold? Without any diagnostic checks, the map can be misleading.
Chapter 3.1.2: I am unsure whether it is needed for the paper. I do not require the chapter to be removed, I leave this to the Authors, but I believe it's not so interesting as the rest of the document.
Figures 3 and 7: an option to have an idea of the location would be to add to the name of the locations on the x axis a number and reproduce these numbers in the map. In any case, something to help the localisation of the different locations on the maps should be done. I leave it to the author's preferences how. Please increase the y axis of the bar plot series.
Lines 255 and everywhere within the document specify the meaning of low-frequency waves? What are you referring to?
Line 274: check the typo, “is” is missing in the sentence.
Lines 284-285: is this temporal shift something realistic?
Lines 288-290: the sentence seems not finished. If these values do not represent that, then they represent what…?
Lines 292-293: The sentence “This result is expected since most ocean surface waves are forced by momentum transfer from the atmosphere to the ocean by winds or by atmospheric pressure via the inverse barometric effect.” might also be removed.
Lines 299-302: I do not see the link between the following sentences and the paper. I suggest to remove because not relevant for this paper, or therwise justify the reason to be mentioned. “However, with decreasing levels of sea ice due to climate change (e.g. Meier et al., 2022a, b, and references therein), the contribution of storm surge to ESLs is likely to increase in the future in regions that are currently covered by annual sea ice. In addition, with decreasing ice cover, the average wave loads and annual wave energy flux are expected to increase by about 5% and up to 82% respectively (Najafzadeh et al., 2022).”
Chapter 4.1.2: can be removed and for each excluded components specify what is the expected effect(s) on the main outcomes. It is already partially done, but I believe is interesting can be detailed a bit more addressing the effects rather than the reason why each component was not considered.
Lines 319-320: “However, the potential contribution of wave setup can be substantial in specific locations.“ can you be more specific?
Lines 375-378: if you consider the two statistics completely independent, the final event resulting from the sum of the two components having the same probability of exceedance is larger than accounting for the correlation between the components. Can you please revised the text?
Line 389: “…which could serve as a peak-over-threshold for GPD statistics,…” I think is redundant and slightly misleading.
Citation: https://doi.org/10.5194/nhess-2024-198-RC2
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 165 | 38 | 10 | 213 | 3 | 2 |
- HTML: 165
- PDF: 38
- XML: 10
- Total: 213
- BibTeX: 3
- EndNote: 2
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1