the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Increased Intensity and Frequency of Global Coastal Compound Wind and Precipitation Extremes: Implications for Sea Level Anomalies
Abstract. Coastal flooding and damage can result from compound extremes of wind and precipitation that elevate sea level anomalies. However, the global patterns and impacts of such conditions are poorly understood. Here we analyze observational and model data to reveal a positive correlation between wind and precipitation extremes across most of the global coastline, especially at higher latitudes. We also show that these variables exhibit stronger dependence on higher quantiles, indicating more frequent and severe compound conditions. Moreover, we demonstrate that sea level anomalies are enhanced during compound conditions compared to normal conditions, implying increased coastal flooding risk. We project that both the intensity and frequency of compound conditions will rise in 2020–2100 compared to 1940–2014 under two emission scenarios, with larger changes at high latitudes. Our findings highlight the need for assessing and managing the risks and impacts of compound extremes on coastal communities and infrastructure.
- Preprint
(2320 KB) - Metadata XML
-
Supplement
(1311 KB) - BibTeX
- EndNote
Status: open (until 08 Apr 2025)
-
RC1: 'Comment on egusphere-2024-3799', Anonymous Referee #1, 19 Mar 2025
reply
This study tried to analyze the global coastal compound wind and precipitation "extremes" for coastal flood risk assessment. However, this study focused on 90 percentiles of monthly “mean” values. These statistics cannot represent extreme events at all. 90 percentiles of monthly mean values suppose to represent just the seasons with the severest wind and precipitation. Therefore, I think this study analyzed just seasonal synchronicity of wind, precipitation and sea level anomaly. The authors subtracted the monthly mean for removing seasonality, but variations were not subtracted. The season with larger mean values should take larger variations. Furthermore, monthly mean SLA derived from CMIP6 global climate models represents large scale thermodynamic sea level variance and not coastal extremes such as storm surges. I think this study didn’t take an appropriate approach to achieve objectives.
Citation: https://doi.org/10.5194/egusphere-2024-3799-RC1 -
RC2: 'Comment on egusphere-2024-3799', Anonymous Referee #2, 31 Mar 2025
reply
The presented study analyses the long-term co-occurrence of stronger wind and precipitation, as well as a possible connection to deviations from the mean sea level for the global coast. Extremes are defined as months where wind and/or precipitation exceed the 90th percentile relative to the period 1993-2020. The authors use wind and precipitation data from the ERA5 reanalysis, sea level from a global gridded satellite altimetry product, and wind, precipitation and sea level anomalies from eleven CMIP6 models. The analysed time periods are 1993-2020 (ERA5 and altimetry), as well as 1940-2014 (historical simulations) and 2020-2100 (future simulations). Co-occurrences of variables exceeding the 90th percentile threshold in a respective month are derived by computing correlations, but also as the joint probability computed with different copula models.
While the manuscript is well-written, follows a logical structure and poses an interesting research question, the study has some major shortcomings:
- The formulation of the research questions as well as the context given for the study diverge from the actual analysis done. The authors announce to analyse the impacts of meteorological extreme events and their significance for coastal flooding. Readers usually interpret "extreme events" as episodic extreme events such as storm surges. However, instead of episodic extreme events, the authors actually study anomalies in long-term climate patterns and sea level changes. I see two possibilities forward:
a) I think the study of long-term changes is also an interesting research question that could increase our understanding of climate change effects on coastal sea level. However, this should be clearly communicated throughout the paper. In this case, the entire manuscript starting from the literature review and ending with the significance and implications needs to be rewritten.
b) If the aim was to study episodic extreme events, then a reader would expect that hourly or daily datasets are used instead of monthly means.
- Throughout the paper, there are several loosely defined terms such as "extreme conditions", "CWPE conditions", or "normal conditions". The phenomenon in question, for example "months where wind and precipitation exceed the 90th percentile" should be precisely defined and referred to consistently throughout the paper (possibly as an alias).
- While there is no precise definition for the word "extreme", I was a bit surprised by the relatively low threshold of the 90 % quantile. I would encourage the authors to explain why they chose this low threshold and not a higher percentile.
- I have concerns about the suitability of the sea level anomaly data used in the study. Looking up the cited dataset reveals that it's a gridded altimetry product. Conventional satellite radar altimetry was originally designed for the open ocean. Especially the quality of low resolution sensors (the only altimetry sensors available for most of the studied period) suffers in the vicinity of land. Obtaining high-quality sea level observations close to coasts therefore requires the use of specialised retracking algorithms. When using an open ocean product close to the coast, the uncertainties can easily exceed several decimeters. Additionally, we would expect that the tropospheric correction has less accuracy in periods with strong precipitation. The authors should explain how they deal with these issues, how the dataset they used was derived, what type of post-processing and which corrections they applied, how close to the coast the resulting sea level observations are, which sea level variations they expect between the closest observations and the actual coastline, and which accuracies they expect.
- If the goal of the study was to analyse episodic extreme events, then short-frequency sea level changes for example due to tides, wave setup, wind and atmospheric pressure need to be considered.
- The changes of sea level anomalies between "extreme" and "normal conditions" are presented as their absolute deviation from zero, where a mean of 1 cm in sea level anomaly is interpreted as an increase of 1 cm due to precipitation and/or wind. The underlying assumption is that during "normal conditions", sea level is the same as the mean sea level. In this context, the authors should explain which mean sea surface solution the results are related to, e.g. over which time period it was averaged, what is the horizontal resolution, is it a global or a local solution. Additionally, I think the interpretation of these results requires a discussion of the differences between regional and global sea level changes, and which role effects e.g. from self-attraction, loading and density differences play. Another approach that could bypass some of these issues would be to consider differences in sea level between months with strong wind/precipitation and months without strong wind/precipitation.
- In general, the study is lacking an assessment of accuracies of the employed datasets, and the significance of the results in view of these accuracies.
- The ERA5 reanalysis is continuously referred to as "observations". I think it should be made clear to the reader that it's a combination of observations and models.
- In section 2.2 Statistical analysis, I find it difficult to understand which processing was applied to which dataset. To enhance reproducibility, the authors should clearly state which steps where applied to which datasets. Furthermore, the division in subsections should be revised in a way that one subsection covers only topic to help the reader.
- Starting from equation (3), I am not able to follow the proposed methodology to compute the joint probability. I assume that equation (4) is the same as equation (3), but rewritten. From equation (4) it seems like C(0.9,1) and C(1,0.9) are the same as u and v. But u and v were defined before as FX(x) and FY(y). I think the meaning of the different variables and the computational steps should be explained. Also, a short introduction to the copula model and Sklar’s theorem would be highly appreciated. The same accounts for the introduction of the "Probability Multiplication Factor" (equation (5)).
- Section 3 reports a mean correlation of 0.12 +- 0.38 (between precipitation and wind) and 0.21 +- 0.23 (between precipitation and sea level anomaly) which are referred to as "positive correlation". The authors should explain how they come to the conclusion that these correlations are significant. I would instead suggest to include Figure S1 in the main manuscript to show that some areas indeed have a positive correlation, while others show a negative correlation.
- I'm wondering if the finding that wind and precipitation (and following that sea level anomaly) are more coupled in higher latitudes corresponds to the prevailing wind and precipitation climate in these regions. So before looking at their co-occurrence, it would be interesting to see maps only of wind and precipitation.
- Figure 2 shows the joint probability of wind and precipitation, averaged over all 11 used CMIP6 models. However, different models can exhibit very different behaviour. I would encourage the authors to report how the individual models perform and if it's possible that in some areas large positive or negative values average each other out.
- Figure 6 is interpreted as wind having less effect as precipitation on sea level anomalies. But especially for figure 6a) it seems as wind is having a quite strong effect, but more in the negative direction leading to lower sea levels.
- The results sections 3.2 and 3.3 as well as the discussion section should quantify the findings (e.g. how much does it increase?) similar as is done in section 3.1.
- I appreciate the structure of numbered research questions at the end of the introduction and their answers in the discussions. However, the questions asked are not the same as the answers given. I would advise the authors to formulate questions and answers in a way that they correspond to each other.
- Generally, I see shortcomings in scientific rigour. Two examples:
In the introduction, the authors cite a paper on the impact of snowfall anomalies on sea level rise from West Antarctic ice loss (Davison et al. 2023) as a source for the effects of wind and precipitation on coasts (line 49). None of the other three papers cited here are directly relevant to the statement made. Thus, these citations neither provide proof for the statements, nor relevant further literature for the reader.
In section 2.1, the focus on data below 60 degree latitude is justified with the decreasing quality of satellite data, due to "increased cloud cover ... and reduced solar radiation" (line 127 ff). To my knowledge, wind, precipitation and sea level are obtained from radar observations (when using satellite data).
- In my opinion, excluding small islands from the study should be mentioned as a limitation, as especially small developing islands are above average affected by climate change.
- Not taking into account wind direction is mentioned as a limitation. Regarding the huge effect that wind can have on coastal sea level (see e.g. Dangendorf 2013, https://doi.org/10.1007/s10236-013-0614-4), I think the authors should explain in detail their reasons for leaving out wind direction and how much they think the results could change.
- This is not a complete review. I have more minor comments, but I suggest that the authors first work on the bigger issues.
Citation: https://doi.org/10.5194/egusphere-2024-3799-RC2
Viewed
HTML | XML | Total | Supplement | BibTeX | EndNote | |
---|---|---|---|---|---|---|
139 | 28 | 7 | 174 | 28 | 7 | 5 |
- HTML: 139
- PDF: 28
- XML: 7
- Total: 174
- Supplement: 28
- BibTeX: 7
- EndNote: 5
Viewed (geographical distribution)
Country | # | Views | % |
---|---|---|---|
United States of America | 1 | 58 | 34 |
China | 2 | 22 | 12 |
Japan | 3 | 15 | 8 |
Germany | 4 | 14 | 8 |
Netherlands | 5 | 14 | 8 |
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1
- 58