Extreme storm surges exhibit significant seasonal and interannual variability influenced by large-scale climate modes. The goal of our work is to investigate the seasonality of storm surge extremes and the influence of interannual climate variability at the global scale, which to date is not fully understood due to lack of observations with long-records.  

To achieve this goal, we use storm surge levels derived from the Global Tides and Surge Model (GTSM) forced with the extended ERA5 climate reanalysis data spanning 1950-2022. Our methodology consists of two main steps. First, we classify the dataset into four seasons (Winter-DJF, Spring-MAM, Summer-JJA, Autumn-SON) and compute the number of events per season. Next, we conduct extreme value analysis on selected thresholds and explore their connections with climate modes.

Preliminary findings indicate that extreme surge events are more frequent and pronounced at higher latitudes during SON, with notable peaks in DJF. This is particularly significant in the North Sea and funnel-shaped coastlines such as Rio de la Plata, Arafura Sea, and Hudson Bay. In contrast, regions like the South China Sea, the Bay of Bengal, the Yellow Sea, and southern Australia experience more frequent surge extremes from JJA to SON with variations in peak season.

Equatorial regions, especially around Africa, have negligible surge extremes except for occasional tropical cyclones from late DJF, with peaks in MAM in Mozambique and Madagascar. Similarly, there are occasional tropical cyclone events in parts of the Caribbean with peaks in JJA.

The study findings have broader implications for understanding the global distribution and spatio-temporal variation of extreme surge events, which could help to provide guidance on the impacts of climate change in the future. Overall, the preliminary findings underpin the need to further explore what the drivers of storm surge variability are. 

How to cite: Apolola, A., Ward, P., Antolínez, J., and Muis, S.: Global estimation of storm surge seasonality and the effect of interannual variability., EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-2825, https://doi.org/10.5194/egusphere-egu24-2825, 2024.

The present study introduces a high-resolution Coastal Tide/Storm surge model (CTSM). Recently, the NIMS (National Institute of Meteorological Sciences) developed the CTSM, based on the NEMO ocean model to enhance forecasting capabilities for sea conditions and storm surges. The CTSM is constructed with a two-dimensional barotropic sigma coordinates and has a 1 km horizontal resolution. It consists of Tide/Surge model and Tide model, and their residual is used as surge forecasts. The surge forecasts are then added to the harmonically predicted tides to give forecasts of total water level at the 30 tidal stations around Korea Peninsula. Based on the sensitivity studies, the constant values of 0.0275 and 1024 hPa are adopted as the Charnock coefficient, bottom friction and reference pressure of the model, respectively.
In addition, this study investigated the effect of temporal and spatial variations of Charnock coefficient on the surge forecasts during Typhoon HINNAMNO, which caused substantial damage to the Korean Peninsula in 2022. For this, the 2-D Charnock coefficients derived from an operational Coastal wave model are added to the CTSM. It was found that the Charnock values generally exceeded the model’s constant value of 0.0275 during typhoon period. This alteration in Charnock coefficient impacts on the surge forecasts especially near the coastal regions, showing about 10% increase in the sea level.

How to cite: La, N. and Chang, P.-H.: Sensitivity analysis of coastal storm surge forecasting using NEMO-based CTSM model, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-7151, https://doi.org/10.5194/egusphere-egu24-7151, 2024.

Storm surges along the German North Sea coast are tidal surges that reach a peak of 1.5 metres or more above mean high water (MHW). Severe and very severe storm surges exceed 2.5 and 3.5 metres above mean high water respectively. Storm surges along the German North Sea coast are triggered by westerly winds from approx. 7 to 8 Beaufort. At the Hamburg, St. Pauli gauge, the long-term average is 5 to 6 storm surges per year; less frequent are severe (2 in 3 years) and very severe storm surges (every 5 years). The period of occurrence is essentially limited to the winter half-year.

The Federal Maritime and Hydrographic Agency (BSH) is responsible for warning of storm surges along the Baltic and North Sea coasts and the tidal river sections of the Ems, Weser, Jade and Elbe. Warnings of storm surges are distributed in very different ways. Warnings have been broadcast on the radio for many decades. In Hamburg, firecrackers are set off by the police. In recent years, dissemination via the BSH website and a customisable telephone distribution system has become established. Since the 2021/22 storm surge season, warnings have been fed into warning apps such as NINA via the Modular Warning System (MoWaS) of the Federal Office of Civil Protection and Disaster Assistance (BBK) and thus reach more directly affected citizens.

The possibilities offered by new media make it necessary to further develop warning strategies. For example, we are currently working closely with the BBK on the automated provision of warnings via the NINA warning app. This leads to faster and more precise distribution of storm surge forecasts.

The warnings and forecasts described take place against the background of the mean sea level rise and the associated rise in mean high water level. It is also the responsibility of the Federal Maritime and Hydrographic Agency to monitor this and to provide and analyse it at tide gauges such as Cuxhaven Steubenhöft, where measurements have been taken for around 180 years.

When creating the forecasts, we set up the Flood Early Warning System (FEWS), which pools and helps to process the data and creates and publishes reports. With the help of our developed Model Output Statistics System (BSH-MOS), precise and individualised forecasts up to one week into the future are possible for up to 40 gauges. Among other things, MOS evaluates water level measurements, wind forecasts from the German Weather Service (DWD) and area-based modelled water level forecasts from the BSH model system.

How to cite: Schenk, L. and Methe, M.: Storm surge warning for the German North Sea coast, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-10307, https://doi.org/10.5194/egusphere-egu24-10307, 2024.

The Dutch coast is characterized by dikes, dunes, and structural barriers with low-lying, densely populated hinterland, which makes the area very vulnerable to coastal flooding. Therefore, the reliability of flood forecasting models is of great importance: accurate short-term forecasts (up to 2 weeks lead time) are necessary for operational decision-making processes (e.g. closing the storm surge barriers on time), while mid-term forecasting (seasonal) is useful for the planning of maintenance, for example. However, the uncertainty of forecasts naturally increases with longer lead times, which means that the extent of a storm is often only known on short term, leaving little time to take safety measures (Pardowitz et al., 2016). With a changing climate, the uncertainty in forecasting might even increase. Improving our understanding of the characteristics of storm evolutions in present and future climate plays a fundamental role to reduce uncertainty in forecasting models.

In recent years, research into storms and resulting extreme sea levels along the Dutch coast has been boosted by the availability of long time series of meteorological data in the current climate, from the seasonal forecasting system (SEAS5) by the European Centre for Medium-Range Weather Forecasts (ECMWF) (ECMWF, 2021). For these synthetic time series of wind data, the Royal Dutch Meteorological Institute (KNMI) calculated corresponding sea levels (van den Brink, 2020). As a result, a period of 8,000 years of simulated meteorological and hydraulic data of the current climate have become available for many Dutch coastal locations. Compared to the limited availability of measurements from coastal stations (up to 50 or 100 years for a limited number of stations) these long time series are a great source of synthetic storm information.

The aim of this study is to explore the potential of using storm characteristics derived from these long synthetic time series of wind and corresponding water level for operational forecasting at the Dutch coast. First, physical and statistical properties of storm characteristics and their mutual correlations are analyzed. Storm characteristics consist of the temporal and spatial evolution of wind speed, wind direction and surge height, the duration of wind speed and storm surge above a certain threshold and the phase difference between the maximum storm surge and high tide. Mutual correlations between these characteristics are derived using copulas. Previous analyses result in strong correlations between wind speed and surge height, although it varies significantly depending on the location and combination of wind direction, duration and phase (Caspers & Kindermann, 2023). Still, this strong correlation suggests potential to be used for the forecasting of resulting storm surges from wind speed. Consequently, the correlations and other storm evolution properties found from these synthetic time series are compared to the observations of storms in recent years to investigate whether the findings from synthetic data agree with the characteristics of observed storm evolutions, in order to explore their potential for the short and mid-term forecasting of storm impact at the Dutch coast.

How to cite: Kindermann, P., Morales Napoles, O., and Antonlínez, J.: An exploration of the potential of using storm characteristics from long synthetic time series of wind and water levels for operational forecasting, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-19736, https://doi.org/10.5194/egusphere-egu24-19736, 2024.

Early 2022, four severe storms (Corrie, Dudley, Eunice and Franklin) raged over the Netherlands, of which the latter three hit the Dutch coast in only four days. The question is how well the Dutch flood protection system can deal with such a series of storms. Will there be enough time to recover from the previous storm?

The Maeslant barrier is a storm surge barrier near Rotterdam that exists of two enormous floating sector doors. In rest, they are safely located in the dry docks along the shore. Yet, in case of the most severe storms the doors are floated to the middle of the river and submerged to retain storm surges from the sea. After the storm they are floated up again and moved back to the docks. During its operation the Maeslant barrier is likely to be more vulnerable for small damages, that may lead to the temporal unavailability of the surge retaining function.

This study investigates 1) the probability that the barrier needs to close shortly after a previous closure and 2) the flood risk as a result of failure of the second closure. Herewith, we distinguish two different phenomena. The two-top storm is a single storm during which the barrier needs to close twice and open in between as a result of the astronomical tide. The twin storm (or even triplet or multiple storm) is a cluster of severe storms that succeed each other shortly.

The probabilities of both phenomena are estimated from a data-analysis of a long record of sea level observations at Hoek van Holland, close to Rotterdam. The associated flood risk is estimated from a simple conceptual model of the failure probability of the Maeslant barrier.

How to cite: Bakker, A. and Rovers, D.: Twin storms and the performance of storm surge barriers, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-13233, https://doi.org/10.5194/egusphere-egu24-13233, 2024.

Coastal regions in the Netherlands face persistent challenges from sea level extremes, prompting a comprehensive exploration of their spatial variability. Our study explores the nuances of extreme sea level events across the country, using the observed sea level data from the GESLA-3 (Global Extreme Sea Level Analysis) dataset. We analyse 16 stations with observational periods spanning from 38 to 68 years.

The total observed sea level is detrended and split into two components: (i) the tidal component, derived using harmonic analysis, and (ii) the non-tidal residual, calculated by subtracting the obtained tidal signal from the observed sea-level records. Extremes of both total sea level and non-tidal residual are then identified using the Peak over Threshold method, opting for a 70th percentile threshold. This choice allows us to examine less severe scenarios, suitable for risk assessments or planning purposes.

Our preliminary analysis of extreme event characteristics, such as the duration and intensity of an event, indicates significant spatial differences across stations. Correlation coefficients between stations, particularly for total extreme sea level characteristics and extreme non-tidal residual characteristics (duration and intensity), show a noticeable pattern that consistently reveals higher values between stations with similar latitudes across all variables. Moreover, the distributions of total extreme sea level characteristics exhibit noteworthy differences as well - for example, in southern regions, the distributions of intensity are more broadly dispersed and skewed to the right, signifying higher events than those in the northern counterparts. However, this distinction is less pronounced when focusing solely on the non-tidal residual, possibly since the total sea level is influenced by factors such as the river inflow, prevalent in the south, and tidal propagation behaviour in the North Sea.

As we progress with our analysis, we plan to apply a supervised learning method for classifying extreme events based on storm characteristics, and conduct a clustering analysis to reveal hidden spatial patterns of extreme events, for both total sea level and non-tidal residual. Furthermore, we aim to explore the interactions between surges and tides across different classes of extreme events, unravelling the underlying driving mechanisms of enhanced compound events.

In summary, our ongoing study of sea level extremes in the Netherlands, from spatial dynamics to event characteristics, will provide a solid foundation for understanding the driving mechanisms behind the extremes, gaining insights about their natural variability, and evaluating the impacts of changing climatic conditions.

How to cite: Pupić Vurilj, M., Alvarez Antolinez, J. A., Morales Napoles, O., Muis, S., and Mendez, F. J.: Unravelling Spatial Variability of Sea Level Extremes in the Netherlands: Insights from Observational Data, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-15703, https://doi.org/10.5194/egusphere-egu24-15703, 2024.

Projections of sea level rise are vital for assessing the impacts of climate change, especially in coastal regions. Present sea level rise projections are primarily focused on monthly water levels, but tend to underrepresent the critical role of storm surges. There are some studies that also provide projections of storm surges along global coastlines based on numerical models using meteorological forcing data from Global Climate Models (GCMs). However, those applications are limited by coarse meteorological inputs as well as the computational demands placed by running numerical models for an ensemble of different GCMs and climate change scenarios.

We propose a stochastic deep learning model trained on model output from numerical surge models. It is designed to capture the spatial and temporal dependencies that are characteristic of storm surge time series. Our approach generates potential storm surge scenarios that are consistent with GCM outputs but are not directly determined by those meteorological inputs. A second advantage is that the trained machine learning model has lower computational demands than traditional numerical models which makes it possible to explore different GCMs and climate change scenarios.

How to cite: Treu, S., Mengel, M., and Frieler, K.: A Stochastic Deep Learning Approach for Projecting Storm Surges in the Context of Climate Change, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-19209, https://doi.org/10.5194/egusphere-egu24-19209, 2024.

GNSS-IR represents an innovative technique for monitoring water levels. By analyzing the frequency of interference patterns between the direct GNSS signals and the signals reflected off the water surface, GNSS-IR offers a robust alternative to traditional tide gauges. GNSS-IR provides various advantages, including cost savings, convenient implementation, and accurate separation of vertical land motion. Recently, commercial companies started to adopt GNSS-IR for operational water level monitoring campaigns. The Lomb-Scargle Periodogram (LSP) is widely used to determine the frequency of interference patterns in the GNSS signal-to-noise ratio (SNR) data. Subsequently, the frequency/period can be converted into reflector height and water level. The LSP retrieves only one dominant frequency for each satellite and each channel, ascending or descending, over a time period longer than 20 min. Consequently, the temporal resolution of GNSS-IR water level measurements with LSP is lower compared to traditional tide gauges. High-temporal-resolution water level data would be valuable for applications like coastal hydrodynamics and hurricane studies. To address the temporal resolution, we developed a Single-Cycle Periodogram (SCP) analysis. The SCP analysis uses the LSP retrieval as a priori value and determines the period for each SNR cycle by tracking the maximum/minimum point corresponding to constructive/destructive interferences. Due to the reduced data span, the SCP suffers from noise. To improve the data quality of the interference patterns, we installed a GNSS antenna 90 degree tilted, facing the horizon, taking advantage of the antenna gain characterises. Such an experimental installation exists at the Onsala Space Observatory, with a relative small reflector height of approximately 3 m. Usually a small reflector height GNSS-IR installation results in low temporal resolution due to few interference fringes. However, using the proposed SCP analysis, preliminary results from 26 days of data indicate a significant increase in the number of water level retrievals. The LSP method yields approximately 200 unevenly distributed results per day, with occasional gaps exceeding 30 min. The SCP method gives approximately 10 times more retrivals. Furthermore, using the nearby traditional tide gauge (in the Swedish observation network of sea level) as a reference, the SCP retrievals, averaged over 6 min, provide a higher accuracy compared to the unevenly distributed LSP results.

How to cite: Feng, P., Haas, R., and Elgered, G.: Improving the temporal resolution of GNSS-IR water level monitoring using single-cycle periodogram, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-17544, https://doi.org/10.5194/egusphere-egu24-17544, 2024.

Recently, global trends in tidal amplitudes have been estimated from satellite radar altimetry data by including several constituents with linearly changing amplitudes into the harmonic analysis least squares problem. However, changes in tidal amplitudes do not have to be linear. The assumption of linearity can potentially obscure the true time-varying evolution of tidal amplitudes. Revealing deviations from linearity could be useful for attribution of physical mechanisms responsible for changes in tidal amplitudes and may have implications for future projections.

To address this limitation, an algorithm that estimates time-varying amplitudes without making any assumptions about the temporal shape is desired. To that end, we propose to use a state space time series model, for which time-varying parameters are estimated using a Kalman filter. Unlike the conventional least squares problem, the state space approach allows the value of a parameter to vary at each time step, providing a more flexible representation of the dynamic nature of tidal amplitude changes.

We apply the model to global TOPEX/Poseidon and Jason altimetry data from 1993-2023 at satellite crossover locations, aiming to identify if and where tidal amplitude changes are deviating from linearity. Provisional results show that, in many locations, the M₂ amplitude trend is close to linear during the considered timescale. Nevertheless, there are some regions where the estimated M₂ amplitude trends are clearly deviating from linearity. However, these results should be interpreted with caution since the 95% confidence intervals around the estimated amplitudes are often of similar magnitude as the temporal variability of the amplitude. One potential strategy to mitigate this issue involves increasing the number of samples per time series by binning altimetry observations, as opposed to restricting the analysis solely to crossover locations. To fully understand whether the generated time-varying amplitudes are reliable, the state space model will be thoroughly tested with synthetic data.

How to cite: Haakman, K., Slobbe, C., and Verlaan, M.: Estimation of time-varying tidal amplitudes using a state space model, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-10850, https://doi.org/10.5194/egusphere-egu24-10850, 2024.

Recent model results in combination with observations have provided a first coherent picture of secular changes in ocean tides since 1993. Strengthening of ocean stratification has been identified as an important driver of the observed secular trends, where the barotropic tide is primarily affected through enhanced tidal conversion at topography. These changes are responsible for open-ocean trends in the order of 0.1 mm yr^-1 for the barotropic M₂ tide, increasing to magnitudes comparable to the tidal response to sea level change (0.2—0.4 mm yr^-1) in several coastal regions. This has ramifications for global projections of future extreme sea levels, which either neglect changes in tides or consider them solely as a function of sea level rise. In this study, we employ a global high-resolution (1/12°) internal-tide permitting numerical ocean model to quantify future changes in ocean tides until 2100 as a result of upper-ocean warming and the concomitant increase in stratification. We simulate the evolution of leading tidal constituents in 5-year average time slices and use EC-Earth3P HighResMIP density data to constrain the model’s background stratification. As the Representative Concentration Pathway (RCP) in the EC-Earth3P simulation is a high greenhouse gas emission scenario (RCP8.5), we also consider data from a CM2.6 coupled global climate model, which is more closely aligned with a medium stabilisation scenario (RCP6.0).

How to cite: MacPherson, L., Opel, L., Schindelegger, M., Arns, A., and Vafeidis, A.: Changes in ocean tides by the end of the 21st century in response to stronger stratification, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-19316, https://doi.org/10.5194/egusphere-egu24-19316, 2024.