This session encourages contributions addressing dynamic mechanisms across an entire spectrum of atmospheric and ocean extremes, event attribution studies and projections under future climate. Relevant submissions also encompass new observation techniques, new modeling and machine learning methods to marine extremes forecasting, novel detection strategies and, finally, ecosystem or socio-economic impact assessments, relevant for prevention, mitigation and adaptation policies.
The 2013–14 UK winter storm series featured an unusually high number of storm events that caused 10 deaths, widespread flooding, coastal erosion, and damage to critical infrastructure. Official direct damage estimates of £1.3 billion were likely too low because they did not account for indirect disruptions, highlighting the broad economic and structural impacts of large storms. The observational analysis drew on wave, tidal, wind, and precipitation data from 44 tide gauges, 52 wave buoys, and 15 climate stations, giving good coverage of the UK coastline. Findings showed that thresholds for storm surge and wave height were exceeded more often than in previous winters, indicating unusually frequent episodes of storm surges and extreme waves. The maximum storm surge recorded at Lowestoft during Storm Xaver reached 2.2 m and contributed to extensive damage along the East Coast.
To complement observations, storm surge propagation was simulated using the Regional Ocean Modelling System (ROMS), providing a detailed 1 km resolution around the entire UK coastline. The simulations were validated against tide gauge data and successfully reproduced surge amplification from the North Atlantic into the UK’s shallow coastal zones, supporting their use in a nationwide coastal hazard prediction framework. Sensitivity tests using multiple nested domain configurations, along with a heuristic method for assigning land–ocean categories to coastal grid cells, improved numerical stability and revealed an optimal domain setup that balanced performance and computational cost. Further analysis offers insights into how climate change, storm tracks, and cyclogenesis influence surge maxima. These results provide practical guidance for early warning systems, infrastructure planning, and coastal management in a changing climate, and they can be applied to global disaster-risk resilience modelling.
How to cite: Shaw, D., Šepić, J., Heller, V., Heidarzadeh, M., Takagawa, T., and Iwamoto, T.: Spatio-temporal analysis of extreme coastal hazards during the 2013–14 UK winter storms: integrated observations and high resolution synoptic CFD storm-surge modelling, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-1148, https://doi.org/10.5194/egusphere-egu26-1148, 2026.
Forecast uncertainty is essential for operational decision-making in storm surge forecasting. Current probabilistic operational systems typically estimate uncertainty by ensembling deterministic numerical models forced by meteorological ensemble members. While this approach captures atmospheric uncertainty, it neglects uncertainty in the oceanic response and surge propagation, leading to forecasts that are commonly underdispersed.
We present a framework that explicitly accounts for these missing sources of uncertainty, including those arising from waves and other coupled processes. Building on the deterministic tidal response method of Munk and Cartwright, we model storm surges as conditional time-invariant stochastic processes. These processes are defined by distributions over nonlinear impulse-response functions that map gravitational, meteorological, and other forcings to total sea level. The response distributions can be flexibly conditioned on recent observations, such as in-situ gauge data, while remaining robust when such data are unavailable.
To learn these processes, we introduce Mixture Density Neural Processes (MDNPs), a Bayesian neural architecture that combines the expressiveness of neural networks with the stochastic function modeling capabilities of Gaussian processes. The models are trained on in-situ observational data at locations used for operational decision-making but, due to the conditional time-invariant formulation, do not require such data to generate subsequent forecasts.
We demonstrate state-of-the-art performance against operational baselines from the UK, the Netherlands, and the United States. We further show how MDNPs can be coupled with traditional numerical models to improve both forecast accuracy and uncertainty calibration. Currently being trialed in UK and Dutch operational systems, the approach maintains high performance during extreme events, producing calibrated forecasts even for previously unseen peaks. We attribute this robustness on tail events to the impulse-response formulation and discuss the broader applicability of the framework to multiscale and compound coastal hazards.
How to cite: Monahan, T., Adcock, T., Polton, J., and Roberts, S.: A conditional time-invariant framework for probabilistic storm surge forecasting , EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-6880, https://doi.org/10.5194/egusphere-egu26-6880, 2026.
Severe coastal floods affected the northern Adriatic Sea in December 2019, despite the absence of clear warnings from operational forecasting systems. This study investigates the physical mechanisms underlying these events using a combination of several approaches, including analysis of in situ wind and sea-level observations, atmospheric reanalysis products, simulations performed with a high-resolution hydrodynamic model, and a simplified analytical model. The results reveal a previously unrecognised wind-induced resonance mechanism. Specifically, a quasi-periodic wind forcing, associated with a sequence of successive cyclones over the Adriatic region, efficiently excited the basin’s fundamental barotropic mode. This resonant response led to a substantial amplification of sea-level oscillations, contributing significantly to the observed flooding. The findings identify a new type of resonance, complementing the four resonance mechanisms previously described in the Adriatic-related literature, and highlight the importance of accounting for resonance effects in coastal flooding assessment and forecasting.
How to cite: Bajo, M., Arpaia, L., Ferrarin, C., and Orlić, M.: A previously undetected resonant mechanism influencing the Adriatic sea-level dynamics, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-7193, https://doi.org/10.5194/egusphere-egu26-7193, 2026.
Marine ecosystems are increasingly affected by extreme events such as heat waves, oxygen depletion, and algal blooms. Potential consequences include fish kills due to oxygen depletion and beach closures due to algal blooms. These can cause enormous ecological, social and economic damage. With extreme marine events intensity thresholds, which induce physiological stress or even mortality on marine organisms, are more likely passed with potential consequences ranging from the species to community level, raising concerns over ecosystem stability and habitat preservation (Daru & Rock 2023, Antão et al., 2020).
The Baltic Sea, located in north-eastern Europe, is an ideal location for studying extreme marine events due to its susceptibility to climate change and anthropogenic activities such as excessive nutrient inputs. In order to understand highly dynamic extreme events, high-frequency observations over sufficient time spans are necessary. However, observations are often limited in terms of both space and time. To overcome these limitations we used both available daily resolved station data and high-resolution 3D outputs from the coupled hydrodynamic-biogeochemical model MOM-ERGOM. For the western Baltic Sea in particular, high-frequency measurements from 2011 to 2024 enabled a more detailed analysis of the similarities and differences between the areas, which will be presented at the conference. Despite localized differences in the identification of extreme marine events between the model and measurements, spatial analysis remains a powerful tool for understanding extreme events in coastal areas. The long time series also facilitates the evaluation of potential influences on the monitoring and assessment of water quality. Moreover, we investigated cascading and compounding extreme marine events to improve our understanding of the dynamics to which marine organisms are exposed when subjected to multiple stressors. This knowledge can help us design more realistic multi-stressor experiments and ultimately assess the impact of extreme marine events on organisms.
How to cite: Piehl, S., de Ramos, B., Friedland, R., Mars, R., Neumann, T., and Wolf, F.: Extreme marine events in the western Baltic Sea: a data- and model-based approach, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-17680, https://doi.org/10.5194/egusphere-egu26-17680, 2026.
Ocean extreme events, particularly those being spatially and temporally compound, pose substantial threats to ecosystem stability and remain insufficiently explored within current frameworks. Such extremes are expected to affect key biological groups, specifically plankton, which form the base of marine food webs. This study introduces a novel approach by coupling sea surface temperature extremes – Marine Heatwaves (MHWs) and Marine Cold-Spells (MCSs) – with Mixed Layer Depth (MLD) dynamics, a critical yet overlooked combination of extreme events. Focusing on the Mediterranean Sea, a "miniature ocean" and global climate change hotspot, we use an integrated multi-scale approach, covering the full water-column from surface to depth. Combining long-term (1998 - 2023) satellite observations, BGC-Argo derived datasets, numerical models, and in-situ measurements, we investigate the dynamic mechanisms driving plankton variability during the productive winter-spring bloom period. We demonstrate that MHW-shallow MLD compound events intensify vertical stratification and nutrient depletion, leading to reduced productivity. Conversely, MCS-deep MLD compounds stimulate primary production by enhancing vertical nutrient transport. Vertically integrated responses are more robust (80 - 88% consistency) than surface observations. In-situ evidence suggests a trophic cascade: MCS-deep MLD compounds favour larger phytoplankton, whereas MHW-shallow MLD extremes drive shifts toward smaller phytoplankton groups, with implications for the energy transfer efficiency to higher trophic levels. As climate-driven warming increases MHW frequency and suppresses MCSs, this framework enhances our capacity to predict ecological risks and offers a scalable tool for developing mitigation and adaptation strategies in a warming ocean.
How to cite: Theodorou, I., Gittings, J. A., Livanou, E., Kournopoulou, A., Pateraki, M., Organelli, E., Zervoudaki, S., and Raitsos, D. E.: A novel Framework for Assessing Plankton Responses to Compound SST-MLD extremes, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-20900, https://doi.org/10.5194/egusphere-egu26-20900, 2026.
Marine heat waves (MHWs) are among the most impactful oceanic extremes, yet their spatial structure remains poorly constrained in climate models. Here we show that low-resolution (LR) climate models systematically overestimate the spatial coherence of marine heat waves, producing events that are unrealistically connected across large ocean regions. Using the same LR–high-resolution (HR) model hierarchy previously employed in Gou et al. (2024) and Gou et al.(2025), we quantify MHW spatial structure using decorrelation length scales and snapshot-based connectivity diagnostics, and compare simulations to satellite-based observations. Observed MHWs exhibit rapid spatial decorrelation and fragmented event patterns. In contrast, LR simulations show decorrelation lengths that are too long and MHW snapshots dominated by basin-scale connected components. Increasing horizontal resolution substantially reduces spatial coherence, increases the number of effective spatial degrees of freedom, and brings both decorrelation scales and event connectivity closer to observations. The strongest resolution sensitivity is found in the Southern Ocean, where LR models produce unrealistically synchronized extremes. Our results demonstrate that model resolution fundamentally controls the spatial organization of marine heat waves, with important implications for impact assessments and future projections.
How to cite: Ma, Y., Lohmann, G., and Gou, R.: Climate models overestimate the spatial coherence of global marine heat waves, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-5949, https://doi.org/10.5194/egusphere-egu26-5949, 2026.
Global ocean warming has led to a projected decrease in the frequency and intensity of marine cold spells (MCSs). However, the East/Japan Sea presents a climatic paradox, as winter MCSs continue to occur despite a clear, long-term warming trend. This study investigates the mechanisms underlying the sustained occurrence of these events. The analysis indicates that extreme coastal cold anomalies are closely linked to the intensified southward flow of cold and less saline North Korea Cold Water (NKCW) from East Korea Bay. Composite analyses and vertical profiles showed that cold, low-salinity water resulted from the intensified southward flow of the NKCW, which is associated with the weakened northward flow of the East Korea Warm Current. The time-lag correlation indicated that the low-salinity water causing MCSs in winter originated from freshwater inflows through the Korea Strait in summer and autumn. Low-salinity water could intensify upper ocean stratification and enhanced surface cooling, resulting in an increase in winter MCS events. Considering that the CMIP6 climate change scenario project indicate an increase in the Yangtze River discharge, which is the primary freshwater source through the Korea Strait, these findings suggest that the potential for coastal MCSs could be sustained in a warming ocean.
How to cite: Tak, Y.-J., Kim, C., Kim, G.-D., Jeong, S. J., Jung, H. K., Kim, Y.-Y., and Cho, Y.-K.: Winter marine cold spells driven by intrusion of cold, low-salinity water in the western coast of the East/Japan Sea , EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-6246, https://doi.org/10.5194/egusphere-egu26-6246, 2026.
Intense high-frequency sea-level oscillations (HFOs) in the Mediterranean Sea, sometimes leading to destructive meteotsunamis, are generated by specific meteorological conditions that are spatially limited (from several tens to a few hundred kilometers). Although the physical mechanisms driving extreme HFOs are well understood, existing forecasting systems based on hydrodynamic models remain unreliable and computationally demanding.
To address these limitations, we developed deep-learning models (CNNs and ViTs) to predict HFOs using data from the Adriatic tide-gauge station Bakar, which provides a long record (2003–2025) but is not particularly prone to meteotsunamis. Models trained at Bakar can, however, be transferred to meteotsunami-prone Adriatic locations with shorter data records (Stari Grad, Vela Luka, Mali Lošinj, Sobra, etc.). Models were trained using measured 1-minute sea levels together with two sources of simulated atmospheric data (2D and 3D fields): hourly ERA5 data at 30 km resolution and 3-hourly CERRA data at 5.5 km resolution.
We will present model architectures and predictions of HFO amplitudes as a function of (i) forecasting horizon (up to several days, using different input windows of 6 h and 24 h), (ii) atmospheric data source (ERA5 vs. CERRA), and (iii) different combinations of training, validation, and testing periods. The main findings are as follows: (i) daily HFO amplitudes remain reasonably predictable over multi-day horizons, with comparable results from CNN and ViT approaches; (ii) forecast skill is higher for low-amplitude HFOs (up to ~12 cm); (iii) higher-amplitude events (10–40 cm) are generally underestimated; (iv) higher-resolution atmospheric forcing (CERRA) does not improve forecast skill, suggesting that meteotsunami-triggering atmospheric disturbances are not better represented at higher resolution; and (v) the choice of training, validation, and testing intervals has little effect on forecasting of small-amplitude events but affects forecasts of larger-amplitude HFOs.
How to cite: Medugorac, I., Metličić, N., Rus, M., Urbas, M., Šepić, J., Kristan, M., and Ličer, M.: Deep-learning prediction of high-frequency sea levels in the Adriatic Sea, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-9073, https://doi.org/10.5194/egusphere-egu26-9073, 2026.
Mediterranean cyclones are often weaker and smaller than other mid-latitude cyclones that develop over open oceans. However, they have a great impact on the Mediterranean basin which is highly populated on the coasts, causing heavy precipitation, windstorms, storm surges and sea wave extremes. The marine and coastal hazard related to cyclones can be quantified with several parameters like the maximum storm surge, the maximum significant wave height and other intensity indices based on waves or sea levels. The aim of this study is to investigate the relationship between different marine and coastal hazard parameters.
In this work we consider a dataset of more then 1000 cyclones which took place in the period 1994-2020 in the Mediterranean basin. The reference dataset is provided by a composite approach that uses different detection and tracking methods, proposed by Flaounas et al. (2023). To analyse the ocean response a coupled hydrodynamic-wave numerical model is used to account for the wave-current interaction. The modelling system consists in the SHYFEM (System of HydrodYnamic Finite Element Modules) hydrodynamic model, two-way coupled with the WW3 (WAVEWATCH III) wave model. The CERRA data are used as meteorological forcing in terms of wind and mean sea level pressure. To assess the marine hazard the following parameters based on sea levels and sea waves are considered: the maximum storm surge, the maximum significant wave height, the mean cyclone influence area, the total storm wave energy, the storm power index and the storm erosion potential index. Relationship between different measures are analysed through Pearson and mutual information correlation coefficients.
First results shows a strong correlation between the maximum significant wave height and the intensity indices based on waves. On the contrary there is
a weak correlation between the maximum storm surge and the intensity index based on sea levels. Other results concerning the relationship between hazard parameters will be presented. Ongoing investigations aim to include in the analysis some atmospheric features such as mean sea level pressure and wind in order to characterize the relationship between atmospheric anc ocean parameters.
How to cite: Pavan, S., Ferrarin, C., Bajo, M., Barbariol, F., Benetazzo, A., and Davison, S.: Relationship between marine and coastal hazard parameters of Mediterranean cyclones, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-12209, https://doi.org/10.5194/egusphere-egu26-12209, 2026.
Extratropical cyclones (ETCs) are low-pressure systems that occur in the mid- to high latitudes of both hemispheres. They typically form through cyclogenesis over the ocean near upper-tropospheric jet streams, and less frequently through transition from other storm types. As high-impact weather events, they are often associated with extreme winds, heavy precipitation, and storm surges, and can result in loss of life and extensive property damage. Consequently, any change in their activity (e.g., due to global warming) may have profound impacts on socio-economic systems and human well-being.
In this study, we tracked and identified extreme ETCs during the extended winter season (ONDJFM) using sea level pressure (SLP) tracking (tracks with minimum SLP <981 hPa) from a bias-corrected general circulation model (bGCM) over the North Atlantic domain for the SSP5-8.5 scenario (2041–2100). Historical track conditions based on ERA5 (regridded and temporally sampled to match the bGCM data) were analysed to evaluate how well the bGCM represents extreme ETCs in the Baltic Sea region during 1981–2010. Identified extreme events were assessed based on their tracks and associated 850-hPa wind fields to identify potential extreme storm surge candidates for Pärnu Bay, Estonia. Selected events were then dynamically downscaled using the Weather Research and Forecasting model (WRF v.4.6) to a 20 km resolution grid covering the Baltic Sea. The WRF output (wind and air pressure) subsequently forced the Finite Volume Community Ocean Model (FVCOM) for the Baltic Sea. Simulated water level fluctuations were compared against a local historical extreme storm Gudrun on 9 January 2005, which caused a record-high storm surge of 2.75 m in Pärnu. All simulations used identical numerical domain configurations and parameterization schemes, with only atmospheric forcing for water level simulations (the background water levels prior to the storm Gudrun were approximately 70 cm above the long-term average).
Storm track frequency during the historical period was underestimated by 3.1%, while intensity (850-hPa winds and mean SLP) was overestimated at the extreme end (90th percentile), with a spatial track bias. Overall, the bGCM tends to overestimate ETC intensity while slightly underestimating the frequency of extreme events in the Baltic Sea region. In contrast, future regional ETCs showed an increase in frequency of 14–23%. Changes in intensity depended on the future time slice and metric considered. The most extreme future event occurred in February 2088 (SSP5-8.5), exhibiting a storm track similar to the 2005 Gudrun storm but tracked few degrees northward. This minimum SLP reached 937.6 hPa over the Gulf of Bothnia, compared to 962.6 hPa for the Gudrun. Owing to its unique track and wind field (maximum 850-hPa wind speed of 76 m/s compared to 50 m/s for Gudrun), this event generated multiple high storm surges concurrently across major northeastern gulfs. In Pärnu, the maximum storm surge reached 2.86 m, compared to 2.28 m during Gudrun (excluding background water level). If such an event were to occur under projected sea level rise conditions, the resulting local impacts would be severe, particularly if accompanied by elevated background water levels.
How to cite: Mäll, M., Suursaar, Ü., Tõnisson, H., and Nakamura, R.: Dynamical downscaling of extreme extratropical cyclones for storm surge analysis in the eastern Baltic Sea under future climate scenarios, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-15398, https://doi.org/10.5194/egusphere-egu26-15398, 2026.
In May 2023, a low-pressure system named Minerva by the Italian weather service Meteo Aeronautica Militare brought heavy rainfall and strong winds to Italy and neighbouring countries, resulting in widespread damage and tragic loss of life.
During this event, autonomous instruments, along with satellite and modelling products, monitored the Tyrrhenian Sea, the Sicily Channel, the Ionian Sea and the south Adriatic Sea. In this work, we have combined these data to investigate the physical and biogeochemical responses of the upper layer of the Mediterranean Sea to Storm Minerva.
The upper ocean feedback typically consists of the sea surface cooling and subsurface warming (“heat pump” effect), driven by a combination of different physical processes. This effect can be more or less pronounced and can have different characteristics depending on the oceanic conditions encountered by the storm on its path.
Our findings reveal that Minerva triggered three different responses in the subbasins studied, which were influenced by the circulation structures and thermohaline conditions in the affected areas. This result underscores the importance of understanding the regional oceanographic characteristics when assessing the impacts of extreme weather events and how this information contributes to improved forecasting and mitigation strategies for future events.
How to cite: Vallortigara, C., Martellucci, R., Pirro, A., Reale, M., Mauri, E., Poulain, P.-M., and Menna, M.: The impact of storm Minerva on the upper layer of the Mediterranean Sea, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-18856, https://doi.org/10.5194/egusphere-egu26-18856, 2026.
