Estuaries are particularly vulnerable to flooding from storm surges, a risk worsened by climate change. While numerical models are essential for flood risk management, most storm surge models rely on atmospheric forcing data with coarse spatial (tens of kilometers) and temporal (hours) resolutions-significantly lower than the model's own grid resolution. This mismatch may compromise prediction accuracy. This study evaluates the impact of the spatial and temporal resolution of atmospheric forcing data on storm surge modeling within the Scheldt river-estuary-North Sea continuum for the record-breaking Storm Xaver (December 2013). Atmospheric forcings were incorporated at spatial resolutions ranging from 2 km to 30 km and at temporal resolutions from 15 min to 6 h. Using an unstructured-mesh multiscale hydrodynamic model, we assessed how these variations influenced the accuracy of storm surge simulations. Our findings indicate that spatial resolution has the greatest influence on model performance, with finer resolutions (2-5 km) improving peak surge predictions in estuarine areas. Temporal resolution enhancements provide additional benefits, but only when combined with high spatial resolution. The impact of temporal refinement diminishes rapidly as spatial resolution coarsens beyond 10 km. Notably, the timing of peak surges remains stable across all resolution combinations. The best results are obtained with 2 km and 15 min atmospheric forcing resolution, while 5 km spatial resolution also shows good performance. This study underscores the importance of aligning atmospheric forcing resolution with the hydrodynamic model's spatial scale to achieve optimal accuracy in storm surge predictions within this estuary.

The re-establishment of seagrass meadows following dieback events depends on the availability of viable propagules, particularly vegetative fragments that facilitate recovery beyond the local meadow through long-distance dispersal. The dispersal of vegetative fragments by ocean currents, waves and wind can be predicted by biophysical models. Among the model parameters, the duration of fragment buoyancy is an important determinant of dispersal but remains poorly quantified for tropical seagrass species. Yet, few empirical studies have assessed fragment dispersal traits and only for a small number of seagrass taxa. This limitation is particularly pronounced in tropical ecosystems, including the Great Barrier Reef (GBR), Australia, where tropical species exhibit diverse life histories and form extensive mixed-species meadows. This study aims to improve the accuracy of biophysical dispersal models for tropical seagrass by generating robust, species-specific data. We quantified the buoyancy duration of fragments from three species-Halophila ovalis, Halodule uninervis, and Zostera muelleri-over 48 days, and assessed whether initial morphological traits influenced buoyancy, finding species type was the primary determinant rather than fragment size. We then incorporated these empirical estimates into a biophysical model to evaluate their effects on dispersal. Our results highlight major differences between species. Z. muelleri floated the longest (24.7 ± 3.0 days); H. uninervis sank the fastest; and H. ovalis was intermediate, generating broken fragments available for further dispersal. Integrating these experimental derived buoyancy values into a biophysical model reduced the mean predicted dispersal distances by 44% on average compared to previous models. These findings highlight interspecific dispersal behaviours and provide useable empirical data to refine future modelling studies. Such improvements are essential for predicting seagrass recovery, guiding restoration site selection, and informing management strategies that maintain connectivity and ecosystem resilience.

Unstructured-mesh ocean models are increasingly used for coastal applications due to their ability to represent complex geometries and apply local grid refinement where needed. However, their broader use has been hindered by their high computational cost, particularly for models based on the Discontinuous Galerkin finite element (DG-FE) method, which involves significantly more degrees of freedom than traditional finite volume or continuous finite element approaches. The rapid emergence of GPU-based high-performance computing architectures now offers a pathway to address this limitation, as DG-FE formulations are inherently well suited to massively parallel, element-wise computations. Here, we present a full 3D DG-FE ocean model implementation optimized for both single- and multi-GPU systems, with support for both NVIDIA and AMD architectures. We detail the computational strategies employed to achieve high performance, including memory layout optimization, kernel-level parallelization, and matrix-free solvers for key vertical processes. Benchmark results demonstrate that a single HPC-grade GPU (e.g. NVIDIA A100) delivers performance equivalent to approximately 1500 CPU cores, while replacing a 128-core CPU node with a 4xA100 GPU node yields a speedup of around 50x. Weak-scaling efficiency is maintained up to 1024 GPUs. We further demonstrate the model's capabilities on a real-world application in the Great Barrier Reef, achieving a spatial resolution five times finer than the most accurate existing model while maintaining a physical-to-numerical time ratio of 100. These results highlight how GPU-accelerated DG-FE methods can dramatically advance the capabilities of unstructured-mesh ocean modeling, enabling ultra-high-resolution coastal simulations that were previously infeasible.

Abstract. Storm surges are a major coastal hazard that can develop within a few hours, so accurate short-term predictions are essential for early warning and emergency response. Operational hydrodynamic models remain too computationally demanding to deliver on-demand, high-frequency forecasts at the required resolution, while purely data-driven models suffer from the limited duration of tide-gauge records. We propose a hybrid machine-learning framework that combines tide-gauge observations from 22 stations along the southern North Sea coast with surge simulations from the COupled Hydrodynamical–Ecological model for REgioNal and Shelf seas (COHERENS) and ERA5 atmospheric forcing. Two architectures — a Transformer and a parallel LSTM+Transformer — are trained on this combined input and compared against (i) the same architectures trained on gauge observations alone and (ii) the COHERENS operational baseline. For 2-hour-ahead nowcasting at 10-minute resolution, the hybrid LSTM+Transformer reaches an average root-mean-square error (RMSE) of 0.055 m (0.033–0.100 m across stations), substantially better than COHERENS (0.164 m; 0.137–0.185 m) and modestly better than the pure data-driven baseline (0.061 m; 0.043–0.100 m); the hybrid Transformer reaches 0.096 m (0.067–0.121 m). The hybrid models also reproduce the magnitude and timing of extreme events well. Applied recursively, the hybrid LSTM+Transformer degrades to an average RMSE of 0.146 m at the 12-hour horizon, remaining well below the COHERENS baseline. Inference takes less than one minute on a single GPU, making the framework directly suitable for operational nowcasting. The approach should generalize to other coastal regions with comparable observational coverage.

Coral reefs are increasingly vulnerable to climate change, and this vulnerability is further exacerbated by land-based pollution that degrades water quality and coral habitat. Assessing reef exposure is challenging because highly variable ocean currents and pollutant discharges make traditional in situ sampling inadequate. To address this, we use the multi-scale ocean model SLIM to simulate the dispersal of seven specific analytes from the four main inlets of southeast Florida between September 2018 and December 2021. Our simulations reveal that while pollutant plumes are primarily transported northward by the Florida Current, southward transport also occurs frequently nearshore, resulting in wider environmental footprints than the predominantly northward Florida Current alone would suggest. Our model results show that the analyte plumes originating from Government Cut and Baker's Haulover inlets extend southward into Biscayne Bay. Because of its southern position and high discharge, Government Cut emerges as the dominant pollution source, affecting nearly all monitored reef sites over the 200 km study extent. Simulated analyte concentrations matched approximately 40% of observed values, confirming the model's utility in capturing the reef’s spatial and temporal exposure patterns. Our results suggest that reducing pollutant loads at Government Cut would significantly mitigate land-based pollution over the coral reefs in the Kristin Jacobs Coral Aquatic Preserve (KJCAP). As exposure to land-based pollution is poorly quantified for most reef systems, the modeling framework developed here could support investigations into the drivers of coral disease and inform targeted management actions to strengthen reef resilience under climate warming.

Abstract. Estuaries are particularly vulnerable to flooding from extreme events such as storm surges, and this vulnerability can be exacerbated by climate change. Numerical models are valuable tools for supporting flood prevention and planning in these regions. However, despite recent improvements in storm surge modeling, most models use atmospheric forcing data with spatial resolutions of several tens of kilometers and temporal resolutions of a few hours, hence much coarser than their own resolution. This discrepancy may have an impact on the overall model accuracy. Here, we evaluate the impact of atmospheric forcing data’s spatial and temporal resolution on storm surge modeling within the Scheldt river-estuary-North Sea sea continuum. Atmospheric forcings were incorporated at spatial resolutions ranging from 2 km to 30 km and at temporal resolutions from 15 minutes to 6 hours. Using an unstructured-mesh multiscale hydrodynamic model, we assessed how these variations influenced the accuracy of storm surge simulations. Our findings indicate that increasing spatial resolution significantly improves the accuracy of peak surge predictions in estuarine areas, while higher temporal resolution further enhances model performance only at the finest spatial resolution. The effect of the temporal resolution diminishes as spatial resolution becomes coarser, suggesting that spatial resolution is more critical for improving storm surge forecasts in estuaries like the Scheldt. The timing of peak surges remained consistent across all configurations. The best results are obtained with 2 km and 15 min atmospheric forcing resolution. This study underscores the importance of aligning atmospheric forcing resolution with the hydrodynamic model's spatial scale to achieve optimal accuracy in storm surge predictions for estuaries.

In the face of rapidly compounding climate change impacts, including ocean acidification (OA), it is critical to understand present-day stress exposure and to anticipate the biogeochemical conditions experienced by vulnerable ecosystems like coral reefs. To meaningfully predict nearshore carbonate chemistry, we must account for the complexity of the local benthic community, as well as connectivity between habitats and relevant endmember carbonate chemistry. Here, we adopt a system-scale approach to predict site-scale effects of benthic metabolism on the carbonate system of the Florida Reef Tract (FRT). We utilize bimonthly carbonate chemistry data from ten cross-shelf transects spanning 250 km of the FRT to model changes in dissolved inorganic carbon (DIC) and total alkalinity (TA). Benthic habitat maps were used to broadly classify communities known to impact carbonate chemistry. A SLIM 2D hydrodynamic model with mesh resolution reaching 100 m over reefs and along the coastline was used to determine the relevant water mass histories and identify the upstream benthic communities shaping local carbonate chemistry. These historical metabolic footprints, or “flowsheds”, were used to build predictive models of the change in DIC and TA at each station. The best predictive models included the chemical impacts of benthic ecosystem metabolism, as defined by water mass trajectories, weighted endmember chemistry, volume, time, and other environmental parameters (light, temperature, salinity, chlorophyll-a, and nitrate). Considering water mass for 5 days prior to sample collection yielded the highest model skill.

The resilience of seagrass meadows strongly depends on the dispersal of their propagules, which fosters recovery and replenishment after disturbances. However, predicting dispersal patterns across dynamic coastal environments and large spatial and temporal scales remains challenging due to the lack of empirical observations. Biophysical models, integrating oceanic and atmospheric drivers with species-specific traits such as buoyancy and lifespan, are commonly used to simulate propagule transport. Yet, few studies account for the interspecific and interannual variability inherent in tropical seagrass ecosystems. Here we present a high-resolution seagrass biophysical dispersal model applied to 11 tropical seagrass species across the entire Great Barrier Reef World Heritage Area (GBRWHA), Australia, and run this model over a 6-year period (2011-2016). We use this model to assess how the interspecific variability in the buoyancy and windage of seagrass propagules affect their dispersal patterns and how these patterns further vary both seasonally and interannually. Our results reveal that species-specific factors such as their windage and buoyancy, as well as the season and region in which they disperse had the largest influence on dispersal distance. H. spinulosa and S. isoetifolium showed the greatest dispersal in the Whitsunday region, while the wet season promoted higher local retention due to lower wind speeds. From a management perspective, this highlights the need to account for species-specific information when devising seagrass management strategies. The outcomes of this research reveal the inherent complexities of predicting multi-species dispersal over large spatial and temporal scales, with broader implications for predicting dispersal in complex coastal ecosystems.

Over the past few decades, Kuwait Bay has experienced significant water quality decline due to growing anthropogenic pressures, including oil and gas extraction and extensive coastal developments, leading to severe eutrophication and marine life mortality. Additionally, the recent construction of a 36 km-long causeway across the Bay and related land reclamation projects has disrupted the Bay's natural flushing processes, allowing pollutants and excess nutrients to accumulate more readily. However, the impact of these new infrastructures on the Bay's circulation patterns and water renewal capacity remains unquantified. Here, we use the multi-scale ocean model SLIM to simulate the fine-scale flow patterns in Kuwait Bay and evaluate water residence time distribution, focusing on its spatial and seasonal variability. By further comparing pre- and post-construction scenarios, we quantify the causeway's influence on Kuwait Bay's hydrodynamics and flushing properties. We find a complete renewal of the Bay within 150-320 days, driven by significant spatial and seasonal variations in water residence time, largely influenced by the prevailing winds and strong tidal flows interacting with the Bay's shallow depths. The introduction of the artificial structures extends the average residence time by only 1.29 days (+3.49 %), but with significant local variations ranging from-66 to +56 days, underlining the causeway's role as a physical barrier, and amplifying the risks of water quality degradation in some regions. From a broader perspective, our findings highlight the large-scale impact of fine-scale hydrodynamic changes in a semi-enclosed coastal system on its flushing processes and water quality.

Coral populations are rapidly declining due to global warming and local anthropogenic stressors, with nearly all living corals at risk if temperatures rise beyond 1.5 ^∘ C. As reversing climate change becomes increasingly unrealistic, effective local actions are essential to mitigate its impacts and support coral recovery through targeted restoration and protection efforts. Biophysical models that simulate coral larval dispersal at reef-scale resolution are crucial for guiding these actions. However, the high computational cost of such models has limited most studies to a few species and spawning events, limiting insights into interannual and interspecific variability. Here, we used the multi-scale ocean model slim to simulate larval dispersal for six key reef-building coral species (Diploria labyrinthiformis, Acropora cervicornis, Pseudodiploria strigosa, Colpophyllia natans, Montastraea cavernosa, and Orbicella faveolata) across Florida’s Coral Reef over a 10-year period (2012–2021), incorporating experimentally calibrated larval dynamics. Our results show that connectivity indicators are most strongly correlated among species with similar spawning windows. Notably, incoming and outgoing connections exhibited the highest interspecific and interannual correlations. By integrating these metrics into a restoration indicator, we identified large clusters of reefs in the Dry Tortugas and northern Broward-Miami regions with significant restoration potential. The in- and out-degrees displayed limited interannual variability, with most fluctuations observed at outer shelf reefs in the Lower Keys and Dry Tortugas, where the ocean circulation is more variable. By providing long-term connectivity estimates for multiple reef-building species, this study offers valuable insights to inform marine conservation strategies.

Acting as a buffer between the Danube and the Black Sea, the Danube Delta plays an important role in regulating the hydro-biochemical flows of this land-sea continuum. Despite its importance, very few studies have focused on the impact of the Danube Delta on the different fluxes between the Danube and the Black Sea. One of the first steps in characterizing this land-sea continuum is to describe the bathymetry of the delta. However, there are no complete, easily accessible bathymetric data on all three branches of the delta to support hydrodynamic, biogeochemical, or ecological studies. In this study, we aim to fill this gap by combining four different datasets, three in the river and one for the riverbanks, each varying in density and spatial distribution, to create a high-resolution bathymetry dataset. The bathymetric data were interpolated on a hybrid curvilinear-unstructured mesh with an anisotropic inverse distance weighting (IDW) interpolation method. The resulting product offers resolutions ranging from 2 m in a connection zone to 100 m in one of the straight unidirectional channels. Cross-validation of the dataset underlined the importance of the data source spatial pattern, with average root mean square errors (RMSEs) of 0.55 %, 6.3 %, and 27.6 % for river segments covered by the densest to coarsest datasets. These error rates are comparable to those observed in bathymetry interpolation in rivers with similar source datasets. The bathymetry presented in this study is the first unique, high-resolution, comprehensive, and easily accessible bathymetric model covering all three branches of the Danube Delta. The dataset is available at 10.5281/zenodo.14055741 .

Egypt, relying heavily on the Nile as its primary water resource, is facing a rising water budget deficit due to increasing consumption, hydroclimatic changes, and upstream river damming. To address the above, innovative management of High Aswan Dam Reservoir (HADR), the third largest artificial reservoir on Earth, and its exchange with the surrounding groundwater system is suggested to develop new agricultural areas. However, the interconnectivity mechanism between the HADR and the fossil Nubian aquifer, the largest transboundary aquifer in Africa, remains speculative due to the lack of in-situ investigations. To address this deficiency, we perform a geophysical survey using aeromagnetic, time-domain electromagnetic, and vertical electrical resistivity sounding in a 330 km(2) pilot area to the northwest of the HADR that is hypothesized to have a dense fracture system that could act as a conduit between these two large water bodies. Our survey results show the presence of normal faults cross the reservoir to the tangential basement and the sedimentary cover that are water-saturated and act as recharges to the Nubian fossil aquifer. These in-situ investigations confirm previous orbital gravity observations by GRACE-FO hypothesizing the interconnectivity between the reservoir and the Nubian aquifer, which was subject to debate. We suggest that such connecting areas between these two water bodies can be optimal sites for future agricultural development using improved management of surface water-groundwater exchanges for irrigation. Finally, our findings highlight upcoming challenges for this linkage if the level of HADR reaches below similar to 160 m above mean sea level (amsl) due to upstream dam operation during the Nile's extended drought periods. Under these conditions, the Nubian aquifer could discharge back into the HADR at the investigated site, changing the water budget of the aquifer and compromising the planned agriculture developments in the adjacent areas, which account for similar to 10 % of the total arable land in Egypt.

Qatar's rapid industrialization, notably in its capital city Doha, has spurred a surge in land reclamation projects, leading to a constriction of the entrance to Doha Bay. By reducing and deflecting the ocean circulation, land reclamation projects have reduced the effective dispersion of wastewater introduced into the bay and hence degraded the water quality. Here, we assess fluctuations in water residence time across three distinct eras (1980, 2000, and 2020) to gauge the impact of successive land reclamation developments. To do this, we couple the multi-scale ocean model SLIM with a Lagrangian model for water residence time within Doha's coastal area. We consider three different topographies of Doha's shoreline to identify which artificial structures contributed the most to increase water residence time. Our findings reveal that the residual ocean circulation in Doha Bay was predominantly impacted by northern developments post-2000. Between 1980 and 2000, the bay's residence time saw a modest rise, of about one day on average. However, this was followed by a substantial surge, of three to six days on average, between 2000 and 2020, which is mostly attributable to The Pearl mega artificial island development. Certain regions of the bay witnessed a tripling of water residence time. Given the ongoing population expansion along the coast, it is anticipated that the growth of artificial structures and coastal reclamation will persist, thereby exacerbating the accumulation of pollutants in the bay. Our findings suggest that artificial offshore structures can exert far-reaching, non-local impacts on water quality, which need to be properly assessed during the planning stages of such developments.

Optimizing hydropower generation from the Nile upstream mega-dams during prolonged droughts while minimizing the downstream water deficit is the cornerstone in resolving the ongoing major water conflict in the Eastern Nile River Basin. A decade of negotiation and mediation has been unsuccessful, mainly due to the hydraulic uncertainties associated with operating the Grand Ethiopian Renaissance Dam during prolonged droughts. Based on the negotiation outcomes, we provide comprehensive assessments of the efficiency of multiple drought-mitigation policies for the impact of dam operation. Our results suggest it can generate almost optimal hydropower without a noticeable downstream deficit during wet, average, and temporary drought flow conditions. For prolonged drought, we identify an ideal operation policy allowing the Grand Ethiopian Renaissance Dam to generate a sustainable energy of 87% of its optimal hydropower without generating additional downstream water deficit. Furthermore, we provide four intermediate policies demonstrating enhanced upstream hydropower generation while minimizing dam-induced downstream water deficits. Our findings attempt to bridge the negotiation disparities in the Nile hydropower mega-dams operations during prolonged drought and foster an actionable and collaborative framework.

Optimizing hydropower generation from mega-dams during prolonged droughts while minimizing downstream water deficits is decisive to resolving the ongoing major conflict on transboundary river management in the highly populous Eastern Nile basin. Our study provides comprehensive assessments of the efficiency of multiple drought-mitigation operation policies based on the outcomes of negotiations. In addition, we develop four novel policies that reduce potential adverse downstream impacts and maximize upstream hydropower generation during prolonged droughts. We use a multi-reservoir hydraulic-energy model with the most up-to-date entries to simulate and optimize the hydropower generation and reservoir level response of the two Nile's largest mega-dams, Grand Ethiopian Resonance Dam (GERD) and Aswan High Dam (AHD), utilizing 100 years of historical flow records. Our results show that, during wet and average flow conditions and a temporary drought, GERD can generate maximum hydropower without a noticeable downstream deficit. However, for prolonged droughts, GERD can still generate sustainable energy from more than 87% of its maximum hydropower while minimizing the dam-induced downstream water budget deficit to a manageable volume. Our up-to-date findings can reduce the negotiations' disparities on operating Nile’s hydropower mega-dams during prolonged drought and help reach a collaborative framework to mitigate the threats of rising hydroclimatic fluctuations.

Since 2014, the stony coral tissue loss disease (SCTLD) has been decimating corals in the Caribbean. Although the trigger of this outbreak remains elusive, evidence suggests waterborne sediment-mediated disease transmission. The outbreak reportedly initiated in September 2014 at a reef site off Virginia Key (VKR), during extensive dredging operations at the Port of Miami. Here we use a high-resolution ocean model to identify the potential driver of the outbreak by simulating the dispersal of dredged sediments, wastewater plumes and disease agents. Our results suggest that VKR could have been impacted by fine sediments produced by dredging operations, especially those involving non-conventional rock-chopping techniques. Wastewater contamination was unlikely. Additionally, our connectivity analysis indicates potential disease transmission from other affected reefs to VKR. Our results therefore suggest that dredging operations might be responsible for the onset of the epidemics. This underscores the need for stricter operational guidelines in future dredging projects.

Climate change poses an existential threat to coral reefs. A warmer and more acidic ocean weakens coral ecosystems and increases the intensity of hurricanes. The wind-wave-current interactions during a hurricane deeply change the ocean circulation patterns and hence potentially affect the dispersal of coral larvae and coral disease agents. Here, we modeled the impact of major hurricane Irma (September 2017) on coral larval and stony coral tissue loss disease (SCTLD) connectivity in Florida's Coral Reef. We coupled high-resolution coastal ocean circulation and wave models to simulate the dispersal of virtual coral larvae and disease agents between thousands of reefs. While being a brief event, our results suggest the passage of hurricane Irma strongly increased the probability of long-distance exchanges while reducing larval supply. It created new connections that could promote coral resilience but also probably accelerated the spread of SCTLD by about a month. As they become more intense, hurricanes' double-edged effect will become increasingly pronounced, contributing to increased variability in transport patterns and an accelerated rate of change within coral reef ecosystems.

Wakatobi National Park (WNP), located in the heart of the Coral Triangle in Indonesia, is one of the most biodiverse marine habitats on Earth. Coral ecosystems within the park, however, are threatened by anthropogenic stressors such as coral diseases, coral mining, blast fishing, invasive species and pollution. This led the Government of Indonesia to establish marine protected zones (MPZs) over 2