Ocean Modelling最新文献

The spatio-temporal resolution of atmospheric forcing plays a key role in the accuracy of simulated storm surges with hydrodynamic numerical models. Here, we generate five hydrodynamic hindcasts of coastal storm surges along the European Atlantic and the Mediterranean Sea coasts, forced with atmospheric fields of varying temporal (hourly and daily) and spatial (0.25$°$ to 2$°$) resolution since 1940. Our results, that are validated with insitu tide gauge observations, show that storm surges obtained with daily forcing underestimate the magnitude of coastal extreme sea level events by up to 50% compared to hourly simulations and observations. Nevertheless, low-resolution simulations capture the temporal variability of storm surges, including strong episodes. Furthermore, taking advantage of the consistent set of coastal storm surge hindcasts, we demonstrate that storm surges forced with daily mean atmospheric fields, when bias corrected via quantile mapping, provide accurate values of daily maxima as calculated by a high-resolution hindcast. This transformation paves the way to obtain daily maxima storm surge estimates from low-resolution atmospheric fields, as those typically provided by large-scale and global climate models, at a lower computational cost.

Submerged topography in shallow waters is fundamental in the propagation and dissipation of ocean waves in the surf and swash zones. However, obtaining accurate bathymetric data in this region is challenging due to the high temporal and spatial environmental variability. The bottom boundary condition can directly affect the accuracy of numerical models used for shallow water simulations. In this study, the performance of the SWASH numerical model in describing wave runup in the swash zone is assessed using different bathymetric boundary conditions. The first method involves using data measured in the surf zone obtained by a Unmanned Aerial Vehicle (UAV), and analyzing it using the cBathy algorithm. The second method utilizes a regular bathymetric mesh generated from Dean’s equilibrium profile combined with beach topography data. The third method relies exclusively on interpolation methods using data from deep waters and beach profiles. This interpolation approach is the most used among SWASH users when detailed or updated surf zone bathymetry is unavailable. Based on the numerical simulations performed, not incorporating data from the surf zone resulted in a 4% increase in the runup estimated and approximately a 2% difference in identifying the swash zone position. The method to obtain bathymetry through the cBathy algorithm, used in this article, is cost-effective and can be used to reduce uncertainties in surf zone numerical simulations, induced by the lack of knowledge about the bottom conditions.

In this short communication, we report initial success in utilizing existing Explainable Artificial Intelligence (XAI) methodology to investigate an emerging precursor of the El Niño-Southern Oscillation (ENSO), manifest as sea surface temperature anomalies (SSTA) in the Western North Pacific (WNP), and its impact on enhancing ENSO prediction accuracy. Our analysis reveals that integrating WNP SSTA with established XAI techniques significantly increases the predictability of ENSO states. We found marked improvement in prediction accuracy, from a 60 % baseline to over 85 % for forecasting moderate warm, cold, and neutral ENSO states one year ahead. For higher magnitude events, precision surpasses 90 %. This work, intended as a follow-up to recent studies, underscores the potential of augmenting emerging XAI with additional SST signals to advance long-term climate forecasting capabilities.

A realistic representation of the Southern Ocean (SO) in climate models is critical for reliable global climate projections. However, many models are still facing severe biases in this region. Using a fully coupled global climate model at non-eddying (1/2${}^{\circ }$) and strongly eddying (1/10${}^{\circ }$) grid resolution in the SO, we investigate the effect of a 0.5 °C, 1.0 °C and 1.6 °C warmer than observed SO on i) the spin-up behaviour of the high-resolution simulation, and ii) the representation of main dynamical features, i.e., the Antarctic circumpolar current (ACC), the subpolar gyres, the overturning circulation and the Agulhas regime in a quasi-equilibrium state. The adjustment of SO dynamics and hydrography critically depends on the initial state and grid resolution. When initialised with an observed ocean state, only the non-eddying configuration quickly builds up a strong warm bias in the SO. The high-resolution configuration initialised with the biased non-eddying model state results in immense spurious open ocean deep convection, as the biased ocean state is not stable at eddying resolution, and thus causes an undesirable imprint on global circulation. The SO heat content also affects the large-scale dynamics in both low- and high-resolution configurations. A warmer SO is associated with a stronger Agulhas current and a temperature-driven reduction of the meridional density gradient at 45${}^{\circ }$S to 65${}^{\circ }$S and thus a weaker ACC. The eddying simulations have stronger subpolar gyres under warmer conditions while the response in the non-eddying simulations is inconsistent. In general, SO dynamics are more realistically represented in a mesoscale-resolving model at the cost of requiring an own spin-up.

Sea surface temperature (SST) is a highly complex spatiotemporal variable, which stems from its susceptibility to non-linear dynamical processes and substantial spatiotemporal variability. In particular, accurately forecasting small-scale SST is a formidable challenge due to the compounded effects of diverse physical processes spanning across various scales. In this study, we employ deep learning methods to mine the ocean’s evolutionary patterns, as the ocean’s dynamic mechanisms are inherently embedded in spatiotemporal data. We propose a hierarchical stacked spatiotemporal self-attention mechanism (HSSSA) network architecture. The hierarchical stacked encoder–decoder architecture provides the capability for feature fusion and extraction at different scales. The spatial self-attention and temporal self-attention modules simultaneously focus on information from different spatial locations and time steps, allowing the exploration of spatiotemporal patterns in the complex dynamics of the ocean. The experiments are conducted on a high-resolution East China Sea dataset ($1/10°\times 1/10°$) to demonstrate the forecast performance of the proposed model for refined ocean variables. The 15-day forecasts indicate that the HSSSA method outperforms the EOF-ARIMA and CNN-Transformer methods.

In this study, a hydrostatic model - the Navy Coastal Ocean Model (NCOM) is used to analyze the temporal evolution of a cold filament under moderate wind (along / cross filament) and surface cooling forcing conditions. The experimental framework adhered to the setup used in large eddy simulations by Sulllivan and McWilliams (2018). For each forcing scenario, the impact of horizontal resolutions is systematically explored through varies model resolutions of 100 m, 50 m, and 20 m; and the influence of horizontal mixing is investigated by adjusting the Smagorinsky constant within the Smagorinsky horizontal mixing scheme. The role of surface gravity waves is also assessed by conducting experiments both with and without surface wave forcing.

The outcomes of our study revealed that while the hydrostatic model is able to predict the correct characteristics/physical appearance of filament frontogenesis, it fails to capture the precise dynamics of the phenomenon. Horizontal mixing parameterization in the model was found to have marginal effect on frontogenesis, and the frontal arrest is controlled by the model's subgrid-scale artificial regularization procedure instead of horizontal shear instability. Consequently, higher resolution is corresponding to stronger frontogenesis in the model. Thus, whether the hydrostatic model can produce realistic magnitude of frontogenesis is purely dependent on the characteristic of the front/filament simulated and model resolution. Moreover, examination of the parameterized effect of surface gravity wave forcing through vertical mixing unveiled a limited impact on frontogenesis, suggesting that the parameterization falls short in representing the real physics of wave-front interaction.

In this study, we developed and validated two wave ensemble prediction systems (WEPS) to forecast wave conditions along the southeastern coast of Australia. Using the SWAN model (GEN3 ST6), we integrated complex bathymetric features with an unstructured grid and validated model outputs against buoy observations from Sydney, Port Kembla, and Batemans Bay. The two WEPS, SWAN-WW3 and SWAN-Pert, utilize different methodologies: SWAN-WW3 derives boundary conditions from NCEP’s Global Wave Ensemble System, while SWAN-Pert employs Latin Hypercube Sampling for boundary perturbations based on historical data. Our results demonstrate that both systems effectively predict significant wave height ($H_s$), with SWAN-Pert showing improved forecast accuracy in certain metrics compared to SWAN-WW3. Despite underdispersion in spread-skill diagrams, both WEPS exhibited good agreement with observed data. Additionally, rank histograms revealed that SWAN-Pert is more reliable at shorter lead times. This study highlights the potential of integrating statistical sampling methods and ensemble systems for enhancing regional wave forecasting accuracy.