Ocean Modelling最新文献

The prediction of ocean temperature using sea surface data is crucial for studying ocean-related events and climate change. However, current temperature predictions mainly focus on surface data and rarely consider the temporal relationship of ocean temperature. In this study, we propose a novel deep-learning model to predict ocean temperature for the next two months, which fully considers both temporal and spatial features. The input consists of satellite remote sensing data from the past month, including weekly sea surface temperature, salinity, height, and velocity. The model consists of four modules: convolutional, attention, residual, and transposed convolutional. We compare the model estimation with reanalysis data and conduct temporal, spatial, and vertical distribution analyses. The results demonstrate that the model can accurately predict ocean temperature at different lead time, depths, and locations. Finally, we compare the predicted temperature with actual sea observations to ensure the model's good performance in practical applications. This study shows the tremendous potential of the proposed model in predicting 4-D ocean temperature, providing powerful data support for ocean-related events and climate change research.

Storm surges can give rise to extreme floods in coastal areas. The Norwegian Meteorological Institute (MET Norway) produces 120 h regional operational storm surge forecasts along the coast of Norway based on the Regional Ocean Modeling System (ROMS), using a model setup called Nordic4-SS. Despite advances in the development of models and computational capabilities, forecast errors remain large enough to impact response measures and issued alerts, in particular, during the strongest storm events. Reducing these errors will positively impact the efficiency of the warning systems while minimizing efforts and resources. Here, we investigate how forecasts can be improved with residual learning, i.e., training data-driven models to predict the residuals in forecasts from Nordic4-SS. A simple error mapping technique and a more sophisticated Neural Network (NN) method are tested. The simple error mapping technique provides a reduction in the Root Mean Square Error (RMSE) of less than 4%. Using the NN residual correction method, the RMSE in the Oslo Fjord is reduced by 36% for lead times of one hour, 9% for 24 h, and 5% for 60 h. Therefore, the residual NN method is a promising direction for correcting storm surge forecasts, especially on short timescales. Moreover, it is well adapted to being deployed operationally, as (i) the correction is applied on top of the existing model and requires no changes to it, (ii) all predictors used for NN inference are already available operationally, (iii) prediction by the NNs is very fast, typically a few seconds per station, and (iv) the NN correction can be provided to a human expert who may inspect it, compare it with the model output, and see how much correction is brought by the NN, allowing to capitalize on human expertise as a quality validation of the NN output. While no changes to the hydrodynamic model are necessary to calibrate the neural networks, they are specific to a given model and must be recalibrated when the numerical models are updated.

Near real-time surface current measurements from shore-based high-frequency (HF) radars have increasingly proved to be an essential observation for ocean data assimilation (DA) into operational forecasting systems. For the first time in Brazil, a high-resolution operational system was developed assimilating HF ocean currents data. The system comprises a well known ocean model, the Regional Ocean Modeling System (ROMS), applied to the Southeastern Brazilian shelf and oceanic regions. The ROMS Restricted B-preconditioned Lanczos 4D-variational DA method is employed using real-time coastal radar, remote sensing, and in situ observations, and the DA solution is used as initial fields to produce hourly forecasts for the next two days. The performance of the system in providing accurate forecasts by using this source of initial condition (IC) was evaluated in an experiment in which multiple sources of IC were used. In situ and remote sensing data were used to assess the quality of predictions obtained in the forecasting experiments. The results indicate that the employed DA technique significantly reduced the misfit between model and assimilated observations, leading to improved forecast results. By using this IC, the system was capable to provide forecasts with errors reduced by up to 85%, 14%, and 12%, respectively for sea surface temperature, velocities, and heights, compared to forecasts based on global models. The system was also able to accurately predict the positioning and intensity of the Brazil Current flow and its spatiotemporal variability along the studied region.

Accurate prediction of Arctic sea ice is essential for ship navigation. The numerical forecast is an important method to predict sea ice. However, currently, it has significant bias from observation data. In this paper, we propose a deep learning-based bias correction model, Ice-BCNet, to post-process the weekly sea ice concentration (SIC) forecast data of MITgcm to improve its accuracy. Different from the existing bias correction models that only consider spatial features, Ice-BCNet embeds Convlstm into UNet, enabling it to extract spatiotemporal features from SIC forecast data. Ice-BCNet also corrects a monthly scale by iteration. Before the correction, we first assimilate the MASIE-AMSR2 (MASAM2) SIC observation into MITgcm to obtain a better numerical output, which can improve the accuracy of bias correction results. We evaluate the Ice-BCNet from the 2022 hindcasting and 2023 forecasting and use the binary accuracy classification coefficient (BACC) to measure the accuracy of the sea ice edge. We compare Ice-BCNet with statistical corrected methods (Simple Bias Correction, SimBC). The weekly corrected SIC’s average RMSE decreased by over 41%, and Ice-BCNet outperforms SimBC in correcting sea ice near the route. The monthly corrected SIC’s RMSE is below 0.1, with a BACC exceeding 94%. Ice-BCNet also shows a better performance in the extreme case of September 2020.

The water motion computed using 3D and 2DH models in tidally dominated shallow waters can, in some cases, differ significantly. In 2DH models, bed friction is typically parametrised in terms of the depth-averaged velocity, whereas in 3D models, typically the near-bed velocity is used. This difference causes the bed shear stress in 2DH models to point towards the depth-averaged velocity, whereas in 3D models, it points towards the near-bed velocity, which are not necessarily the same. Focussing on linearised barotropic models, we derive an exact friction parametrisation for 2DH models such that the same depth-averaged dynamics are described as in the corresponding 3D model. The result is a convolutional friction formulation where the instantaneous friction depends on the present and past velocities, thus modifying the traditional 2DH friction formulation that only depends on the present depth-averaged velocity. In the case of harmonic (tidal) waves, this parametrisation has a clear physical interpretation and shows that the near-bed velocity should be parametrised as a rotated, deformed and phase shifted variant of the depth-averaged velocity. We demonstrate that in certain regions of the parameter space, it may be impossible to calibrate a 2DH model that uses a traditional friction law to reproduce the water levels from a 3D model, showing that the 3D friction formulation can be crucial to capture the 3D dynamics within a depth-averaged model. This phenomenon is explored in detail in a narrow well-mixed estuary.

An ensemble of experiments based on a ¼° global NEMO configuration is presented, including tidally forced and non-tidal simulations, and using both the default z* geopotential vertical coordinate and the z∼ filtered Arbitrary Lagrangian-Eulerian coordinate, the latter being known to reduce numerical mixing. This is used to investigate the sensitivity of numerical mixing, and the resulting model drifts and biases, to both tidal forcing and the choice of vertical coordinate. The model is found to simulate an acceptably realistic external tide, and the first-mode internal tide has a spatial distribution consistent with estimates from observations and high-resolution tidal models, with vertical velocities in the internal tide of over 50 metres per day. Tidal forcing with the z* coordinate increases numerical mixing in the upper ocean between 30°S and 30°N where strong internal tides occur, while the z∼ coordinate substantially reduces numerical mixing and biases in tidal simulations to levels below those in the z* non-tidal control. The implications for the next generation of climate models are discussed.

The large radial sand ridge (RSR) system located in the southern Yellow Sea near the Jiangsu coast, China, is highly impacted by tropical cyclones (TCs). However, the temporal and spatial variations of sediment dynamics and associated morphodynamics in this region under the influence of TCs have been little explored due to the difficulty of implementing direct observation during these extreme events. Taking typhoon Lekima in August 2019 (No. 1909) as an example, this study simulated and comprehensively investigated the dynamic processes in the RSR area under the impacts of TCs based on the Finite Volume Coastal Ocean Model (FVCOM). During the passage of Lekima, the spatial patterns of residual flow (RF), sediment flux (SF) and morphology changes in the RSR area were totally different from that during the pre- and post-Lekima periods, especially in the offshore areas (the seaward edge of sand ridges). This is because TC Lekima can generate strong wind-driven currents and waves, increasing the bottom stress and influencing the sediment transport. Due to the shallow water depth of RSRs, wave height decreased significantly towards the coast, and tidal effects gradually dominated the nearshore sedimentary dynamic processes instead of wave effects. Furthermore, the effects of TCs with different tracks and intensities were discussed in this study, and we found that TCs passing the west/east side of the study domain can induce opposite directions of sediment transport and lead to the spatial asymmetry of geomorphological evolution. This research can contribute to an improved understanding of sedimentary dynamic processes during extreme events and indicates the importance of exploring sediment dynamics response to TCs with different characteristics for reducing TC-induced coastal risks in future climate change scenarios.

Typically, ocean waves comprise both wind sea and swell systems, each exhibiting different characteristics in terms of decay, propagation, and their impact on engineering. Distinguishing between wind sea and short/long swell systems is critical for both scientific research and engineering applications, such as climate assessment, harbor agitation, and structural design, which has led to a growing interest in studies of multimodal wave climate.

This study investigates the origin and characteristics of multimodal waves based on spectral partitioning and wave system tracking, taking Sri Lanka in the North Indian Ocean as a case study. The data is generated from the spectral wave model WAVEWATCH III. Results show that the wave systems mainly originate from the southwest monsoon, northeast monsoon, southeast trade winds, and southern storm belt. Tropical cyclones can occasionally contribute to multimodal waves. Subsequently, four spectral zones of wave origins are defined according to the joint probability density distribution of partitioned mean wave directions and peak periods estimated by a kernel function. Storm belt waves are responsible for over 36 % of the total wave systems. Finally, two ways of describing wave climate based on the partitioned bulk wave parameters are compared. One offers a thorough understanding of the wave climate from the perspective of wave origin; however, it loses the combined information of wave systems. The other method, while lacking a guaranteed coherence between the wave systems over time, preserves the crucial combined information of wave systems, which is useful in generating a reduced dataset comprising representative cases.

Marine heatwaves (MHWs) are widely recognized as prolonged periods of significantly elevated sea surface temperatures, leading to substantial adverse impacts on marine ecosystems. However, a comprehensive understanding of their characteristics and potential changes under climate change in the South China Sea (SCS, 0 ∼ 25°N, 105 ∼ 125°E) remains insufficient. Here, utilizing the OISST V2.0 reanalysis dataset, our study first examines MHW characteristics and their trends in the SCS during the historical period (1982 ∼ 2014). Then, in accordance with the criteria established in this study, GFDL-ESM4, EC-Earth3-Veg, NESM3, EC-Earth3, and GFDL-CM4 are identified from the CMIP6 ensemble of 19 models for their enhanced simulations of historical MHW characteristics. Moreover, considering that the fixed and sliding threshold methods offer distinct perspectives on the future evolution of MHWs, we employ both approaches to evaluate MHW characteristics under projected scenarios for the future period (2015 ∼ 2100) and subsequently compare the disparities between the two methodologies. The outcomes obtained using these methods consistently indicate that MHWs in the SCS are anticipated to intensify and persist for longer durations in the future. Besides, addressing seasonal variability, the peak intensity of MHWs falls in May during both the historical period and the four projected future scenarios. This study provides valuable insights into the behavior of MHWs in the SCS within the context of climate change, underscoring the urgency of adopting effective mitigation strategies. Especially, the use of two definition methods provides a more comprehensive set of information for understanding the future changes of MHWs in the SCS.