Search

Abstract The Southern Ocean's eddy response to changing climate remains unclear, with observations suggesting non‐monotonic changes in eddy kinetic energy (EKE) across scales. Here simulations reappear that smaller‐mesoscale EKE is suppressed while larger‐mesoscale EKE increases with strengthened winds. This change was linked to scale‐wise changes in the kinetic energy cycle, where a sensitive balance between the dominant mesoscale energy sinks—inverse KE cascade, and source—baroclinic energization. Such balance induced a strong (weak) mesoscale suppression in the flat (ridge) channel. Mechanistically, this mesoscale suppression is attributed to stronger zonal jets weakening smaller mesoscale eddies and promoting larger‐scale waves. These EKE multiscale changes lead to multiscale changes in meridional and vertical eddy transport, which can be parameterized using a scale‐dependent diffusivity linked to the EKE spectrum. This multiscale eddy response may have significant implications for understanding and modeling the Southern Ocean eddy activity and transport under a changing climate.

Abstract The Surface Water and Ocean Topography (SWOT) satellite is expected to observe sea surface height (SSH) down to scales approaching ∼15 km, revealing submesoscale patterns that have never before been observed on global scales. Features at these soon‐to‐be‐observed scales, however, are expected to be significantly influenced by internal gravity waves, fronts, and other ageostrophic processes, presenting a serious challenge for estimating surface velocities from SWOT observations. Here we show that a data‐driven approach can be used to estimate the surface flow, particularly the kinematic signatures of smaller scale flows, from SSH observations, and that it performs significantly better than using the geostrophic relationship. We use a Convolutional Neural Network (CNN) trained on submesoscale‐permitting high‐resolution simulations to test the possibility of reconstructing surface vorticity, strain, and divergence from snapshots of SSH. By evaluating success using pointwise accuracy and vorticity‐strain‐divergence joint distributions, we show that the CNN works well when inertial gravity wave amplitudes are relatively weak. When the wave amplitudes are strong, reconstructions of vorticity and strain are less accurate; however, we find that the CNN naturally filters the wave‐divergence, making divergence a surprisingly reliable field to reconstruct. We also show that when applied to realistic simulations, a CNN model pretrained with simpler simulation data performs well, indicating a possible path forward for estimating real flow statistics with limited observations.

Abstract The thickness‐weighted average (TWA) framework, which treats the residual‐mean flow as the prognostic variable, provides a clear theoretical formulation of the eddy feedback onto the residual‐mean flow. The averaging operator involved in the TWA framework, although in theory being an ensemble mean, in practice has often been approximated by a temporal mean. Here, we analyze an ensemble of North Atlantic simulations at mesoscale‐permitting resolution (1/12°). We therefore recognize means and eddies in terms of ensemble means and fluctuations about those means. The ensemble dimension being orthogonal to the temporal and spatial dimensions negates the necessity for an arbitrary temporal or spatial scale in defining the eddies. Eddy‐mean flow feedbacks are encapsulated in the Eliassen‐Palm (E‐P) flux tensor and its convergence indicates that eddy momentum fluxes dominate in the separated Gulf Stream. The eddies can be interpreted to contribute to the zonal meandering of the Gulf Stream and a northward migration of it in the meridional direction. Downstream of the separated Gulf Stream in the North Atlantic Current region, the interfacial form stress convergence becomes leading order in the E‐P flux convergence.

The Southern Ocean is an important sink of carbon via the biological pump. Here authors run high-resolution physical/biogeochemical simulations of an open-Southern Ocean ecosystem forced with a realistic seasonal cycle and confirm that (sub)mesoscale iron transport across the mixing-layer base sustains primary productivity.

Abstract Biological productivity in the Southern Ocean is limited by iron availability. Previous studies of iron supply have focused on mixed‐layer entrainment and diapycnal fluxes. However, the Southern Ocean is a region highly energetic mesoscale and submesoscale turbulence. Here we investigate the role of eddies in supplying iron to the euphotic zone, using a flat‐bottom zonally re‐entrant model, configured to represent the Antarctic Circumpolar Current region, that is coupled to a biogeochemical model with a realistic seasonal cycle. Eddies are admitted or suppressed by changing the model's horizontal resolution. We utilize cross spectral analysis and the generalized Omega equation to temporally and spatially decompose the vertical transport attributable to mesoscale and submesoscale motions. Our results suggest that the mesoscale vertical fluxes provide a first‐order pathway for transporting iron across the mixing‐layer base, where diapycnal mixing is weak, and must be included in modeling the open‐Southern Ocean iron budget.

Abstract We investigate the role of small‐scale, high‐frequency motions on lateral transport in the ocean, by using velocity fields and particle trajectories from an ocean general circulation model (MITgcm‐llc4320) that permits submesoscale flows, inertia‐gravity waves, and tides. Temporal averaging/filtering removes most of the submesoscale turbulence, inertia‐gravity waves, and tides, resulting in a largely geostrophic flow, with a rapid drop‐off in energy at scales smaller than the mesoscales. We advect two types of Lagrangian particles: (a) 2‐D particles (surface restricted) and (b) 3‐D particles (advected in full three dimensions) with the filtered and unfiltered velocities and calculate Lagrangian diagnostics. At large length/time scales, Lagrangian diffusivity is comparable for filtered and unfiltered velocities, while at short scales, unfiltered velocities disperse particles much faster. We also calculate diagnostics of Lagrangian coherent structures:rotationally coherent Lagrangian vortices detected from closed contours of the Lagrangian‐averaged vorticity deviation and material transport barriers formed by ridges of maximum finite‐time Lyapunov exponent. For temporally filtered velocities, we observe strong material coherence, which breaks down when the level of temporal filtering is reduced/removed, due to vigorous small‐scale mixing. In addition, for the lowest temporal resolution, the 3‐D particles experience very little vertical motion, suggesting that degrading temporal resolution greatly reduces vertical advection by high‐frequency motions. Our study suggests that Lagrangian diagnostics based on satellite‐derived surface geostrophic velocity fields, even with higher spatial resolutions as in the upcoming Surface Water and Ocean Topography mission, may overestimate the presence of mesoscale coherent structures and underestimate dispersion.

Due to the geostrophic balance, horizontal divergence-free is often assumed when analyzing large-scale oceanic flows. However, the geostrophic balance is a leading-order approximation. We investigate the statistical feature of weak horizontal compressibility in the Gulf of Mexico by analyzing drifter data (the Grand LAgrangian Deployment (GLAD) experiment and the LAgrangian Submesoscale ExpeRiment (LASER)) based on the asymptotic probability density function of the angle between velocity and acceleration difference vectors in a strain-dominant model. The results reveal a notable divergence at scales between 10 km and 300 km, which is stronger in winter (LASER) than in summer (GLAD). We conjecture that the divergence is induced by wind stress with its curl parallel to the Earth's rotation. [ABSTRACT FROM AUTHOR]

The article discusses a workshop funded by the NSF's Office of Polar Programs to address research priorities for observing the dynamics of the Southern Ocean. The Southern Ocean plays a crucial role in global climate, but its dynamics remain uncertain, impacting climate projections. The workshop identified research priorities, such as understanding ice shelf melt rates, sea ice variability, and ocean circulation, and emphasized the need for observational strategies to address these challenges. Additionally, the article highlights the importance of equity, diversity, and inclusion in fieldwork, data management, and international collaboration within the Southern Ocean research community. [Extracted from the article]

Ocean mesoscale eddies are often poorly represented in climate models, and therefore, their effects on the large scale circulation must be parameterized. Traditional parameterizations, which represent the bulk effect of the unresolved eddies, can be improved with new subgrid models learned directly from data. Zanna and Bolton (2020), https://doi.org/10.1029/2020gl088376 (ZB20) applied an equation‐discovery algorithm to reveal an interpretable expression parameterizing the subgrid momentum fluxes by mesoscale eddies through the components of the velocity‐gradient tensor. In this work, we implement the ZB20 parameterization into the primitive‐equation GFDL MOM6 ocean model and test it in two idealized configurations with significantly different dynamical regimes and topography. The original parameterization was found to generate excessive numerical noise near the grid scale. We propose two filtering approaches to avoid the numerical issues and additionally enhance the strength of large‐scale energy backscatter. The filtered ZB20 parameterizations led to improved climatological mean state and energy distributions, compared to the current state‐of‐the‐art energy backscatter parameterizations. The filtered ZB20 parameterizations are scale‐aware and, consequently, can be used with a single value of the non‐dimensional scaling coefficient for a range of resolutions. The successful application of the filtered ZB20 parameterizations to parameterize mesoscale eddies in two idealized configurations offers a promising opportunity to reduce long‐standing biases in global ocean simulations in future studies. Plain Language Summary: This research focuses on improving the accuracy of ocean models by addressing the challenges of representing the mesoscale eddies on coarse grids. These eddies play a crucial role in the Earth's climate system, but traditional climate models struggle to capture their effects. Here, we implemented a new data‐driven parameterization simulating the physics of the mesoscale eddies into the state‐of‐the‐art ocean model. The parameterization is interpretable and captures key physical processes related to the mesoscale eddies known as energy backscatter. We tested this parameterization in two idealized ocean scenarios and found that it significantly improves the biases in the representation of the mean state and energetics. We propose new filtering schemes which improve the physical and numerical properties of the parameterization. Accurate representation of the mesoscale eddies by the present scheme has the potential to resolve long‐standing biases present in global ocean models and thus allow for more reliable climate simulations. Key Points: A data‐driven mesoscale eddy parameterization is implemented and evaluated in different configurations of the GFDL MOM6 ocean modelWe introduce filtering schemes to reduce the generation of grid‐scale noise and enhance the large‐scale backscatterThe subgrid parameterization improves the representation of the energy distributions and the climatological mean state [ABSTRACT FROM AUTHOR]

On behalf of the journal, AGU, and the scientific community, the editors of Geophysical Research Letters would like to sincerely thank those who reviewed manuscripts for us in 2023. The hours reading and commenting on manuscripts not only improve the manuscripts, but also increase the scientific rigor of future research in the field. With the advent of AGU's data policy, many reviewers have also helped immensely to evaluate the accessibility and availability of data, and many have provided insightful comments that helped to improve the data presentation and quality. We greatly appreciate the assistance of the reviewers in advancing open science, which is a key objective of AGU's data policy. We particularly appreciate the timely reviews in light of the demands imposed by the rapid review process at Geophysical Research Letters. We received 4,512 submissions in 2023 and 5,112 reviewers contributed to their evaluation by providing 8,587 reviews in total. We deeply appreciate their contributions. Plain Language Summary: Individuals in italics provided three or more reviews for GRL in 2023. Key Points: The editors thank the 2023 peer‐reviewers [ABSTRACT FROM AUTHOR]

The Southern Ocean's eddy response to changing climate remains unclear, with observations suggesting non‐monotonic changes in eddy kinetic energy (EKE) across scales. Here simulations reappear that smaller‐mesoscale EKE is suppressed while larger‐mesoscale EKE increases with strengthened winds. This change was linked to scale‐wise changes in the kinetic energy cycle, where a sensitive balance between the dominant mesoscale energy sinks—inverse KE cascade, and source—baroclinic energization. Such balance induced a strong (weak) mesoscale suppression in the flat (ridge) channel. Mechanistically, this mesoscale suppression is attributed to stronger zonal jets weakening smaller mesoscale eddies and promoting larger‐scale waves. These EKE multiscale changes lead to multiscale changes in meridional and vertical eddy transport, which can be parameterized using a scale‐dependent diffusivity linked to the EKE spectrum. This multiscale eddy response may have significant implications for understanding and modeling the Southern Ocean eddy activity and transport under a changing climate. Plain Language Summary: The response of eddies in the Southern Ocean to climate change is not well understood. In this study, we used a channel model that simulates the effects of wind on eddies. We found that smaller eddies have less kinetic energy (KE) when the winds are stronger. On the other hand, larger‐scale eddies have more KE with stronger winds. Similar phenomena are also observed in the observations. By analyzing the eddy's KE budget, the interaction between different scales of eddies and the interaction between the eddies and mean flow are strengthened when the winds get stronger. This leads to a reduction of eddy KE at smaller mesoscale scales and an increase at larger scales. From the observational view, stronger winds weaken smaller eddies and promote larger waves. This change in eddy KE also affects how eddies meridionally transport materials and how eddy diffusivity varies at different scales. Smaller eddies transport materials less when their KE is weakened, while larger eddies become stronger in transporting materials. These findings determine how eddy diffusivity responds to the changed eddy KE at different scales. The multi‐scale response of eddies to wind has important implications for understanding the behavior of Southern Ocean eddies in a changing climate. Key Points: Larger eddies got stronger and smaller eddies got weaker as Southern Ocean westerlies strengthenedBoth flat and ridge channel simulations suggest that these changes may be linked to changes in the inverse energy cascadeThe corresponding changes in scale‐wise meridional and vertical transport are also non‐monotonic [ABSTRACT FROM AUTHOR]

We parameterize sub‐grid scale (SGS) fluxes in sinusoidally forced two‐dimensional turbulence on the β‐plane at high Reynolds numbers (Re ∼25,000) using simple 2‐layer convolutional neural networks (CNN) having only O(1000) parameters, two orders of magnitude smaller than recent studies employing deeper CNNs with 8–10 layers; we obtain stable, accurate, and long‐term online or a posteriori solutions at 16× downscaling factors. Our methodology significantly improves training efficiency and speed of online large eddy simulations runs, while offering insights into the physics of closure in such turbulent flows. Our approach benefits from extensive hyperparameter searching in learning rate and weight decay coefficient space, as well as the use of cyclical learning rate annealing, which leads to more robust and accurate online solutions compared to fixed learning rates. Our CNNs use either the coarse velocity or the vorticity and strain fields as inputs, and output the two components of the deviatoric stress tensor, Sd. We minimize a loss between the SGS vorticity flux divergence (computed from the high‐resolution solver) and that obtained from the CNN‐modeled Sd, without requiring energy or enstrophy preserving constraints. The success of shallow CNNs in accurately parameterizing this class of turbulent flows implies that the SGS stresses have a weak non‐local dependence on coarse fields; it also aligns with our physical conception that small‐scales are locally controlled by larger scales such as vortices and their strained filaments. Furthermore, 2‐layer CNN‐parameterizations are more likely to be interpretable. Plain Language Summary: In this study, we demonstrate that simple, shallow neural networks can be used to effectively model complex turbulent flows in the atmosphere and oceans. By using these simpler NNs, we can improve the efficiency of our simulations and better understand the underlying physics of turbulent flows. We also explore different training techniques to make these models more accurate and robust. Our findings suggest that the stress in these turbulent flows has only a weak spatial dependence on larger‐scale features, which has important implications for our understanding of how turbulence behaves. Overall, our work can help improve climate and weather models by providing a more efficient and interpretable way to simulate turbulence. Key Points: Shallow convolutional neural networks (CNNs) accurately parameterize high Reynolds number forced2D turbulence, with efficient training and high online large eddy simulations accuracyExtensive hyperparameter searching and cyclical learning rates yield robust and accurate online solutions for CNN‐based turbulence modelsThe success of shallow CNNs implies nearly‐local dependence of SGS stresses providing insights into turbulent flow interactions [ABSTRACT FROM AUTHOR]

Global estimates of mesoscale vertical velocity remain poorly constrained due to a historical lack of adequate observations on the spatial and temporal scales needed to measure these small magnitude velocities. However, with the wide‐spread and frequent observations collected by the Argo array of autonomous profiling floats, we can now better quantify mesoscale vertical velocities throughout the global ocean. We use the underutilized trajectory data files from the Argo array to estimate the time evolution of isotherm displacement around a float as it drifts at 1,000 m, allowing us to quantify vertical velocity averaged over approximately 4.5 days for that depth level. The resulting estimates have a non‐normal, high‐peak, and heavy‐tail distribution. The vertical velocity distribution has a mean value of (1.9 ± 0.02) × 10−6 m s−1 and a median value of (1.3 ± 0.2) × 10−7 m s−1, but the high‐magnitude events can be up to the order of 10−4 m s−1. We find that vertical velocity is highly spatially variable and is largely associated with a combination of topographic features and horizontal flow. These are some of the first observational estimates of mesoscale vertical velocity to be taken across such large swaths of the ocean without assumptions of uniformity or reliance on horizontal divergence. Plain Language Summary: Vertical velocity in the ocean is a fundamental part of how water circulates throughout the globe. This impacts the temperature, salt, nutrients, and currents that make up the ocean. However, vertical velocities are very small and are, therefore, difficult to measure. In particular, the vertical velocities of ocean events that occur on roughly a weekly to monthly time scale (mesoscale) are poorly understood. We have developed a method for estimating these mesoscale vertical velocities across the globe using an array of autonomous robots called Argo floats. Our results show that vertical velocities vary greatly depending on location, with the largest values occurring where there is a combination of relatively shallow ocean depths and larger horizontal velocities. These estimates are some of the first of their kind to be made from observations across such large swaths of the ocean. Key Points: Five‐day averaged vertical velocities from Argo observations near 1,000 m are non‐normally distributed, with a high peak and heavy tailsMesoscale vertical velocities are on the order of centimeters per day, but high‐magnitude events can be on the order of meters per dayVertical velocities estimated from Argo floats are spatially variable and correlated with topographic features and horizontal surface flow [ABSTRACT FROM AUTHOR]

Oceanic macroturbulence is efficient at stirring and transporting tracers. The dynamical properties of this stirring can be characterized by statistically quantifying tracer structures. Here, we characterize the macroscale (1–100 km) tracer structures observed by two Seagliders downstream of the Southwest Indian Ridge in the Antarctic Circumpolar Current (ACC). These are some of the first glider observations in an energetic standing meander of the ACC, a region associated with enhanced ventilation. The small‐scale density variance in the mixed layer (ML) was relatively enhanced near the surface and base of the ML, while being muted at mid‐depth in the ML, suggesting the formation mechanism to be associated with ML instabilities and eddies. In addition, ML density fronts were formed by comparable contributions from temperature and salinity gradients. In the interior, along‐isopycnal spectra and structure functions of spice indicated that there is relatively lower variance at smaller scales than would be expected based on non‐local stirring, suggesting that flows smaller than the deformation radius play a role in the cascade of tracers to small scales. These interior spice anomalies spanned across isopycnals, and were found to be about 3–5 times flatter than the aspect ratio that would be expected for O(1) Burger number flows like interior QG dynamics, suggesting the ratio of vertical shear to horizontal strain is greater than N/f. This further supports that small‐scale flows, with high‐mode vertical structures, impact tracer distributions. Plain Language Summary: Ocean eddies and fronts are efficient at stirring and generating tracer structures spanning a wide range of scales. This stirring is important for uptake of tracers, particularly in the Southern Ocean where sloping density surfaces provide a direct pathway from the surface into the interior. While the qualitative nature of stirring seems clear, quantitatively characterizing properties of this stirring is necessary to build accurate parameterizations and understand how the ocean is ventilated. One way to characterize this is by quantifying the distribution of the tracer concentration as a function of scale. Here, we do this characterization by analyzing data collected by two gliders in a region where the mean current of the Southern Ocean meanders as it flows over topography, and is likely to be a hot spot for ventilation. This analysis reveals that relatively small eddies generated by a process known as mixed layer instability are active in the mixed layer and play an important role in setting the near‐surface stratification. Surprisingly, tracer structure below the mixed layer also indicates that flows with small horizontal and vertical scales play a role in stirring. This challenges conventional wisdom, which assumes that only the largest eddies are important for stirring. Key Points: Glider observations were used to analyze tracer structure at scales ≤100 km in the Antarctic Circumpolar CurrentThe vertical structure of density variance in the mixed layer (ML) is suggestive of ML instabilities and eddiesSmall‐scale flows contribute to stirring in the interior, as indicated by passive tracer structure [ABSTRACT FROM AUTHOR]

The Surface Water and Ocean Topography (SWOT) satellite is expected to observe sea surface height (SSH) down to scales approaching ∼15 km, revealing submesoscale patterns that have never before been observed on global scales. Features at these soon‐to‐be‐observed scales, however, are expected to be significantly influenced by internal gravity waves, fronts, and other ageostrophic processes, presenting a serious challenge for estimating surface velocities from SWOT observations. Here we show that a data‐driven approach can be used to estimate the surface flow, particularly the kinematic signatures of smaller scale flows, from SSH observations, and that it performs significantly better than using the geostrophic relationship. We use a Convolutional Neural Network (CNN) trained on submesoscale‐permitting high‐resolution simulations to test the possibility of reconstructing surface vorticity, strain, and divergence from snapshots of SSH. By evaluating success using pointwise accuracy and vorticity‐strain‐divergence joint distributions, we show that the CNN works well when inertial gravity wave amplitudes are relatively weak. When the wave amplitudes are strong, reconstructions of vorticity and strain are less accurate; however, we find that the CNN naturally filters the wave‐divergence, making divergence a surprisingly reliable field to reconstruct. We also show that when applied to realistic simulations, a CNN model pretrained with simpler simulation data performs well, indicating a possible path forward for estimating real flow statistics with limited observations. Plain Language Summary: Satellite measurements of sea surface height (SSH) have for the past few decades provided weekly global estimates of upper ocean currents at scales larger than approximately 100 km. The new Surface Water and Ocean Topography satellite promises to improve the resolution of these SSH observations. However, these new observations will introduce a new challenge, since a simple physics‐based diagnostic relationship does not exist between the SSH and upper ocean currents for the finer scales that will now be visible. Here we show that a neural network can be used to estimate the surface flow from SSH observations. In particular, our trained neural networks are able to use SSH to predict aspects of upper‐ocean currents that may be particularly useful for improving estimates of oceanic transport of heat and other tracers. Key Points: Neural networks reasonably reconstruct surface vorticity, strain and divergence, from sea surface heightNeural networks naturally filter wave divergence, leaving only the desired divergence associated with frontsTransfer learning shows promise when task‐specific data is limited but data from reasonably close simulations is available [ABSTRACT FROM AUTHOR]

Ocean models at intermediate resolution (1/4°), which partially resolve mesoscale eddies, can be seen as Large eddy simulations of the primitive equations, in which the effect of unresolved eddies must be parameterized. In this work, we propose new subgrid models that are consistent with the physics of two‐dimensional flows. We analyze subgrid fluxes in barotropic decaying turbulence using Germano (1986, https://doi.org/10.1063/1.865568) decomposition. We show that Leonard and Cross stresses are responsible for the enstrophy dissipation, while the Reynolds stress is responsible for additional kinetic energy (KE) backscatter. We utilize these findings to propose a new model, consisting of three parts, that is compared to a baseline dynamic Smagorinsky model. The three‐component model accurately simulates the spectral transfer of energy and enstrophy and improves the representation of KE spectrum, resolved KE and enstrophy decay in a posteriori experiments. The backscattering component of the new model (Reynolds stress) is implemented both in quasi‐geostrophic and primitive equation ocean models and improves statistical characteristics, such as the vertical profile of eddy KE, meridional overturning circulation and cascades of kinetic and potential energy. Plain Language Summary: Ocean models at intermediate resolution contain missing physics term that accounts for the contribution of unresolved mesoscale eddies, which needs to be parameterized. Mesoscale eddies obey complex physics which should be accounted for when proposing a parameterization. Here we consider the interscale transfer of kinetic energy and enstrophy in a barotropic fluid and propose new subgrid models which capture this transfer. Our strategy is to split the subgrid contribution into three parts and propose a model for each term separately. This approach results in excellent a priori performance and improves online simulations. We demonstrate that our analysis of subgrid fluxes generalizes well across flow regimes: the new parameterization of energy redistribution improves barotropic, quasi‐geostrophic and primitive equation ocean models. Key Points: We propose a three‐component subgrid model consistent with the physics of two‐dimensional fluids using Germano (1986, https://doi.org/10.1063/1.865568) decompositionThe new subgrid model accurately predicts the spectral transfer of energy and enstrophy and improves a posteriori experimentsA backscattering component (Reynolds stress) improves coarse‐grid ocean models based on quasi‐geostrophic and primitive equations [ABSTRACT FROM AUTHOR]

Ocean dynamics at the submesoscale play a key role in mediating upper‐ocean energy dissipation and dispersion of tracers. Observations of ocean currents from synoptic mesoscale surveys at submesoscale resolution (250 m–100 km) from a novel airborne instrument (MASS DoppVis) reveal that the kinetic energy spectrum in the California Current System is nearly continuous from 100 km to sub‐kilometer scales, with a k−2 spectral slope. Although there is not a transition in the kinetic energy spectral slope, there is a transition in the dynamics to non‐linear ageostrophic interactions at scales of O $\mathcal{O}$(1 km). Kinetic energy transfer across spatial scales is enabled by interactions between the rotational and divergent components of the flow field at the submesoscale. Kinetic energy flux is patchy and localized at submesoscale fronts. Kinetic energy is transferred both downscale and upscale from 1 km in the observations of a cold filament. Plain Language Summary: Ocean dynamics at scales of 100 m–10 km, called the submesoscale, are important because they are associated with large velocity gradients and non‐linear interactions. Large gradients lead to vertical velocity, which facilitates ocean‐atmosphere interactions and ocean biological processes. Velocity gradients and non‐linear processes combine to transfer kinetic energy from the large‐scale flow to small‐scale perturbations. This can lead to instabilities that dissipate energy in the ocean surface layer (rather than the seafloor). Here we analyze novel observations that provide insight into ocean dynamics through the distributions of velocity gradients and energy transfer at 1 km scale. Dynamics at these scales have previously been modeled, but have not been observed directly. We observe a transition where non‐linear dynamics become more important at scales of order 10 km. We also introduce new interpretations of spectral analysis (analysis of energy and correlations across scales). Moreover, we analyze covariance of velocity gradient quantities and flow energetics to demonstrate that energy flux is episodic and localized at fronts. Together, these observations demonstrate that fronts play an important role in boundary‐layer kinetic energy processes and highlight the evolution of upwelling filaments. Key Points: Remote sensing observations reveal a kinetic energy spectrum with a continuous slope from 100 to 1 km in an eastern boundary regionBetween 1 and 10 km, ageostrophic non‐linear interactions become dynamically importantCross‐scale kinetic energy transfers computed from 2D velocity observations are associated with shear strain in the observed front [ABSTRACT FROM AUTHOR]

Using a probabilistic neural network and Lagrangian observations from the Global Drifter Program, we model the single particle transition probability density function (pdf) of ocean surface drifters. The transition pdf is represented by a Gaussian mixture whose parameters (weights, means, and covariances) are continuous functions of latitude and longitude determined to maximize the likelihood of observed drifter trajectories. This provides a comprehensive description of drifter dynamics allowing for the simulation of drifter trajectories and the estimation of a wealth of dynamical statistics without the need to revisit the raw data. As examples, we compute global estimates of mean displacements over 4 days and lateral diffusivity. We use a probabilistic scoring rule to compare our model to commonly used transition matrix models. Our model outperforms others globally and in three specific regions. A drifter release experiment simulated using our model shows the emergence of concentrated clusters in the subtropical gyres, in agreement with previous studies on the formation of garbage patches. An advantage of the neural network model is that it provides a continuous‐in‐space representation and avoids the need to discretize space, overcoming the challenges of dealing with nonuniform data. Our approach, which embraces data‐driven probabilistic modeling, is applicable to many other problems in fluid dynamics and oceanography. Plain Language Summary: Data collected using satellite tracked drifting buoys (drifters) have been used to estimate statistics of ocean dynamics. The estimation procedures employed are typically developed on a case‐by‐case basis and provide a limited characterization of the dynamics. We show that, by modeling the full probability distribution from which these statistics derive, we can estimate efficiently a range of statistics simultaneously and provide a more comprehensive description. To do this we leverage probabilistic neural networks. Our model provides a convenient means of estimating statistics, and can be used to simulate drifter trajectories. Our simulations reproduce the clustering behavior in ocean gyres seen in previous studies on the formation of garbage patches. Key Points: We use probabilistic neural networks to learn the transition probability density of surface drifters from observationsWe show that modeling the full transition density is an efficient way to estimate many dynamical statistics simultaneouslyOur model can also be used to simulate transport at the ocean near‐surface and achieves higher skill scores than existing models [ABSTRACT FROM AUTHOR]

Isopycnal eddy mixing across continental slopes profoundly modulates the ocean circulation and biogeochemistry. Yet this process must be parameterized in coarse‐resolution ocean models via an isopycnal eddy diffusivity prescribed with the Redi scheme. In this work, we evaluate the skill of physics‐based and data‐driven Redi variants in predicting the cross‐slope exchanges using a suite of offline‐mode parameterized tracer simulations for wind‐driven upwelling continental slope fronts, which commonly arise around the margins of subtropical gyres. The tested physics‐based Redi variants range from a constant eddy diffusivity to a recently proposed, bathymetry‐aware diffusivity augmented by the artificial neural network (ANN) that infers the mesoscale eddy kinetic energy from the mean flow and topographic quantities. Moreover, a purely data‐driven eddy diffusivity is learned by the ANN from the output data set of an eddy‐resolving model, whose solutions serve as the ground truth against which the parameterized tracer simulations are compared. Among all tested Redi variants, the ANN‐learned diffusivity and the bathymetry‐aware diffusivity outperform others in reproducing the tracer solutions of the eddy‐resolving model. However, a physics‐based Redi variant with local deficiencies can introduce global errors in the predicted tracer distribution, which calls for ongoing efforts in constraining the shelf‐to‐ocean transition of the isopycnal eddy diffusivity. A purely data‐driven diffusivity can nearly reproduce the diagnosed diffusivity from the eddy‐resolving model, which highlights the efficacy of machine learning techniques for parameterizing eddy processes across steep topography. This work serves as a key step toward parameterizing the isopycnal eddy mixing in ocean models with continental slopes. Plain Language Summary: Turbulent eddies across continental slopes drive the exchanges between shelf seas and open oceans, which are essential for coastal ecosystems and the global climate. These eddies are characterized by length scales smaller than the grid spacing of today's ocean climate models, resulting in unresolved eddy‐induced exchanges. This issue is usually remedied by eddy parameterizations that infer eddy fluxes from properties explicitly resolved in climate models. In this work, we evaluate extant parameterizations in constraining cross‐slope exchanges of oceanic tracers (such as salt and nutrients) by comparing simulations that utilize parameterizations with simulations that can explicitly resolve mesoscale eddies. Our results reveal the necessity to adopt more sophisticated parameterizations in non‐eddying simulations for cross‐slope eddy tracer exchanges than those currently used in climate models. Moreover, machine learning techniques are found to be effective in parameterizing cross‐slope eddy exchanges. Key Points: Data‐driven and physics‐based parameterizations of isopycnal eddy diffusivity across continental slopes are tested prognosticallyArtificial neural network‐inferred diffusivities and bathymetry‐aware diffusivities outperform previously tested variants over steep slopesMachine learning techniques effectively augment physics‐based mesoscale eddy parameterizations [ABSTRACT FROM AUTHOR]

Oceanic submesoscale dynamics with horizontal scales < 20 km have similar temporal and spatial scales as internal gravity waves (IGWs), but they differ dynamically and have distinct impacts on the ocean. Separating unbalanced submesoscale motions (USMs), quasi-balanced submesoscale motions (QBMs), and IGWs in observations remains a great challenge. Based on the wave–vortex decomposition and the vertical scale separation approach for distinguishing IGWs and USMs, the long-term repeat Oleander observations in the Gulf Stream region provide an opportunity to quantify these processes separately. Here in this study, the role of USMs in the divergence is emphasized, which has confounded the wave–vortex decomposition of wintertime data in previous analyses. We also adopt the vertical filtering approach to identify the USMs by applying a high-pass filter to the vertical scales, as USMs are characterized by smaller vertical scales. This approach is tested with submesoscale-permitting model data to confirm its effectiveness in filtering the submesoscale velocity perturbations, before being applied to the compiled velocity data of the Oleander dataset (years 2005–18). The results show that the averaged submesoscale eddy kinetic energy by USMs can reach ∼1 × 10−3 m² s−2 at z = −30 m in winter, much stronger than found in other seasons. Importantly, this study exemplifies the possibility of obtaining USMs from existing ADCP observations and reveals the seasonal dynamical regimes for the submesoscales. [ABSTRACT FROM AUTHOR]