Imperial College London

ProfessorPavelBerloff

Faculty of Natural SciencesDepartment of Mathematics

Professor in Applied Mathematics
 
 
 
//

Contact

 

+44 (0)20 7594 9662p.berloff Website

 
 
//

Location

 

745Huxley BuildingSouth Kensington Campus

//

Summary

 

Publications

Publication Type
Year
to

91 results found

Davies J, Shevchenko I, Berloff P, Sutyrin GGet al., 2024, Linear instability and weakly nonlinear effects in eastward dipoles, Physica D: Nonlinear Phenomena, Vol: 460, ISSN: 0167-2789

The linear instability and weakly nonlinear dynamics of eastward-propagating, steady-state Larichev–Reznik vortex dipoles are explored in terms of two-dimensional normal-mode analysis. To extract the fastest growing normal modes, we apply both breeding methodology based on solving the initial-value problem, as well as a direct-solution approach through the full-spectrum eigenproblem involving large matrices. We find that the amplification rate of dipole instability decreases with respect to increase in dipole intensity. In our study, both approaches yield consistent results and are systematically compared to provide guidance for further studies of vortex structures. We consider nonlinear self-interaction of the fastest growing mode, along with the induced eddy fluxes, their divergence and mechanical energy balance. Through this analysis, we find that the unstable mode leads to weakening of the dipole by extracting its energy and exchanging potential vorticity content in the down-gradient sense, thus, providing a nonlinear physical mechanism for the dipole destruction. In particular, we highlight the fundamental importance of the west-east asymmetry of the normal mode for destruction to be realized. In summary, we consider this work to be a foundational demonstration of useful methodology for future studies of dynamics and stability of isolated vortices without simplifying spatial symmetries, such as ubiquitous vortices in geophysical fluids.

Journal article

Meacham J, Berloff P, 2024, Clustering as a mechanism for enhanced reaction of buoyant species, Journal of Marine Systems, Vol: 243, ISSN: 0924-7963

Buoyant material has a tendency to form dense clusters at the ocean surface. This has been observed in distributions of marine life and floating plastic contaminants. The main mechanism behind this is that particles with positive/neutral buoyancy do not behave as passive tracers in stratified flows. It could be expected that coextensive clustering between plankton and toxic ocean contaminants could lead to enhanced ecological risk. However, such interactions cannot be sufficiently modelled in a standard passive tracer approximation. Given the large uncertainty in the form of converging currents and how to model interactions of buoyant tracers, we opt for an idealised modelling approach. The simplicity of our model allows easy interpretation of the novel physical considerations. We demonstrate that the global dynamics of our biogeochemical model are significantly altered by clustering forces. Most notably, a new balance in the ecosystem exists in which reactions are dominated entirely by those within the dense clusters. This greatly enhances the impact of destructive pollutants through efficient mixing. There is evidence that this equilibrium will be robust moving to more complex and realistic models.

Journal article

Shevchenko I, Berloff P, 2023, On a probabilistic evolutionary approach to ocean modelling: From Lorenz-63 to idealized ocean models, Ocean Modelling, Vol: 186, ISSN: 1463-5003

In this study we develop an alternative way to model the ocean reflecting the chaotic nature of ocean flows and uncertainty of ocean models — instead of making use of classical deterministic or stochastic differential equations we offer a probabilistic evolutionary approach (PEA) that capitalizes on the use of probabilistic dynamics in phase space. The main feature of the data-driven version of PEA proposed in this work is that it does not require to know the physics behind the flow dynamics to model it. Within the PEA framework we develop two probabilistic evolutionary methods, which are based on probabilistic evolutionary models using quasi time-invariant structures in phase space. The methods have been tested on complete and incomplete reference data sets generated by the Lorenz 63 system and by an idealized two-layer quasi-geostrophic model. The results show that both methods reproduce large- and small-scale features of the reference flow by keeping the probabilistic dynamics within the phase space of the reference flow. The proposed approach offers appealing benefits and a great flexibility to ocean modellers working with mathematical models and measurements. The most remarkable one is that it provides an alternative to the mainstream ocean parameterizations, requires no modification of existing ocean models, and is easy to implement. Moreover, it does not depend on the nature of input data, and therefore could work with both numerically-computed flows and real measurements from different sources (drifters, weather stations, etc.).

Journal article

Davies J, Sutyrin GG, Crowe MN, Berloff PSet al., 2023, Deformation and destruction of north-eastward drifting dipoles, Physics of Fluids, Vol: 35, ISSN: 1070-6631

We systematically study the evolution of Larichev-Reznik dipoles in an equivalent-barotropic quasi-geostrophic beta-plane model in high-resolution numerical simulations. Our results shed light on the self-organization and rich dynamics of dipolar vortices, which are ubiquitous in geophysical flows. By varying both dipole strength and initial angle α0 of dipole tilt to the zonal direction, we discover new breakdown mechanisms of the dipole evolution. The dipoles are quickly destroyed by Rossby wave radiation, if initial tilt is too large or the dipole is too weak; otherwise, via damped oscillations, the dipoles tend to adjust themselves to different states drifting eastward. Two competing physical mechanisms that govern dipole transformations are found: (1) spontaneous dipole instability due to a growing critical linear mode and (2) meridional separation of dipole partners that accumulates over the adjustment period and prevents the above instability. Which mechanism prevails depends on the initial tilt and dipole strength, and the details of this are discussed.

Journal article

Kurashina R, Berloff P, 2023, Low-frequency variability enhancement of the midlatitude climate in an eddy-resolving coupled ocean–atmosphere model. Part I: anatomy, Climate Dynamics, Vol: 61, Pages: 1997-2023, ISSN: 0930-7575

This study investigated the coupling of the wind-driven ocean gyres with the atmospheric westerly jet using an idealised, eddy-resolving, coupled model. An empirical orthogonal function analysis of the low-pass filtered data showed that the ocean gyre variability is dominated by meridional shifts of the western boundary current extension (WBCE) and changes in the strength of the subtropical inertial recirculation zone. On the other hand, the atmospheric potential vorticity (PV) variability is dominated by the growth of standing Rossby wave patterns, while its pressure variability is dominated by a zonally-asymmetric meridional shift of the atmospheric jet. Damping sea surface temperature (SST) variability in the atmosphere was shown to weaken its PV variability and reduce the zonal asymmetry of the jet-shift mode. Singular value decompositions revealed a positive feedback between meridional shifts of the WBCE and the growth of standing Rossby wave disturbances in the atmospheric jet. The atmosphere’s response is controlled by shifts in the meridional eddy heat flux over the SST front which triggers the growth of baroclinic instabilities. This instability growth eventually leads to a large-scale, barotropic pressure response over the eastern ocean basin, or an aforementioned meridional shift of the atmospheric jet. Reduction in the atmospheric resolution inhibits the ability of atmospheric eddies to resolve length scales associated with meridional shifts of the SST front and WBCE. The lack of resolution consequently weakens the influence of ocean gyre variability on the atmospheric jet and reduces the strength of the positive feedback.

Journal article

Kurashina R, Berloff P, 2023, Low-frequency variability enhancement of the midlatitude climate in an eddy-resolving coupled ocean-atmosphere model-part II: ocean mechanisms, Climate Dynamics, Vol: 61, Pages: 2025-2044, ISSN: 0930-7575

This paper investigates the spatial inhomogeneity of the time-averaged, quasigeostrophic, double-gyre circulation response to fixed, realistic, large-scale modes of wind-stress forcing. While the companion paper of this study focused on understanding the anatomy of low-frequency, midlatitude climate variability in an idealised, eddy-resolving coupled model, this paper looked at understanding the nature of the wind-induced ocean gyre response using an ocean-only configuration of the same model. Our analysis revealed two, time-averaged responses to an east–west dipole, wind-stress curl anomaly in the ocean basin. Firstly, wind-stress anomalies in the western ocean basin led to changes in relative strength of the inertial recirculation zones and jet-axis tilt. This is consistent with an advection-dominated, nonlinear adjustment of the ocean gyres to anomalous forcing. Secondly, wind-stress curl anomalies in the eastern ocean basin was found to induce a largely independent response involving meridional shifts of the western boundary current extension (WBCE). The effects of time-averaged advection in this region are weak and the discovery of westward-propagating Rossby waves along the WBCE revealed the response is more akin to a baroclinic Rossby wave adjustment.

Journal article

Shevchenko I, Berloff P, 2023, A hyper-parameterization method for comprehensive ocean models: advection of the image point, Ocean Modelling, Vol: 184, ISSN: 1463-5003

Idealized and comprehensive ocean models at low resolutions cannot reproduce nominally-resolved flow structures similar to those presented in the high-resolution solution. Although there are various underlying physical reasons for this, from the dynamical system point of view all these reasons manifest themselves as a low-resolution trajectory avoiding the phase space occupied by the reference solution (the high-resolution solution projected onto the coarse grid). In order to solve this problem, a set of hyper-parameterization methods has recently been proposed and successfully tested on idealized ocean models. In this work, for the first time we apply one of hyper-parameterization methods (Advection of the image point) to a comprehensive, rather than idealized, general circulation model of the North Atlantic.The results show that the hyper-parameterization method significantly outperforms the coarse-grid ocean model by reproducing both the large- and small-scale features of the Gulf Stream flow. The proposed method is much faster than even a single run of the coarse-grid ocean model, requires no modification of the model, and is easy to implement. Moreover, the method can take not only the reference solution as input data but also real measurements from different sources (drifters, weather stations, etc.), or combination of both. All this offers a great flexibility to ocean modellers working with mathematical models and/or measurements.

Journal article

Meacham J, Berloff P, 2023, On clustering of floating tracers in random velocity fields, Journal of Advances in Modeling Earth Systems, Vol: 15, ISSN: 1942-2466

In this paper, we investigate the aggregation of a floating tracer into clusters. Motivated by observations of dense patches of buoyant material in the real ocean (e.g., microplastic pollutants, plankton, and sargassum), we develop an idealized model that can reproduce the clustering process. A stochastic, kinematic 2D velocity field is chosen to represent turbulent oceanic surface currents, with a weakly divergent component. Lagrangian particles are introduced and we track their concentrations. We differ from delta-correlated fields used in previous studies by including finite time correlations. Clustering in these fields can be compared to the traditional setting, through global measures and cluster detection algorithms. The enhanced velocity fields can be deformed using various interpolation methods. We can then investigate the sensitivity of clustering to the representation of temporal/spatial velocity structure to inform future studies of this phenomenon. We find coherency of time-correlated velocities leads to significantly faster rates of clustering, causing a larger number of longer lived/more populated clusters to form. Clustering is likely relevant to a host of biogeochemical processes of urgent interest, such as phytoplankton blooms and the ecological risk of microplastic pollutants. This work aims to establish an accurate basis for clustering simulations, to enable further exploration.

Journal article

Davies J, Sutyrin GG, Berloff P, 2023, On the spontaneous symmetry breaking of eastward propagating dipoles, Physics of Fluids, Vol: 35, ISSN: 1070-6631

The spontaneous symmetry breaking of weak eastward propagating dipoles is revealed from high-resolution numerical simulations in an equivalent-barotropic quasigeostrophic beta-plane model. The evolution of initialized Larichev–Reznik dipoles is found to depend on the initial intensity, characterized by the dipole drift speed divided by the maximum Rossby wave phase speed. We found that weak dipoles lose symmetry and eventually disintegrate due to growing perturbations. This instability is attributed to a critical, linear growing mode that is extracted from the potential vorticity anomaly. The growth rate of perturbations is found to decrease with the increasing dipole intensity.

Journal article

Lu Y, Kamenkovich I, Berloff P, 2022, Properties of the lateral mesoscale eddy-induced transport in a high-resolution ocean model: beyond the flux-gradient relation, Journal of Physical Oceanography, Vol: 52, Pages: 3273-3295, ISSN: 0022-3670

Lateral mesoscale eddy-induced tracer transport is traditionally represented in coarse-resolution models by the flux–gradient relation. In its most complete form, the relation assumes the eddy tracer flux as a product of the large-scale tracer concentration gradient and an eddy transport coefficient tensor. However, several recent studies reported that the tensor has significant spatiotemporal complexity and is not uniquely defined, that is, it is sensitive to the tracer distributions and to the presence of nondivergent (“rotational”) components of the eddy flux. These issues could lead to significant biases in the representation of the eddy-induced transport. Using a high-resolution tracer model of the Gulf Stream region, we examine the diffusive and advective properties of lateral eddy-induced transport of dynamically passive tracers, reevaluate the utility of the flux–gradient relation, and propose an alternative approach based on modeling the local eddy forcing by a combination of diffusion and generalized eddy-induced advection. Mesoscale eddies are defined by a scale-based spatial filtering, which leads to the importance of new eddy-induced terms, including eddy-mean covariances in the eddy fluxes. The results show that the biases in representing these terms are noticeably reduced by the new approach. A series of targeted simulations in the high-resolution model further demonstrates that the approach outperforms the flux–gradient model in reproducing the stirring and dispersing effect of eddies. Our study indicates potential to upgrade the traditional flux–gradient relation for representing the eddy-induced tracer transport.

Journal article

Shevchenko I, Berloff P, 2022, A method for preserving nominally-resolved flow patterns in low-resolution ocean simulations: Constrained dynamics, Ocean Modelling, Vol: 178, Pages: 1-6, ISSN: 1463-5003

Inability of low-resolution ocean models to simulate many important aspects of the large-scale general circulation is a common problem. In the view of physics, the main reason for this failure are the missed dynamical effects of the unresolved small scales of motion on the explicitly resolved large-scale circulation. Complimentary to this mainstream physics-based perspective, we propose to address this failure from the dynamical systems point of view, namely, as the persistent tendency of phase space trajectories representing the low-resolution solution to escape the right region of the corresponding phase space, which is occupied by the reference eddy-resolving solution. Based on this concept, we propose to use methods of constrained optimization to confine the low-resolution solution to remain within the correct phase space region, without attempting to amend the eddy physics by introducing a process-based parameterization. This approach is advocated as a novel framework for data-driven hyper-parameterizations of mesoscale oceanic eddies in non-eddy-resolving models. We tested the idea in the context of classical, baroclinic beta-plane turbulence model and showed that non-eddy-resolving solution can be substantially improved towards the reference eddy-resolving benchmark.

Journal article

Ryzhov EA, Berloff P, 2022, On transport tensor of dynamically unresolved oceanic mesoscale eddies, Journal of Fluid Mechanics, Vol: 939, ISSN: 0022-1120

Parameterizing mesoscale eddies in ocean circulation models remains an open problem due to the ambiguity with separating the eddies from large-scale flow, so that their interplay is consistent with the resolving skill of the employed non-eddy-resolving model. One way to address the issue is by using recently formulated dynamically filtered eddies. These eddies are obtained as the field errors of fitting some given reference ocean circulation into the employed coarse-grid ocean model. The main strengths are (i) no explicit spatio-temporal filter is needed for separating the large-scale and eddy flow components, (ii) the eddies are dynamically translated into the error-correcting forcing that perfectly augments the coarse-grid model towards reproducing the reference circulation. We uncovered physical properties of the eddies by interpreting involved nonlinear eddy/large-scale interactions via the classical flux-gradient relation. We described the eddies in terms of their full, space–time dependent transport tensor, which was made unique by constraining it to be the same for the potential vorticity, momentum and buoyancy fluxes. Both diffusive and advective parts of the transport tensor were found to be significant. The diffusive tensor component is characterised by polar eigenvalues and is further decomposed into isotropic and filamentation components. The latter component completely dominates, therefore, it should be taken into account by eddy parameterizations, which is not yet the case. We also showed that spatial inhomogeneities of the transport tensor components are important. Comparing these properties with those obtained for more common, locally filtered eddies revealed that they are distinctly different.

Journal article

Haigh M, Berloff P, 2022, On the stability of tracer simulations with opposite-signed diffusivities, Journal of Fluid Mechanics, Vol: 937, Pages: 1-11, ISSN: 0022-1120

Many recent studies have diagnosed opposite-signed diffusion eigenvalues to be a prevalent feature of the transfer tensor for diffusive tracer transport by oceanic mesoscale eddies. This diagnosed tensor, which we refer to as the diffusion tensor, therefore accounts for tracer filamentation effects. The preferential orientation of this filamentation is quantified by the principal axis of the diffusion tensor, namely the diffusion axis. Parameterisations of eddy diffusion commonly invoke a diffusion tensor, typically one with non-negative eigenvalues to avoid numerical issues. Motivated by the need to parameterise tracer filamentation, in this study we examine diffusion of a Gaussian tracer patch with imposed opposite-signed diffusion eigenvalues, and in particular we focus on the time scale for the onset of instability. For a fixed diffusion axis, numerical instability is an inevitable consequence of persistent up-gradient fluxes associated with the negative eigenvalue. For typical oceanic scales and diffusion magnitudes, this time scale is of the order of 100 days, but is shorter for larger negative eigenvalues or for finer grid resolutions. We show that imposing a time-dependent diffusion axis can lead to simulations with no onset of instability after 100 000 days of tracer evolution. Although motivated by oceanographic fluid dynamics, our results have much broader applications since diffusive processes are present in a wide range of fluid flows.

Journal article

Shevchenko I, Berloff P, 2022, A method for preserving nominally-resolved flow patterns in low-resolution ocean simulations: Dynamical system reconstruction, Ocean Modelling, Vol: 170, Pages: 101939-101939, ISSN: 1463-5003

Accurate representation of large-scale flow patterns in low-resolution ocean simulations is one of the most challenging problems in ocean modelling. The main difficulty is to correctly reproduce effects of unresolved small scales on the resolved large scales. For this purpose, most of current research is focused on development of parameterizations directly accounting for the small scales. In this work we propose an alternative to the mainstream ideas by showing how to reconstruct a dynamical system from the available reference solution data (our proxy for observations) and, then, how to use this system for modelling not only large-scale but also nominally-resolved flow patterns at low resolutions. This approach is advocated as a part of the novel framework for data-driven hyper-parameterization of mesoscale oceanic eddies in non-eddy-resolving models. The main characteristic of this framework is that it does not require to know the physics behind large–small scale interactions to reproduce both large and small scales in low-resolution ocean simulations. We tested it in the context of a three-layer, statistically equilibrated, steadily forced quasigeostrophic model for the beta-plane configuration and showed that non-eddy-resolving solution can be substantially improved towards the reference eddy-resolving benchmark. The proposed methodology robustly allows to retrieve a system of equations governing reduced dynamics of the observed data, while the additional adaptive nudging counteracts numerical instabilities by keeping solutions in the region of phase space occupied by the reference fields. Remarkably, its solutions simulate not only large-scale but also small-scale flow features, which can be nominally resolved by the low-resolution grid. In addition, the proposed method reproduces realistic vortex trajectories. One of the important and general conclusions that can be drawn from our results is that not only mesoscale eddy parameterization is possible in pri

Journal article

Shevchenko I, Berloff P, 2021, On a minimum set of equations for parameterisations in comprehensive ocean circulation models, Ocean Modelling, Vol: 168, Pages: 1-7, ISSN: 1463-5003

The complexity of comprehensive ocean models poses an important question for parameterisations: is there a minimum set of equations that should be parameterised, on the one hand, to reduce the development to a minimum, and, on the other hand, to ensure an accurate representation of large-scale flow patterns? This work seeks to answer the question to assist modern parameterisations be more selective in their targets. For this, we considered a model of the North Atlantic and studied contributions of different model equations to the accuracy of representation of the Gulf Stream at low resolution. Our results suggest that one should focus on parameterising the tracer equations for temperature and salinity, and may leave the other equations in the hydrodynamic part, as well as the atmospheric model unmodified. They also suggest that parameterisations representing only Kinetic Energy Backscatter cannot be fully efficient and the main focus should be shifted towards developing parameterisations of combined Potential/Kinetic Energy Backscatters.

Journal article

Haigh M, Berloff P, 2021, On co-existing diffusive and anti-diffusive tracer transport by oceanic mesoscale eddies, Ocean Modelling, Vol: 168, ISSN: 1463-5003

A common approach for parameterising eddy transport of passive tracers by mesoscale eddies in the ocean is by invoking a transport tensor. The symmetric part of this tensor, the diffusion tensor, quantifies diffusive eddy tracer transport. Recent studies have diagnosed opposite-signed eigenvalues (diffusivities) of the diffusion tensor from eddy-resolving simulations, while all current parameterisations implement only positive diffusivities. For opposite-signed eigenvalues the associated diffusive eddy tracer flux is not necessarily down-gradient and therefore may not mix the tracer by transferring variance to the small scales. In this study we explore such diffusive eddy fluxes by using an eddy-resolving simulation of passive tracers with a relaxation (source/sink) forcing. After confirming that the diffusion tensors for different tracer pairs have opposite-signed eigenvalues, we show that the corresponding diffusive eddy tracer flux drives a net down-gradient transfer of variance, as would be guaranteed when the diffusion tensor eigenvalues are both positive. Locally up-gradient fluxes are common, with their frequency strongly dependent on the relaxation profile. The effects of weakening/strengthening the relaxation on the frequency of down-gradient fluxes is different for each tracer. However, for all tracers considered the amplitude of the net down-gradient transfer weakens as the relaxation strengthens, a consequence of the homogeneous diffusion dissipating less eddy variance. Our results indicate that for oceanic tracers with sources/sinks the parameterised diffusive eddy tracer fluxes should not be globally down-gradient.

Journal article

Agarwal N, Ryzhov E, Kondrashov D, Berloff Pet al., 2021, Correlation-based flow decomposition and statistical analysis of the eddy forcing, Journal of Fluid Mechanics, Vol: 924, Pages: 1-30, ISSN: 0022-1120

We present a comprehensive study of the mesoscale eddy forcing in the ocean by proposing spatially local filtering of the high-resolution double-gyre ocean circulation solution into its large- and small-scale (eddy) components. The large-scale component is dominated by the mid-latitude gyres, the western boundary currents and their highly transient eastward jet extension; the eddy component is concentrated around the eastward jet and strongly interacts with it. The proposed decomposition method achieves flow filtering based on the local spatial correlations. This is different from the existing decomposition methods, e.g. classical Reynolds decomposition and moving-average (spatial) filtering with a constant filter size based on the first baroclinic Rossby deformation radius. Next, we characterize the dynamical impacts of the resulting eddy forcing on the large-scale flow in terms of their mutual time-lagged spatial correlations, formulated as product integral characteristics. Its temporal statistics uncover robust causality between the eddy forcing and the induced large-scale potential vorticity anomalies – referred to as the eddy backscatter. The results also prove the significance of the transient eddy forcing and the time lag dependence of the eddy backscatter. We argue that these properties are to be considered by eddy parametrization schemes. We further used the decomposed eddy fields to augment a coarse-resolution ocean model. The augmented solution statistically reproduces the missing eastward jet extension, enhances the eddy activities around it and recovers the essential large-scale low-frequency variability. This justifies a reduced-order statistical emulation of the eddies – an emerging methodology for including eddy effects in non-eddy-resolving ocean models.

Journal article

Haigh M, Sun L, McWilliams JC, Berloff Pet al., 2021, On eddy transport in the ocean. Part II: The advection tensor, Ocean Modelling, Vol: 165, Pages: 1-17, ISSN: 1463-5003

This study considers the isopycnal eddy transport of mass and passive tracers in eddy-resolving doublegyre quasigeostrophic oceanic circulation. Here we focus on advective transport, whereas a companion paperfocuses on eddy-induced diffusive tracer transport. To work towards parameterising eddy tracer transport wequantify the eddy tracer flux using a transport tensor with eddies defined using a spatial filter, which leadsto results distinct from those obtained via a temporal Reynolds eddy decomposition. The advection tensoris the antisymmetric part of the transport tensor, and is so named since the associated tracer transport canbe expressed as advection of the large-scale tracer field by a rotational eddy-induced velocity (EIV) 𝒖𝑐∗ withstreamfunction 𝐴. The EIV 𝒖𝑐∗is fastest (∼ 1 m s−1) where eddy activity is strongest, e.g., in the upper layer,near the eastward jet and western boundary current. Our results suggest that a stochastic closure for the eddytransport would be most suitable since 𝐴 exhibits a probabilistic distribution when conditioned on, for example,the large-scale relative vorticity. Consistent with closures in ocean circulation models, we quantify eddy mass(isopycnal layer thickness) fluxes as eddy-induced advection by the thickness EIV 𝒖ℎ∗. The divergent part of𝒖ℎ∗– the only part relevant for mass transport in the quasigeostrophic limit – tends to be oriented down thethickness gradient suggesting it quantifies some baroclinic eddy effects similar to those parameterised by theGent & McWilliams (GM90) EIV. Although 𝒖ℎ∗has some qualitative similarities to 𝒖𝑐∗, our results suggest thateddy-induced tracer advection is driven by more than just the thickness-determined EIV and, in turn, morethan just the GM90 EIV.

Journal article

Agarwal N, Kondrashov D, Dueben P, Ryzhov E, Berloff Pet al., 2021, A comparison of data-driven approaches to build low-dimensional ocean models, Journal of Advances in Modeling Earth Systems, Vol: 13, ISSN: 1942-2466

We present a comprehensive inter-comparison of linear regression (LR), stochastic, and deep-learning approaches for reduced-order statistical emulation of ocean circulation. The reference data set is provided by an idealized, eddy-resolving, double-gyre ocean circulation model. Our goal is to conduct a systematic and comprehensive assessment and comparison of skill, cost, and complexity of statistical models from the three methodological classes. The model based on LR is considered as a baseline. Additionally, we investigate its additive white noise augmentation and a multi-level stochastic approach, deep-learning methods, hybrid frameworks (LR plus deep-learning), and simple stochastic extensions of deep-learning and hybrid methods. The assessment metrics considered are: root mean squared error, anomaly cross-correlation, climatology, variance, frequency map, forecast horizon, and computational cost. We found that the multi-level linear stochastic approach performs the best for both short- and long-timescale forecasts. The deep-learning hybrid models augmented by additive state-dependent white noise came second, while their deterministic counterparts failed to reproduce the characteristic frequencies in climate-range forecasts. Pure deep learning implementations performed worse than LR and its simple white noise augmentation. Skills of LR and its white noise extension were similar on short timescales, but the latter performed better on long timescales, while LR-only outputs decay to zero for long simulations. Overall, our analysis promotes multi-level LR stochastic models with memory effects, and hybrid models with linear dynamical core augmented by additive stochastic terms learned via deep learning, as a more practical, accurate, and cost-effective option for ocean emulation than pure deep-learning solutions.

Journal article

Sun L, Haigh M, Shevchenko I, Berloff P, Kamenkovich Iet al., 2021, On non-uniqueness of the mesoscale eddy diffusivity, Journal of Fluid Mechanics, Vol: 920, Pages: 1-27, ISSN: 0022-1120

Oceanic mesoscale currents (‘eddies’) can have significant effects on the distributions of passive tracers. The associated inhomogeneous and anisotropic eddy fluxes are traditionally parametrised using a transport tensor (K-tensor), which contains both diffusive and advective components. In this study, we analyse the eddy transport tensor in a quasigeostrophic double-gyre flow. First, the flow and passive tracer fields are decomposed into large- and small-scale (eddy) components by spatial filtering, and the resulting eddy forcing includes an eddy tracer flux representing advection by eddies and non-advective terms. Second, we use the flux-gradient relation between the eddy fluxes and the large-scale tracer gradient to estimate the associated K-tensors in their entire structural, spatial and temporal complexity, without making any additional assumptions or simplifications. The divergent components of the eddy tracer fluxes are extracted via the Helmholtz decomposition, which yields a divergent tensor. The remaining rotational flux does not affect the tracer evolution, but dominates the total tracer flux, affecting both its magnitude and spatial structure. However, in terms of estimating the eddy forcing, the transport tensor prevails over its divergent counterpart because of the significant numerical errors induced by the Helmholtz decomposition. Our analyses demonstrate that, in general, the K-tensor for the eddy forcing is not unique, that is, it is tracer-dependent. Our study raises serious questions on how to interpret and use various estimates of K-tensors obtained from either observations or eddy-resolving model solutions.

Journal article

Berloff P, Ryzhov E, Shevchenko I, 2021, On dynamically unresolved oceanic mesoscale motions, Journal of Fluid Mechanics, Vol: 920, Pages: 1-24, ISSN: 0022-1120

The problem of defining oceanic mesoscale eddies remains generally unresolved because there is no unique local spatio-temporal filter that can be used for extracting the eddies, and it is unclear what part of the eddy field cannot be actually resolved and needs to be parameterized in a coarse-grid model. We propose using of the coarse-grid model itself for reconstructing dynamically unresolved eddies, which are actually field errors on the top of the dynamically resolved, large-scale reference flow. The novelties and strengths of the approach are that (i) no spatio-temporal filtering is ever needed, (ii) field errors are dynamically translated into the error-correcting forcing and (iii) the latter exactly augments the coarse-grid model solution towards the reference flow. After implementation of the proposed approach, we study statistical properties of the field errors, show their robustness and reveal their significant differences from the locally filtered eddies. We argue that dynamical effects of unresolved eddies can be ultimately parameterized by emulating field errors and closing them on the dynamically resolved flow. So far, our results are limited to the quasigeostrophic approximation, but this serves as a proof of concept and starting point for the follow-up extension into the primitive equations, which are used routinely in the comprehensive oceanic general circulation models.

Journal article

Haigh M, Sun L, McWilliams JC, Berloff Pet al., 2021, On eddy transport in the ocean. Part I: The diffusion tensor, Ocean Modelling, Vol: 164, Pages: 1-15, ISSN: 1463-5003

This study provides an interpretation of isopycnal eddy transport for mass and passive tracers in double-gyreeddy-resolving oceanic circulation. This paper focuses on a transport/diffusion tensor representation of theeddy tracer flux, and a companion paper will focus on advective eddy-induced tracer and mass transports. Weuse a spatial filter to separate the large and small scales, which leads to results distinct from those obtained viaa temporal Reynolds eddy decomposition. To work towards a parameterisation, we relate the eddy tracer fluxto the large-scale tracer gradient via the transport tensor 𝑲. The symmetric part of 𝑲 is the diffusion tensor,𝑺, which parameterises diffusive fluxes and whose mixing properties are determined by the signs of its eigenvalues. The eigenvalues of 𝑺 are robustly of opposite sign (polar) and thus quantify filamentation of the tracervia both up- and down-gradient fluxes. Given the prevalence of polar eigenvalues – which are also obtained forReynolds eddy fluxes – representing their associated effects should be a target of future eddy tracer transportclosures. Given the inherent inhomogeneity and anisotropy of the eddy-induced transport, we argue that a fulltransport tensor is better suited to this task than scalar coefficients or diagonal tensors. The diffusion axis,which represents the direction of preferential mixing, tends to align with the large-scale velocity vector andcontours of large-scale relative vorticity and layer thickness. Strong shears can inhibit this alignment. We showthat the large-scale velocity gradient matrix may be suitable for parameterising the transport tensor, in particular at depth. Furthermore, since entries of 𝑲 and 𝑺 exhibit probabilistic distributions when conditioned on certain large-scale flow features, we suggest that a stochastic closure for the eddy transport would be most suitable.

Journal article

Kurashina R, Berloff P, Shevchenko I, 2021, Western boundary layer nonlinear control of the oceanic gyres, Journal of Fluid Mechanics, Vol: 918, Pages: 1-26, ISSN: 0022-1120

This study examines the influence of flow nonlinearity in western boundary layers upon the turbulent wind-driven ocean gyres. Our analysis involves comparisons between large-scale circulation properties of the linear and nonlinear states, as well as a Lagrangian particle analysis of relevant flow features. We find that the so-called counter-rotating gyre anomalies, which are nonlinear circulation features embedded in the gyres, are consistent in shape with the linear, weakened, wind-curl response created by the geometric wind effect. However, the linear response is far too weak without considering nonlinear effects. Within the western boundary layer lobe of these features, the nonlinear boundary layer has a pivotal impact upon the global circulation. Effects of potential vorticity advection inhibit viscous relative vorticity fluxes through the western boundary. This creates a significant potential vorticity imbalance between the gyres. Consequently, this generates an accumulation of enstrophy downstream in the inertial recirculation zones, which in turn supports the eastward jet. However, within the ocean basin, the growing imbalance is eventually rectified by inter-gyre potential vorticity exchanges owing to nonlinear fluxes. The Lagrangian particle analysis reveals the inter-gyre exchange mechanism, where particles seeded within the western boundary layer migrate between the gyres and weaken the eastward jet extension.

Journal article

Shevchenko I, Berloff P, 2021, A method for preserving large-scale flow patterns in low-resolution ocean simulations, Ocean Modelling, Vol: 161, Pages: 1-6, ISSN: 1463-5003

It is typical for low-resolution simulations of the ocean to miss not only small- but also large-scale patterns of the flow dynamics compared with their high-resolution analogues. It is usually attributed to the inability of coarse-grid models to properly reproduce the effects of the unresolved small-scale dynamics on the resolved large scales. In part, the reason for that is that coarse-grid models fail to at least keep the coarse-grid solution within the region of phase space occupied by the true solution (the high-resolution solution projected onto the coarse grid). In this paper we offer a solution to this problem by computing the image point in the phase space restricted to the region of the true flow dynamics. The proposed method shows encouraging results for both low- and high-dimensional phase spaces, it takes a near-zero effort to implement into existing numerical codes and has ample room for further improvements.

Journal article

Kamenkovich I, Berloff P, Haigh M, Sun L, Lu Yet al., 2021, Complexity of mesoscale eddy diffusivity in the ocean, Geophysical Research Letters, Vol: 48, Pages: 1-12, ISSN: 0094-8276

Stirring of water by mesoscale currents (“eddies”) leads to large‐scale transport of many important oceanic properties (“tracers”). These eddy‐induced transports can be related to the large‐scale tracer gradients, using the concept of turbulent diffusion. The concept is widely used to describe these transports in the real ocean and to represent them in climate models. This study focuses on the inherent complexity of the corresponding coefficient tensor (“K‐tensor”) and its components, defined here in all its spatio‐temporal complexity. Results demonstrate that this comprehensive K‐tensor is space‐, time‐, direction‐ and tracer‐dependent. Using numerical simulations with both idealized and comprehensive models of the Atlantic circulation, we show that these properties lead to upgradient eddy fluxes and the potential importance of all tensor components. The uncovered complexity of the eddy transports calls for reconsideration of how they are estimated in practice, included in the general circulation models and theoretically interpreted.

Journal article

Davies J, Khatri H, Berloff P, 2021, Linear stability analysis for flows over sinusoidal bottom topography, Journal of Fluid Mechanics, Vol: 911, Pages: 1-25, ISSN: 0022-1120

This is an ocean motivated study which investigates the impacts of sinusoidal bottom topography on baroclinic instability of zonal vertically sheared flows in the two-layer quasigeostrophic model. The corresponding linear stability problem is solved by assuming Fourier-mode solutions in both the zonal and meridional directions. In the presence of variable topographic features, the Fourier modes become coupled due to phase shifts in the wavevectors. The spectral discretisation method used in this work retains the primary relationship between different Fourier modes; thus, the linear stability eigenproblem can be solved for any periodic topography. Moreover, this method does not need any additional assumptions, such as considering small-amplitude or large-scale bottom irregularities, as in some previous studies. In this work, the eigenproblem is solved for a range of topographic amplitudes and wavenumbers, and their effects on the growth rates and shapes of the most unstable eigenmodes are discussed. In general, both the zonal and meridional variations in topography tend to suppress the baroclinic instability. However, it is found that only meridionally varying topography affects the magnitudes of the fastest growth rates. In this instance, unstable modes appear to form two clusters well separated in the zonal wavenumber axis and growth rate maxima occur at two distinct zonal wavenumbers. Dependencies of the characteristics of these clusters on the values of topography amplitude and ridge width are reviewed. Finally, doubly periodic numerical simulations are used to verify the results from the linear stability analysis.

Journal article

Ryzhov EA, Kondrashov D, Agarwal N, McWilliams JC, Berloff Pet al., 2020, On data-driven induction of the low-frequency variability in a coarse-resolution ocean model, Ocean Modelling, Vol: 153, Pages: 1-11, ISSN: 1463-5003

This study makes progress towards a data-driven parameterization for mesoscale oceanic eddies. To demonstrate the concept and reveal accompanying caveats, we aimed at replacing a computationally expensive, standard high-resolution ocean model with its inexpensive low-resolution analogue augmented by the parameterization. We considered eddy-resolving and non-eddy-resolving double-gyre ocean circulation models characterized by drastically different solutions due to the nonlinear mesoscale eddy effects. The key step of the proposed approach is to extract from the high-resolution reference solution its eddy field varying in space and time, and then to use this information to improve the low-resolution analogue model.By interactively coupling both the continuously supplied history of the eddy field and the explicitly modeled low-resolution large-scale flow, we obtained the additional eddy forcing term which modified the low-resolution model and significantly augmented its solutions. This eddy forcing term represents the action of the eddy field, its coupling with the large-scale flow and is a key dynamical constraint imposed on the augmentation procedure.Although the augmentation drastically improved the low-resolution circulation patterns, it did not recover the robust, intrinsic, large-scale low-frequency variability (LFV), which is an important feature of the high-resolution solution. This is by itself an important (negative) result that has significant implication for any data-driven eddy parameterization, especially, given the fact that we used the most complete information about the space–time history of the eddy fields. Note, when we supplied the reference (true) eddy forcing, rather than just the eddy field, the LFV was recovered. This suggests that the LFV is crucially dependent on the details of the space–time eddy forcing/large-scale flow correlations, which are not fully respected by the proposed augmentation procedure.In order to overcome the def

Journal article

Stepanov D, Ryzhov EA, Berloff P, Koshel Ket al., 2020, Floating tracer clustering in divergent random flows modulated by an unsteady mesoscale ocean field, Geophysical and Astrophysical Fluid Dynamics, Vol: 114, Pages: 690-714, ISSN: 0309-1929

Clustering of tracers floating on the ocean surface and evolving due to combined velocity fields consisting of a deterministic mesoscale component and a kinematic random component is analysed. The random component represents the influence of submesoscale motions. A theory of exponential clustering in random velocity fields is applied to characterise the obtained clustering scenarios in both steady and unsteady time-dependent mesoscale flows, as simulated by a comprehensive realistic, eddy-resolving, general circulation model for the Japan/East Sea. The mesoscale flow field abounds in transient eddy-like patterns modulating and branching the main currents, and the underlying time-mean flow component features closed recirculation zones that can entrap the tracer. The submesoscale flow component is modelled kinematically, as a divergent random velocity field with a prescribed correlation radius and variance. The combined flow induces tracer clustering, that is, the exponential growth of tracer density in patches with vanishing areas. The statistical topography methodology, which provides integral characteristics to quantify the emerging clusters, uncovers drastic dependence of the clustering rates on whether the mesoscale flow component is taken to be steady or time-dependent. The former situation favours robust exponential clustering, similar to the theoretically understood case of purely divergent and zero-mean random velocity. The latter situation, on the contrary, hinders exponential clustering due to significant advection of the tracer out of the nearly enclosed eddies, at the rate faster than the clustering rate.

Journal article

This data is extracted from the Web of Science and reproduced under a licence from Thomson Reuters. You may not copy or re-distribute this data in whole or in part without the written consent of the Science business of Thomson Reuters.

Request URL: http://wlsprd.imperial.ac.uk:80/respub/WEB-INF/jsp/search-html.jsp Request URI: /respub/WEB-INF/jsp/search-html.jsp Query String: respub-action=search.html&id=00524879&limit=30&person=true