Invariant parameterization of geostrophic eddies in the ocean

Alexander Bihlo; Elsa Dos Santos Cardoso-Bihlo; Roman O. Popovych

REVIEW 3 major objections 4 minor 1 cited by

Invariant parameterization of geostrophic eddies in the ocean

T0 review · 3 major / 4 minor · reviewed 2026-08-14 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read This paper constructs the first higher-order invariant closure for ocean eddy parameterization and shows numerically that it outperforms the standard non-invariant closure in decaying-turbulence tests.

desk verdict Solid symmetry analysis undercuts its own 'first higher-order invariant closure' claim via a permissive generalized definition, and the numerics are confounded; worth refereeing but needs reframing. read the letter →

arxiv 1908.06345 v1 pith:YNVOJCZP submitted 2019-08-17 physics.ao-ph math-phmath.APmath.MP

Alexander Bihlo , Elsa Dos Santos Cardoso-Bihlo , Roman O. Popovych This is my paper · ORCID

classification physics.ao-phmath-phmath.APmath.MP MSC 35B0676M6086A05

keywords invariantparameterizationLiesymmetriesbarotropicvorticityequationbeta-planeeddyclosureturbulentkineticenergyequivalencetransformationsFofonoffvortices

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper extends symmetry-preserving (invariant) parameterization of unresolved ocean eddies from first-order closures to one-and-a-half order closures, where an explicit equation for turbulent kinetic energy $k$ is part of the model. The central construction is system (23): the Reynolds-averaged barotropic vorticity equation closed by an eddy vorticity flux with coefficient $\kappa = \alpha L_{\mathrm{eddy}}(2k)^{1/2}$ and an invariant hyperdiffusion term $\tilde A k^{5/4}\nabla^4\eta$, together with a kinetic energy equation closed by eddy diffusivity $\nu = 2\tilde\alpha L_{\mathrm{eddy}}(2k)^{1/2}$. The authors show that the family of such systems is invariant, in a generalized sense, under the essential subgroup of the symmetry group of the barotropic vorticity equation, with $L_{\mathrm{eddy}}$ transforming as a length scale. They verify numerically that these invariant schemes give better average results than standard non-invariant closure models, including more robust emergence of Fofonoff vortices and no unstable integrations in 72 runs.

What carries the argument

The carrying object is the one-and-a-half order closure model (23), built from the Reynolds-averaged barotropic vorticity equation plus a prognostic equation for turbulent kinetic energy $k$. Its invariant nature rests on three pieces: the prolongation of the symmetry group of the vorticity equation to $k$ (which transforms as $\tilde k = e^{-4\varepsilon_3}k$), the differential invariants $I_1 = \psi_{xx}/\sqrt{|\psi_x|}$ and $I_2 = k/\psi_x^2$ with the invariant differentiation operators (15), and the generalized interpretation of scale transformations as equivalence transformations that act on the eddy mixing length $L_{\mathrm{eddy}}$ via $\tilde L_{\mathrm{eddy}} = e^{-\varepsilon_3}L_{\mathrm{eddy}}$. The nonlinear invariant hyperdiffusion term $\tilde A k^{5/4}\nabla^4\eta$ replaces standard linear hyperdiffusion, which broke scale invariance.

What would settle it

Compute the action of a generalized Galilean boost $(t,x,y,\psi,k) \to (t, x+f(t), y, \psi-f_t(t)y, k)$ on system (23): the $k$-equation's conversion term $-\kappa\nabla\psi\cdot\nabla\eta$ will not be form-preserving, showing the closure is not invariant under the full symmetry group. Alternatively, repeat the 72-run decaying-turbulence experiment on a doubly periodic domain where Galilean invariance is physical; if the invariant model then loses its advantage or becomes unstable, the essential-subgroup assumption is the culprit.

Watch

Extended reading notes

Core claim

The paper's central claim is that a one-and-a-half order closure for geostrophic eddies can be made invariant: the closed system (23), with $\kappa$ and $\nu$ as in (18) and (22), admits the prolonged action of the essential subgroup $\bar G_{\mathrm{ess}}$ (scale transformations, translations, and streamfunction gauging) of the maximal Lie symmetry group of the barotropic vorticity equation, with the eddy mixing length $L_{\mathrm{eddy}}$ scaling as $e^{-\varepsilon_3}$ so that the scale transformation acts as an equivalence transformation on the class of closed systems. This gives the first example of an invariantly closed system using a closure of order higher than one. The paper also introduces generalized invariant parameterization, in which symmetries of the original model are preserved as equivalence transformations of the class of closed models rather than as strict symmetries of a single model. Numerically, the invariant model produces Fofonoff vortex solutions more robustly than the non-invariant model: higher mean anti-correlation, less scatter, and no unstable integrations.

Load-bearing premise

The construction assumes that fixed Dirichlet boundaries make generalized Galilean boosts 'not natural,' so only the essential subgroup needs preserving; if Galilean invariance were required of the closed model, system (23) would not qualify as invariant.

Editorial extensions

If this is right

Symmetry preservation in parameterization is not limited to first-order closures; one-and-a-half order closure models that carry turbulent kinetic energy can be made invariant in the generalized sense.
The equivalence-transformation interpretation gives a practical rule: closure constants with a physical dimension, like the eddy mixing length, may be rescaled by symmetries, while dimensionless constants must not be; this distinguishes $L_{\mathrm{eddy}}$ from $\tilde A$.
Invariant hyperdiffusion of the form $\tilde A k^{5/4}\nabla^4\eta$ is self-refining: as resolved eddy kinetic energy decreases with finer resolution, the hyperdiffusion strength decreases automatically, unlike grid-spacing-tuned $A$.
Numerically, the invariant model reproduces the linear $\eta$\textendash$\psi$ relation of Fofonoff states over a wider parameter range and delays diffusive homogenization of the gyres.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

Beyond the paper: the same equivalence-transformation trick could be applied to other closure constants with physical dimensions, such as eddy turnover time in $\kappa = 2\gamma T_{\mathrm{eddy}}k$, turning each into a scale-covariant parameter and enlarging the set of invariant closure families.
Beyond the paper: the requirement to preserve only the essential subgroup is a modeling choice tied to fixed Dirichlet boundaries; on periodic beta-plane domains, Galilean invariance becomes natural, and a testable prediction is that an invariant closure must then include generalized Galilean boosts, which would change the allowed form of the energy-conversion term.
Beyond the paper: the nonlinear hyperdiffusion term suggests a concrete diagnostic, namely comparing spectral energy fluxes of the invariant and non-invariant models to see whether the $k^{5/4}$ multiplier alters the inertial-range cascade in a way that explains the delayed homogenization.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

The symmetry-analysis core is genuinely good, but you should not trust the headline claims as stated. The 'first higher-order invariant closure' is undercut by the paper's own generalized definition, and the numerical comparison is confounded.

What's actually new and worth keeping: The prolongation of the barotropic vorticity equation's symmetry group to turbulent kinetic energy is clean and useful—scale transformations act on k as e^{-4ε}, Galilean boosts act only on the mean streamfunction. The first computation of the maximal Lie invariance algebra of the beta-plane Euler equations (8) and the potential-symmetry observation for ∂y is a nice side result. The explicit construction of system (23), with κ = αL_eddy√(2k), ν = 2α~L_eddy√(2k), and k^{5/4} hyperdiffusion, is internally consistent, and the invariance proof checks out. The move to let symmetries act as equivalence transformations on closure parameters is a reasonable extension of [32], even if it is more permissive than the authors acknowledge.

Now the soft spots. The stress-test note is essentially right: treat A and ν as arbitrary elements of the class (as the experiments do), and the 'standard non-invariant' model (17) with r=0 also maps to itself under scale, with A and ν rescaling appropriately (the specific exponents in the note are off by sign, but the structural point stands). So the claim to be the first higher-order invariant closure reduces to a choice about which constants are allowed to transform. That is a convention, not a structural difference. The paper's distinction—L_eddy has physical justification to rescale, A~ does not—is informal and not part of the formal definition in Section 2.

The numerics are suggestive but weaker than claimed. The invariant model changes two things relative to (17): the hyperdiffusion becomes nonlinear k^{5/4} and the eddy energy diffusivity becomes k-dependent. You can't attribute the difference to invariance alone. The reported means C_inv=469±51 vs C_std=445±66 overlap substantially, and the instability count (4 vs 0) is not analyzed. No error bars, no significance test.

Also, dropping Galilean invariance because fixed boundaries break it is defensible, but it means the 'essential subgroup' is a domain-dependent choice, and the generalized invariance claim is scoped to that choice.

Who should read this: people interested in symmetry-based closure construction will find the invariant-theoretic machinery useful. The paper deserves a serious referee, but the authors should be asked to (1) clarify why their model is the first higher-order invariant closure under a definition that also admits (17), or tighten the definition, and (2) run a controlled experiment that isolates invariance from the change in hyperdiffusion nonlinearity.

My recommendation: send to peer review with these requests. The mathematical core is sound; the claims need reframing.

Referee Report

3 major / 4 minor

Summary. The paper extends the invariant-parameterization framework to higher-order (one-and-a-half order) closures and introduces a generalized notion in which symmetries of the original model are preserved as equivalence transformations of a class of closed systems, with closure parameters treated as arbitrary elements. The specific application is to geostrophic eddies in a barotropic ocean: the authors compute the maximal Lie symmetry algebra of the barotropic vorticity equation on the beta-plane, compute for the first time the Lie symmetries of the underlying two-dimensional Euler equations, prolong the symmetries to the turbulent kinetic energy, and construct a closed system (Eq. (23)) that they claim is invariant under the essential subgroup of the symmetry group in this generalized sense. They then compare this invariant model with the standard non-invariant model (17) in numerical experiments of freely decaying two-dimensional turbulence, measuring the anti-correlation function as an indicator of Fofonoff-vortex emergence, and report that the invariant model gives, on average, better and more robust results.

Significance. If substantiated, the paper would make a useful contribution: it would provide the first higher-order symmetry-preserving closure, a new potential symmetry interpretation for the Euler equations on the beta-plane, and a generalized framework for treating closure parameters as equivalence-transformation variables. The symmetry computations and the invariance verification for Eq. (23) appear internally consistent, and the explicit prolongation of the symmetry action to the turbulent kinetic energy is a clean and reusable result. However, the significance is limited by two issues: the claimed novelty of the first higher-order invariant closure is undermined by the permissiveness of the generalized definition, and the numerical evidence does not isolate the effect of symmetry preservation from other changes between the two compared models.

major comments (3)

[Section 2 and Section 5 (after Eq. (22))] The generalized definition of invariant parameterization in Section 2 explicitly allows equivalence transformations to act on the arbitrary elements of the class, i.e., on the closure parameters. Under this definition, the 'standard non-invariant' model (17) with r=0 is already generalized scale-invariant: with κ given by (18) and with the scale transformation acting on the parameters as L_eddy' = e^{-ε}L_eddy, A' = e^{-5ε}A, and ν' = e^{-3ε}ν (or the inverse convention), the class of systems (17) is mapped to itself. The paper itself notes in Section 5 that scale invariance of (17) can be restored by letting A transform. Therefore, the statement that 'the above class gives the first example for an invariantly closed system employing a closure of order higher than one' (Section 5, after Eq. (23)) is a convention about which parameters are allowed to transform, not a structural property of the closure. The formal definition does not distinguish (23) from (17); the distinction relies on the informal physical-justification requirement about dimensionless constants. The authors should either refine the definition to include a formal criterion (e.g., which closure parameters are required to be absolute invariants) or substantially revise the novelty claim.
[Section 6, numerical experiments] The numerical comparison does not support the general claim that 'the invariant parameterization schemes give, on average, better results than the standard non-invariant closure models' (Abstract and Section 6.2). The invariant model (23) differs from the standard model (17) in at least three respects simultaneously: the hyperdiffusion term is changed from A∇⁴η to ~Ak^{5/4}∇⁴η, the eddy energy diffusivity is changed from a constant ν to ν = 2~αL_eddy(2k)^{1/2}, and the symmetry properties are altered. The 4 unstable non-invariant runs are mentioned, but it is not stated whether they are excluded from the reported means and standard deviations; the difference in means of the anti-correlation function (469 vs 445) is not assessed with any statistical test or confidence interval. To attribute the observed improvement to symmetry preservation, the authors should include controlled experiments, for example comparing (23) with (17) in which A and ν are allowed to scale as equivalence transformations, or comparing (23) with (20), and report significance measures.
[Section 5, choice of the essential subgroup] The selection of G_ess as the symmetry group to preserve is a load-bearing modeling assumption. The paper excludes generalized Galilean boosts because fixed Dirichlet boundaries are not natural for moving frames, but the same boundary argument also breaks scale invariance, which is nevertheless retained through the equivalence-transformation interpretation. The asymmetry between treating scale transformations as equivalence transformations and treating Galilean boosts as non-natural is not derived from a formal principle. Since the central claim of invariance of model (23) depends on this choice, the authors should state a more principled criterion for selecting the essential subgroup, or at least discuss how the results would change if Galilean boosts were also treated as equivalence transformations on the boundary conditions.

minor comments (4)

[Section 6.1] There is a typo in the boundary conditions description: 'no normal follow' should be 'no normal flow'.
[Section 6.1] The setting ~A = A/k₀^{5/4} matches the initial hyperdiffusion strength at t = 0 only; since k evolves in time, the assertion that the diffusion strength is 'similar over the entire integration' should be quantified or stated more cautiously.
[Eq. (24)] The anti-correlation function C is reported with values around 469 in non-dimensional units, but the definition (24) is dimensionally dependent on the choice of scaling; a brief statement of the normalization or units would improve interpretability.
[Section 4] The claim that the Lie symmetries of the system (8) 'have not been done in the literature before' would be easier to verify if the authors indicated the search terms or briefly explained why the existing symmetry analyses of the beta-plane equations do not cover this system.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the invariant closure (23) is obtained by imposing explicit scaling constraints on prescribed eddy-transfer forms, and its numerical skill is tested against an external Fofonoff benchmark rather than used to set parameters.

full rationale

The paper's central closure coefficients, κ = α L_eddy (2k)^{1/2} and ν = 2α̃ L_eddy (2k)^{1/2}, are taken from the prior physical closure literature (Marshall & Adcroft and Eden & Greatbatch), not fitted to the Fofonoff outcome measured in Section 6. The invariant hyperdiffusion multiplier k^{5/4} in (21) is derived by requiring each term in the averaged vorticity equation to scale as e^{-2ε3} under the scale subgroup, and the statement that (21) is invariant under all of Ḡ is checked by direct computation; it is not obtained by assuming the target result. The differential invariants I1 and the invariant differentiation operators (15) are cited from the authors' earlier work [4], but this is a parameter-free invariant-theory computation with stated assumptions (ψx ≠ 0) and does not include the closure model (23) as an input; the same generating invariants can be verified independently from the moving frame (16). Thus the self-citation is load-bearing only in the sense of supplying technical infrastructure, not as a substitute for the paper's own derivation. The generalized-invariance notion is a definition, and the choice to let L_eddy transform while holding α, α̃, and Ã fixed is an explicitly defended modeling convention in Section 5 and in the paragraph following Remark 4. A critic could argue this convention is permissive; for example, allowing A and ν to transform could make the 'standard' model (17) scale-invariant in the same generalized sense. But that is a scope or correctness objection, not a reduction of the paper's claims to its inputs. Finally, the numerical comparison is not a fitted prediction: setting Ã = A / k0^{5/4} merely equalizes initial hyperdiffusion strength between the invariant and non-invariant models, and the 72-run ensemble varies prescribed parameter ranges rather than tuning constants to maximize the anti-correlation function C. No fitted parameter is renamed as a prediction, no uniqueness theorem is imported from the authors' prior work to force the chosen ansatz, and no output quantity is equal by construction to an input quantity. Hence no significant circularity is present.

Assumptions & free parameters 5 free parameters · 7 assumptions · 0 invented entities

The paper introduces no new physical entities. It introduces a new modeling concept (generalized invariant parameterization, where symmetries act as equivalence transformations on closure parameters) and a new closure term (k^(5/4) hyperdiffusion), but these are mathematical constructs, not postulated entities with independent falsifiable handles. The closure parameters are free constants inherited from prior work and scanned over ranges in the numerical tests.

free parameters (5)

alpha = 0.01 (fixed in experiments)
Dimensionless eddy mixing coefficient in κ = α L_eddy (2k)^(1/2) (Eq. 18), taken from [26].
L_eddy = range [2π/100, 2π/20] in experiments
Eddy mixing length; central arbitrary element of the class, scaled as e^(-ε3) under the equivalence transformation.
nu = range [10^-4, 10^-3]
Eddy energy diffusivity in Eq. (17); used to set ˜α = ν/L_eddy.
A_tilde = A in [10^-7, 10^-5], with Ã = A/k0^(5/4)
Hyperdiffusion coefficient in the invariant model, multiplying k^(5/4)∇⁴η (Eq. 21).
k0 = range [0.10, 0.25]
Initial turbulent kinetic energy, varied across the 72 runs to assess robustness; an initial condition rather than a closure parameter.

assumptions (7)

domain assumption Reynolds averaging rules, including the Reynolds property ¯ab = ¯a¯b and ab = ¯a¯b + a′b′
Used to derive the averaged vorticity equation (10) and TKE equation (11); standard but an idealization.
domain assumption The turbulent kinetic energy equation (11) is a valid evolution equation for k after averaging and neglecting certain terms (e.g., r=0 for freely decaying turbulence)
Eq. (11) and the neglect of the r k term in Section 5.
ad hoc to paper The closure ansatz κ = α L_eddy (2k)^(1/2) and ν = 2 ˜α L_eddy (2k)^(1/2) (Eqs. 18, 22) is physically appropriate
Adopted from [26] and modified for scaling; not derived from symmetry.
ad hoc to paper The choice of ¯G_ess as the symmetry group to preserve, dropping generalized Galilean boosts
Justified by fixed boundary conditions; this is the premise that the invariance claim rests on.
standard math Standard results from Lie group analysis, including moving frame invariantization and differential invariant theory
Used throughout Sections 4-5; unproved background.
ad hoc to paper The scaling of L_eddy as e^(-ε3) under the equivalence transformation
Needed to keep κ scaling correctly in Eq. 18; introduced in Section 5 as part of generalized invariant parameterization.
domain assumption The standard numerical framework (Arakawa Jacobian, RK2, boundary conditions) is a correct discretization
Section 6.1; standard methods, but no code provided and no convergence tests reported.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Invariant parameterization of geostrophic eddies in the ocean." pith.science (2026). https://pith.science/paper/YNVOJCZP

@misc{pith2026190806345,
  author       = {Pith},
  title        = {Pith review of: Invariant parameterization of geostrophic eddies in the ocean},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/YNVOJCZP}},
  note         = {Machine review of arXiv:1908.06345}
}

read the original abstract

The framework of invariant parameterization is extended to higher-order closure schemes. We also define, for the first time, generalized invariant parameterization schemes, where symmetries of the corresponding original model are preserved as equivalence transformations of related classes of closed system of differential equations. As a particular problem, we consider invariant parameterization schemes for geostrophic eddies in a barotropic ocean. Here the initial model is the barotropic vorticity equation, which is equivalent to the system of incompressible inviscid two-dimensional Euler equations on a midlatitude beta-plane. The maximal Lie invariance algebra of this model is infinite-dimensional, and we intend to preserve it in the course of invariant parameterization, at least partially. The parameterizations proposed for the eddy vorticity flux and the energy flux are of order one and a half since we explicitly consider the equation for the turbulent kinetic energy in the closure models. These parameterizations are therefore the first examples of invariant higher-order closure schemes. Numerical experiments are carried out to assess the performance of these invariant schemes in studies of freely decaying two-dimensional turbulence, and it is verified that the invariant parameterization schemes give, on average, better results than the standard non-invariant closure models do.

Figures

Figures reproduced from arXiv: 1908.06345 by the authors.

**Figure 2.** Upper panel: Zonally-averaged ψ–η scatter plot (left) at t = 500 and time series of the anticorrelation function (right) that are averaged over 72 runs varying the parameterization constants over the intervals Leddy ∈ [2π/20, 2π/100], ν ∈ [10−3 , 10−4 ] and A ∈ [10−5 , 10−7 ], with α = 0.01, ˜α = ν/Leddy, and varying the initial turbulent kinetic energy over k0 = [0.1, 0.25]. Lower panel: Final stream function at t… view at source ↗

**Figure 3.** Upper panel: Zonally-averaged ψ–η scatter plot at t = 500 (left) and time series of the anti-correlation function (right). Lower panel: Final stream function at t = 500 using standard hyperdiffusion (left) and invariant hyperdiffusion (right). The parameterization constants are Leddy = 2π/100, A = 10−7 , ν = 10−3 , starting with uniform initial turbulent kinetic energy set to k0 = 0.15. Overall, the results obtained… view at source ↗

Figures from the paper (1 more)

**Figure 4.** Figure 4: Upper panel: Zonally-averaged ψ–η scatter plot at t = 500 (left) and time series of the anti-correlation function (right). Lower panel: Final stream function at t = 500 using standard hyperdiffusion (left) and invariant hyperdiffusion (right). The parameterization cons…

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

Differential invariants for a class of diffusion equations
math-ph 2019-09 conditional novelty 6.0 of 10

The differential invariant algebra for the equivalence pseudogroup of ut = uxx + f(u, ux) is generated by one invariant I11 and two invariant differentiation operators.

Reference graph

Works this paper leans on

39 extracted references · 35 canonical work pages · cited by 1 Pith paper

[32]

and Bihlo A., Symmetry preserving parameterization schemes, J

Popovych R.O. and Bihlo A., Symmetry preserving parameterization schemes, J. Math. Phys. 53 (2012), 073102 (36 pages), arXiv:1010.3010

arXiv 2012
[1]

Part I, J

Arakawa A., Computational design for long-term numerical integration of the equations of ﬂuid motion: Two-dimensional incompressible ﬂow. Part I, J. Comp. Phys. 1 (1966), 119–143

work page 1966
[2]

Conservative parameterization schemes

Bihlo A. and Bluman G., Conservative parameterization schemes, J. Math. Phys. 54 (2013), 083101 (24 pages), arXiv:1209.4279

work page Pith review arXiv 2013
[3]

Complete group classification of a class of nonlinear wave equations

Bihlo A., Dos Santos Cardoso-Bihlo E.M. and Popovych R.O., Complete group classiﬁcation of a class of nonlinear wave equations, J. Math. Phys. 53 (2012), 123515 (32 pages), arXiv:1106.4801

work page Pith review arXiv 2012
[4]

and Popovych R.O., Invariant parameterization and turbulence modeling on the beta-plane, Phys

Bihlo A., Dos Santos Cardoso-Bihlo E.M. and Popovych R.O., Invariant parameterization and turbulence modeling on the beta-plane, Phys. D 269 (2014), 48–62, arXiv:1112.1917

arXiv 2014
[5]

Lie symmetries and exact solutions of the barotropic vorticity equation

Bihlo A. and Popovych R.O., Lie symmetries and exact solutions of the barotropic vorticity equation,J. Math. Phys. 50 (2009), 123102 (12 pages), arXiv:0902.4099

work page Pith review arXiv 2009
[6]

Lie symmetry analysis and exact solutions of the quasi-geostrophic two-layer problem

Bihlo A. and Popovych R.O., Lie symmetry analysis and exact solutions of the quasi-geostrophic two-layer problem, J. Math. Phys. 52 (2011), 033103 (24 pages), arXiv:1010.1542

work page Pith review arXiv 2011
[7]

and Popovych R.O., Invariant discretization schemes for the shallow-water equations, SIAM J

Bihlo A. and Popovych R.O., Invariant discretization schemes for the shallow-water equations, SIAM J. Sci. Comput. 34 (2012), B810–B839, arXiv:1201.0498

arXiv 2012

Show all 39 references

[8]

and Popovych R.O., Group classiﬁcation of linear evolution equations, J

Bihlo A. and Popovych R.O., Group classiﬁcation of linear evolution equations, J. Math. Anal. Appl. 448 (2017), 982–1005

2017
[9]

and Valiquette F., Symmetry-preserving numerical schemes, in Symmetries and integrability of diﬀerence equations, Springer, pp

Bihlo A. and Valiquette F., Symmetry-preserving numerical schemes, in Symmetries and integrability of diﬀerence equations, Springer, pp. 261–324, 2017

2017
[10]

and Kumei S., Symmetries and diﬀerential equations , Springer, New York, 1989

Bluman G. and Kumei S., Symmetries and diﬀerential equations , Springer, New York, 1989

1989
[11]

and Venaille A., Statistical mechanics of two-dimensional and geophysical ﬂows, Phys

Bouchet F. and Venaille A., Statistical mechanics of two-dimensional and geophysical ﬂows, Phys. Rep. 515 (2012), 227–295

2012
[12]

and Haidvogel D.B., Two-dimensional turbulence above topography, J

Bretherton F.P. and Haidvogel D.B., Two-dimensional turbulence above topography, J. Fluid Mech. 78 (1976), 129–154. 20

1976
[13]

and Pohjanpelto J., Algorithms for diﬀerential invariants of symmetry groups of diﬀer- ential equation, Found

Cheh J., Olver P.J. and Pohjanpelto J., Algorithms for diﬀerential invariants of symmetry groups of diﬀer- ential equation, Found. Comput. Math. 8 (2008), 501–532

2008
[14]

Cummins P.F., Inertial gyres in decaying and forced geostrophic turbulence, J. Mar. Res. 50 (1992), 545–566

1992
[15]

and Zanna L., The statistical nature of turbulent barotropic ocean jets, Ocean Model

David T.W., Marshall D.P. and Zanna L., The statistical nature of turbulent barotropic ocean jets, Ocean Model. 113 (2017), 34–49

2017
[16]

and Greatbatch R.J., Evolution of mean-ﬂow Fofonoﬀ gyres in barotropic quasigeostrophic turbulence, J

Dukowicz J.K. and Greatbatch R.J., Evolution of mean-ﬂow Fofonoﬀ gyres in barotropic quasigeostrophic turbulence, J. Phys. Oceanogr. 29 (1999), 1832–1852

1999
[17]

and Greatbatch R.J., Towards a mesoscale eddy closure, Ocean Model

Eden C. and Greatbatch R.J., Towards a mesoscale eddy closure, Ocean Model. 20 (2008), 223–239

2008
[18]

and Olver P.J., Moving coframes: I

Fels M. and Olver P.J., Moving coframes: I. A practical algorithm, Acta Appl. Math. 51 (1998), 161–213

1998
[19]

and Olver P.J., Moving coframes

Fels M. and Olver P.J., Moving coframes. II. Regularization and theoretical foundations, Acta Appl. Math. 55 (1999), 127–208

1999
[20]

Fofonoﬀ N.P., Steady ﬂow in a frictionless homogeneous ocean, J. Mar. Res. 13 (1954), 254–264

1954
[21]

and Menemenlis D., Can large eddy simulation techniques improve mesoscale rich ocean models?, in Ocean modeling in an eddying regime , Wiley Online Library, pp

Fox-Kemper B. and Menemenlis D., Can large eddy simulation techniques improve mesoscale rich ocean models?, in Ocean modeling in an eddying regime , Wiley Online Library, pp. 319–337, 2008

2008
[22]

and Nadiga B.T., Four-gyre circulation in a barotropic model with double-gyre wind forcing, J

Greatbatch R.J. and Nadiga B.T., Four-gyre circulation in a barotropic model with double-gyre wind forcing, J. Phys. Oceanogr. 30 (2000), 1461–1471

2000
[23]

Katkov V.L., A class of exact solutions of the equation for the forecast of the geopotential, Izv. Akad. Nauk. SSSR Ser. Fiz. Atmosfer. i Oceana 1 (1965), 630–631

1965
[24]

Katkov V.L., Exact solutions of the geopotential forecast equation, Izv. Akad. Nauk. SSSR Ser. Fiz. Atmosfer. i Oceana 2 (1966), 1193

1966
[25]

D 98 (1996), 481–491

Leith C.E., Stochastic models of chaotic systems, Phys. D 98 (1996), 481–491

1996
[26]

and Adcroft A.J., Parameterization of ocean eddies: Potential vorticity mixing, energetics and Arnold’s ﬁrst stability theorem, Ocean Modell

Marshall D.P. and Adcroft A.J., Parameterization of ocean eddies: Potential vorticity mixing, energetics and Arnold’s ﬁrst stability theorem, Ocean Modell. 32 (2010), 188–204

2010
[27]

Oberlack M., Invariant modeling in large-eddy simulation of turbulence , in: Annual research briefs, Stanford University, 1997

1997
[28]

and Pohjanpelto J., Moving frames for Lie pseudo-groups, Canadian J

Olver P.J. and Pohjanpelto J., Moving frames for Lie pseudo-groups, Canadian J. Math. 60 (2008), 1336– 1386

2008
[29]

and Popovych R.O., Group analysis of general Burgers–Korteweg–de Vries equa- tions, J

Opanasenko S., Bihlo A. and Popovych R.O., Group analysis of general Burgers–Korteweg–de Vries equa- tions, J. Math. Phys. 58 (2017), 081511

2017
[30]

Press, New York, 1982

Ovsiannikov L.V., Group analysis of diﬀerential equations , Acad. Press, New York, 1982

1982
[31]

Popovych R.O., Classiﬁcation of admissible transformations of diﬀerential equations, Collection of Works of Institute of Mathematics, Kyiv 3 (2006), 239–254

2006
[33]

and Bihlo A., Inverse problem on conservation laws, arXiv:1705.03547, 2017

Popovych R.O. and Bihlo A., Inverse problem on conservation laws, arXiv:1705.03547, 2017

2017 arXiv
[34]

and Eshraghi H., Admissible transformations and normalized classes of non- linear Schr¨ odinger equations,Acta Appl

Popovych R.O., Kunzinger M. and Eshraghi H., Admissible transformations and normalized classes of non- linear Schr¨ odinger equations,Acta Appl. Math. 109 (2010), 315–359, arXiv:math-ph/0611061

2010 arXiv
[35]

and Talley L.D., Generalizations of Arakawa’s Jacobian, J

Salmon R. and Talley L.D., Generalizations of Arakawa’s Jacobian, J. Comput. Phys. 83 (1989), 247–259

1989
[36]

The basic experiment, Mon

Smagorinsky J., General circulation experiments with the primitive equations: I. The basic experiment, Mon. Wea. Rev. 91 (1963), 99–164

1963
[37]

Fluid Mech

Speziale C.G., Galilean invariance of subgrid-scale stress models in the large-eddy simulation of turbulence, J. Fluid Mech. 156 (1985), 55–62

1985
[38]

13 of Atmospheric Sciences Library, Kluwer Academic Publishers, Dortrecht, 1988

Stull R.B., An introduction to boundary layer meteorology , vol. 13 of Atmospheric Sciences Library, Kluwer Academic Publishers, Dortrecht, 1988

1988
[39]

and Vallis G.K., Emergence of Fofonoﬀ states in inviscid and viscous ocean circulation models, J

Wang J. and Vallis G.K., Emergence of Fofonoﬀ states in inviscid and viscous ocean circulation models, J. Mar. Res. 52 (1994), 83–127. 21

1994

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.

[1] [32]

and Bihlo A., Symmetry preserving parameterization schemes, J

Popovych R.O. and Bihlo A., Symmetry preserving parameterization schemes, J. Math. Phys. 53 (2012), 073102 (36 pages), arXiv:1010.3010

arXiv 2012

[2] [1]

Part I, J

Arakawa A., Computational design for long-term numerical integration of the equations of ﬂuid motion: Two-dimensional incompressible ﬂow. Part I, J. Comp. Phys. 1 (1966), 119–143

work page 1966

[3] [2]

Conservative parameterization schemes

Bihlo A. and Bluman G., Conservative parameterization schemes, J. Math. Phys. 54 (2013), 083101 (24 pages), arXiv:1209.4279

work page Pith review arXiv 2013

[4] [3]

Complete group classification of a class of nonlinear wave equations

Bihlo A., Dos Santos Cardoso-Bihlo E.M. and Popovych R.O., Complete group classiﬁcation of a class of nonlinear wave equations, J. Math. Phys. 53 (2012), 123515 (32 pages), arXiv:1106.4801

work page Pith review arXiv 2012

[5] [4]

and Popovych R.O., Invariant parameterization and turbulence modeling on the beta-plane, Phys

Bihlo A., Dos Santos Cardoso-Bihlo E.M. and Popovych R.O., Invariant parameterization and turbulence modeling on the beta-plane, Phys. D 269 (2014), 48–62, arXiv:1112.1917

arXiv 2014

[6] [5]

Lie symmetries and exact solutions of the barotropic vorticity equation

Bihlo A. and Popovych R.O., Lie symmetries and exact solutions of the barotropic vorticity equation,J. Math. Phys. 50 (2009), 123102 (12 pages), arXiv:0902.4099

work page Pith review arXiv 2009

[7] [6]

Lie symmetry analysis and exact solutions of the quasi-geostrophic two-layer problem

Bihlo A. and Popovych R.O., Lie symmetry analysis and exact solutions of the quasi-geostrophic two-layer problem, J. Math. Phys. 52 (2011), 033103 (24 pages), arXiv:1010.1542

work page Pith review arXiv 2011

[8] [7]

and Popovych R.O., Invariant discretization schemes for the shallow-water equations, SIAM J

Bihlo A. and Popovych R.O., Invariant discretization schemes for the shallow-water equations, SIAM J. Sci. Comput. 34 (2012), B810–B839, arXiv:1201.0498

arXiv 2012