{"id":"1854abfc-3aba-4c30-95a8-877a6857093c","arxiv_id":"1908.06345","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"The authors build one-and-a-half order invariant parameterization schemes for the beta-plane barotropic vorticity equation, preserving scale symmetries as equivalence transformations, and report moderately better Fofonoff vortex formation than a standard non-invariant closure.","lead":"The paper constructs eddy closure schemes for a simple ocean model that preserve the scaling symmetries of the underlying equations, and tests them in simulations of decaying two-dimensional turbulence. It is a proof of concept that symmetry preservation can be extended from simple closures to one-and-a-half order schemes that carry turbulent kinetic energy.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Generalized invariance is too permissive: the standard non-invariant model (17) also qualifies as scale-invariant in the generalized sense if A and ν are treated as transformable constants, undercutting the claimed 'first example' novelty.","rationale":"The reader's verdict flags the Galilean-invariance choice and the unverified equivalence-transformation interpretation. My concern pushes further: the issue is not just whether Galilean boosts are natural, but whether the generalized notion of invariance itself can distinguish any closure model from another. Section 2 defines generalized invariance via the equivalence group of the class of closed systems, putting no formal constraint on which constants may be transformed. The paper's informal 'physical justification' criterion, introduced only in Section 5 when explaining why A cannot be rescaled, is not part of the formal definition. As a result, the standard model (17), which the paper labels 'non-invariant,' can be made invariant under the same scale group by allowing A and ν to transform. Direct substitution confirms this. Therefore, the central claim that (23) is the first higher-order invariant closure is not robust: the comparison model qualifies under the same definition. The numerical experiments then do not compare invariant vs non-invariant closures; they compare two different physical parameterizations (nonlinear vs linear hyperdiffusion, k-dependent vs constant eddy diffusivity), so the invariance property is not isolated. Consequently, the abstract's claim that 'the invariant parameterization schemes give, on average, better results' is not supported as stated. The construction itself is mathematically sound, and the generalized concept is new, but the paper needs to either formalize which parameters may be transformed or weaken the novelty and numerical claims. This is why I keep the reader's CONDITIONAL verdict.","tokens_in":19391,"tokens_out":24388,"duration_ms":224505,"concrete_test":"Perform the analytic substitution: apply the scale transformation (t,x,y,ψ,k)→(e^ε t,e^{-ε}x,e^{-ε}y,e^{-3ε}ψ,e^{-4ε}k) to system (17) with r=0 and κ=αL_eddy(2k)^{1/2}, treating A, ν, and L_eddy as arbitrary constants. Verify that the transformed system has the same form with A' = e^{5ε}A, ν' = e^{3ε}ν, and L_eddy' = e^{-ε}L_eddy. If this holds, the non-invariant model is also invariant in the generalized sense, and both the novelty claim and the 'non-invariant' label used in the numerical comparison are called into question.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's central novelty is the first higher-order invariant closure, built on the generalized notion of invariant parameterization from Section 2. In that definition, a class of closed systems is invariant if its equivalence group contains a prolongation of selected symmetries of the original system to the subgrid-scale fields and to the arbitrary elements of the class. The authors apply this to system (23), letting the scale transformation act on L_eddy via L~_eddy = e^{-ε3}L_eddy while keeping α, ~A, ~α fixed. However, the same construction applies verbatim to the comparison model (17) with r=0 and κ as in (18), which the paper calls 'standard non-invariant.' If the constants A and ν are treated as arbitrary elements — as they are in the Section 6 experiments, where A ∈ [10^{-7},10^{-5}] and ν ∈ [10^{-4},10^{-3}] — then the scale transformation (t,x,y,ψ,k)→(e^ε t,e^{-ε}x,e^{-ε}y,e^{-3ε}ψ,e^{-4ε}k) maps the class to itself with A' = e^{5ε}A, ν' = e^{3ε}ν, L_eddy' = e^{-ε}L_eddy. Direct substitution of the transformed derivatives confirms closure. Thus the standard model qualifies as generalized scale-invariant by the letter of the definition. The only thing distinguishing it from (23) is the informal 'physical justification' criterion in Section 5, which is not part of the formal definition. This makes the 'first example' claim a convention about which constants are allowed to transform, rather than a structural property of the closure. The numerical comparison then reduces to a comparison between a model with nonlinear hyperdiffusion and k-dependent ν versus one with linear hyperdiffusion and constant ν, not between invariant and non-invariant closures.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","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.","tokens_in":19821,"tokens_out":10012,"duration_ms":101745,"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":[{"comment":"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":"Section 2 and Section 5 (after Eq. (22))"},{"comment":"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":"Section 6, numerical experiments"},{"comment":"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.","section":"Section 5, choice of the essential subgroup"}],"minor_comments":[{"comment":"There is a typo in the boundary conditions description: 'no normal follow' should be 'no normal flow'.","section":"Section 6.1"},{"comment":"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.","section":"Section 6.1"},{"comment":"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":"Eq. (24)"},{"comment":"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.","section":"Section 4"}],"recommendation":"major_revision","confidential_remarks":"The paper contains solid symmetry computations, but the framing of the novelty needs careful reconsideration. The generalized invariance notion is broad enough that the standard comparison model also qualifies, and the numerical section does not control for the confounding changes in nonlinearity and diffusivity. I would encourage the editor to ask for a revised version that either formalizes the notion of which parameters may transform or drops the 'first example' claim, and that adds controlled numerical experiments with proper statistical reporting."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","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.\n\nWhat'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.\n\nNow 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.\n\nThe 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.\n\nAlso, 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.\n\nWho 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.\n\nMy recommendation: send to peer review with these requests. The mathematical core is sound; the claims need reframing.","headline":"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.","tokens_in":20302,"tokens_out":9614,"would_cite":false,"duration_ms":88711,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["35B06","76M60","86A05"],"pacs":[],"model":"deepseek-v4-flash","headline":"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.","keywords":["invariant parameterization","Lie symmetries","barotropic vorticity equation","beta-plane","eddy closure","turbulent kinetic energy","equivalence transformations","Fofonoff vortices"],"falsifier":"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.","tokens_in":19175,"feed_emoji":"🌊","tokens_out":6388,"duration_ms":55421,"temperature":0.7,"pith_summary":"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.","feed_headline":"Symmetry-preserving eddy closure beats standard ocean model","feed_subtitle":"First higher-order invariant closure; Fofonoff vortices emerge more reliably and no run destabilizes in 72 tests.","key_machinery":"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.","core_discovery":"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.","pith_inferences":["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."],"forward_implications":["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."],"supporting_citations":[{"why":"Establishes the invariant-parameterization framework as a group classification problem, on which the present higher-order extension builds.","marker":"[32]"},{"why":"Provides the differential invariants and moving-frame construction for the maximal symmetry group of the beta-plane vorticity equation that the one-and-a-half order closure extends.","marker":"[4]"},{"why":"Supplies the one-and-a-half order closure model for eddy vorticity and energy fluxes that the paper makes invariant.","marker":"[26]"},{"why":"Defines the Fofonoff-state test and numerical setup (anti-correlation function, no-normal-flow boundary conditions) used for comparison.","marker":"[39]"},{"why":"Defines one-and-a-half order closure schemes as those carrying equations for second-order correlations such as turbulent kinetic energy.","marker":"[38]"},{"why":"Explains the tendency of Fofonoff vortices to homogenize into gyres under strong diffusion, the tendency the invariant model delays.","marker":"[16]"},{"why":"Provides the Fofonoff analytical steady solution whose emergence is the target of the numerical experiments.","marker":"[20]"}],"fun_headline_variants":["Invariant higher-order closure beats standard eddy models","First invariant closure of order 1.5 for geostrophic eddies","Symmetry-faithful eddy closure: better results, no blow-ups","Eddy closure preserving symmetries wins in numerical tests","Higher-order invariant scheme improves ocean eddy simulation"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"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.","fun_headline_variants_meta":{"raw":{"variants":["Invariant higher-order closure beats standard eddy models","First invariant closure of order 1.5 for geostrophic eddies","Symmetry-faithful eddy closure: better results, no blow-ups","Eddy closure preserving symmetries wins in numerical tests","Higher-order invariant scheme improves ocean eddy simulation"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000602,"raw_usage":{"total_tokens":2829,"prompt_tokens":982,"completion_tokens":1847,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":598,"completion_tokens_details":{"reasoning_tokens":1762}},"tokens_in":598,"tokens_out":1847,"duration_ms":14551,"temperature":1.0,"reasoning_tokens":1762,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T12:48:11.134640+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"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.","supporting_citations":[{"cited_title":"and Adcroft A.J., Parameterization of ocean eddies: Potential vorticity mixing, energetics and Arnold’s first stability theorem, Ocean Modell","cited_arxiv_id":null,"evidence_quote":"Supplies the one-and-a-half order closure model for eddy vorticity and energy fluxes that the paper makes invariant."},{"cited_title":"and Vallis G.K., Emergence of Fofonoff states in inviscid and viscous ocean circulation models, J","cited_arxiv_id":null,"evidence_quote":"Defines the Fofonoff-state test and numerical setup (anti-correlation function, no-normal-flow boundary conditions) used for comparison."},{"cited_title":"13 of Atmospheric Sciences Library, Kluwer Academic Publishers, Dortrecht, 1988","cited_arxiv_id":null,"evidence_quote":"Defines one-and-a-half order closure schemes as those carrying equations for second-order correlations such as turbulent kinetic energy."},{"cited_title":"and Greatbatch R.J., Evolution of mean-flow Fofonoff gyres in barotropic quasigeostrophic turbulence, J","cited_arxiv_id":null,"evidence_quote":"Explains the tendency of Fofonoff vortices to homogenize into gyres under strong diffusion, the tendency the invariant model delays."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the Fofonoff analytical steady solution whose emergence is the target of the numerical experiments."}],"review_version":1}