Modelling the climate and weather of a 2D Lagrangian-averaged Euler-Boussinesq equation with transport noise

Aythami Bethencourt de Leon; Darryl Holm; Diego Alonso-Oran; So Takao

REVIEW 2 major objections 5 minor 14 references

Modelling the climate and weather of a 2D Lagrangian-averaged Euler-Boussinesq equation with transport noise

T0 review · 2 major / 5 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 A stochastic Lagrangian-averaged version of the 2D Euler–Boussinesq equations has unique global strong solutions, and the covariance of its fluctuations evolves in a closed system of deterministic PDEs.

desk verdict First full LA SALT well-posedness and moment-closure analysis for 2D Euler-Boussinesq, but the headline theorem as stated needs an extra H^5 assumption that the proof requires and the statement omits. read the letter →

arxiv 1909.00388 v1 pith:ZOFEQYTG submitted 2019-09-01 math-ph math.MP

Diego Alonso-Oran , Aythami Bethencourt de Leon , Darryl Holm , So Takao This is my paper · ORCID

classification math-phmath.MP MSC 35Q3535R6060H1576B03

keywords Lagrangian-averagedSALTtransportnoiseEuler–Boussinesqequationsglobalwell-posednessclosedmomentstochasticfluiddynamicsKelvincirculationtheoremclimateandweathermodelling

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 constructs a stochastic version of the two-dimensional Euler–Boussinesq equations, the LA-SALT model, in which the drift velocity in the stochastic transport is replaced by its statistical expectation. The central claim is that this model has unique global strong solutions for initial velocity $u_0\in H^2(\mathbb{T}^2,\mathbb{R}^2)$ and initial temperature $\theta_0\in H^3(\mathbb{T}^2,\mathbb{R})$, provided the noise fields are smooth and uniformly elliptic. It also claims that the mean fields obey deterministic equations, that the fluctuations obey linear stochastic transport equations slaved to those means, and that the covariance of the velocity and temperature fields closes into a finite system of PDEs. If correct, the model gives a well-posed mathematical setting in which 'climate' (the expectation) and 'weather' (the fluctuations and their statistics) evolve separately, with climate change driven by deterministic mean dynamics together with fluctuation correlations.

What carries the argument

The load-bearing object is the LA-SALT stochastic transport velocity $dX_t = \mathbb{E}[u^L_t](X_t)\,dt + \sum_k \xi_k(X_t)\circ dW^k_t$: keeping the noise in the Lagrangian label but replacing the drift by its expectation preserves the Lie–Poisson, semidirect-product structure while making the mean-field equations deterministic and the fluctuation equations linear. The analytic machinery is two-stage: energy estimates for the closed vorticity–temperature expectation system give $\mathbb{E}[u]\in C([0,\infty),H^5)$, then the stochastic flow $\varphi_{s,t}$ of $dX_t$, together with Itô’s first formula and the Kunita–Itô–Wentzell formula for $k$-forms, yields an explicit representation of the solution as a push-forward of the initial data. Moment closure is obtained from Itô’s product rule and Cartan’s formula, which lets the exterior derivative commute with Lie transport and makes the $d\theta$ covariance close the velocity–temperature covariance system.

What would settle it

Solve the closed covariance system (3.32), (3.34), (3.36) on $\mathbb{T}^2$ and compare it with the covariance estimated from a large ensemble of solutions of the full SPDE (2.12); disagreement would show the claimed moment closure is false. Alternatively, take deterministic initial data $u_0\in H^2\setminus H^5$, $\theta_0\in H^3$ and check whether the characteristic construction still yields a global $H^2$ solution; failure would show Theorem 4.4 needs its extra regularity assumption stated explicitly.

Watch

Extended reading notes

Core claim

The discovery is that replacing the drift velocity in SALT by its expectation—the LA-SALT modification—turns the stochastic Boussinesq dynamics into a three-level hierarchy. The expectations $\mathbb{E}[u]$ and $\mathbb{E}[\theta]$ satisfy the closed deterministic system (2.13), which after taking the curl becomes the parabolic vorticity–temperature system (2.15). The fluctuations $u'=u-\mathbb{E}[u]$ and $\theta'=\theta-\mathbb{E}[\theta]$ then satisfy linear stochastic transport equations whose coefficients are determined by the already-solved means. The covariance tensors $U^{(2)}=\mathbb{E}[(u')^2]$, $\Theta^{(2)}=\mathbb{E}[(\theta')^2]$, and $\mathbb{E}[u'\otimes d\theta' + d\theta'\otimes u']$ form the closed system (3.32), (3.34), (3.36). On the analytic side, Theorem 4.4 proves unique global strong solutions by first solving the expectation equations and then constructing the stochastic solution along the flow of the characteristic SDE.

Load-bearing premise

The load-bearing premise is that the expected initial velocity $\mathbb{E}[u_0]$ is smooth enough, specifically in $H^5$, even though Theorem 4.4 states only $u_0\in H^2$; if $\mathbb{E}[u_0]$ lacks that smoothness, the stochastic-flow construction in Section 5.2 does not go through as written.

Editorial extensions

If this is right

The expectation equations (2.13) are closed and deterministic, so once they are solved they supply all coefficients of the linear fluctuation equations; climate and weather dynamics can be computed sequentially rather than as one coupled SPDE.
The covariance system (3.32), (3.34), (3.36) is closed, meaning variance growth of the fluctuations is governed by deterministic PDEs driven by gradients of the mean fields.
The $p$-th central moments of any advected scalar field close in the iterated system (3.37), giving equations for non-Gaussianity of temperature fluctuations in this model.
Theorem 4.4 rules out finite-time blow-up for these SPDE solutions under the stated smoothness and ellipticity conditions on the noise fields.

Reading between the lines

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

The proof's extra regularity requirement $\mathbb{E}[u_0]\in H^5$, flagged only in Remark 4.6, means the theorem as stated is slightly stronger than what the characteristic proof supports; readers should check whether weak-solution methods remove the gap.
Because the expectation system becomes parabolic only through the uniform ellipticity condition (4.8), the global regularity result depends essentially on the noise; as the noise fields degenerate toward the deterministic Boussinesq system, whose global regularity is open, the noise is doing real analytic work.
The closed covariance system invites a concrete numerical test: solve (3.32)–(3.36) and compare with ensemble statistics of the full SPDE, which would validate or disprove the moment closure on a discretized torus.
If the same closure persists for other semidirect-product fluid models, it could provide systematic variance-evolution equations for climate models; the paper itself notes that closure does not hold for general tensor advected fields because tensor products do not commute.

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

Dear [Colleague],

Here is my read on Alonso-Orán et al. The genuinely new content is that this is the first concrete LA SALT fluid model where they carry out the whole package: global well-posedness of the stochastic PDE, closed equations for the covariance and higher central moments for advected scalars, and a characteristic-based existence proof. The moment-closure part is the cleanest piece. Proposition 3.6 and the scalar p-th moment formula in Appendix B are correct and genuinely useful additions to the LA SALT toolkit.

But the main theorem overstates what is proven. Theorem 4.4 says H^2 × H^3 initial data gives a global strong solution. In the proof, the characteristics require E[u] ∈ C^{3,α}, which is obtained via Theorem 4.5 from the assumption E[u0] ∈ H^5. That assumption is not in the theorem statement. The authors admit this in Remark 4.6. For deterministic initial data this means you actually need u0 ∈ H^5, and the solution drops to H^2 immediately. So the headline statement is not false exactly—it is just unproven as the theorem is written. It is a fixable mismatch: either state the theorem with E[u0] ∈ H^5 or supply a separate existence argument for merely H^2 data. The mention of weak solutions via [dLT19] doesn't rescue the strong-solution theorem.

The second soft spot is the high-order energy estimates in Section 5.1. They give the L2 and H1 estimates in detail, then say H^4 for vorticity and H^3 for temperature are 'similar' and omit them. Those estimates are needed for Theorem 4.5's H^5 conclusion, which is the foundation of the whole proof. A referee should ask for those calculations before trusting the paper.

I don't think the gaps are fatal. The framework is coherent, the derivations are careful where they are shown, and the covariance closure is a real result. The paper is written for people working on stochastic fluid dynamics and LA SALT; the climate framing in the abstract should be ignored.

My recommendation: send it to peer review, but with a request for major revision that fixes the theorem statement and fills in the missing estimates. The core is worth referee time, but it is not ready in the current form.

Best, [Your name]

Referee Report

2 major / 5 minor

Summary. The paper develops the Lagrangian-averaged SALT (LA SALT) framework for the two-dimensional Euler–Boussinesq equations with transport noise. It derives the stochastic system (2.12), shows that the expectation equations are deterministic and parabolic, and proves that the fluctuation equations are linear stochastic transport equations slaved to the mean. In the 2D EB case, it obtains a closed system of deterministic equations for the second-order covariances (Example 3.7), and it claims global well-posedness for the stochastic system (Theorem 4.4). The proof strategy is to first solve the expectation equations at high regularity (Theorem 4.5), then to construct pathwise solutions to the linear stochastic system by a stochastic characteristic flow (Section 5.2).

Significance. If the regularity hypotheses are corrected, this is a valuable contribution: it gives a concrete stochastic fluid model with unique global strong solutions and closed deterministic moment equations, directly connecting Lorenz's climate/weather distinction to a tractable SPDE. The derivations in Section 3 are transparent and largely self-contained, the noise fields are inputs rather than fitted parameters, and the characteristic solution formula (5.21) together with the covariance closure are concrete and non-circular contributions. The main obstacle is that the headline well-posedness theorem currently overstates what is proved: the proof requires an extra regularity assumption that is not part of the theorem statement, and a key higher-order estimate needed by Theorem 4.5 is only asserted, not demonstrated.

major comments (2)

[Section 4.4, Theorem 4.4 and Remark 4.6] Theorem 4.4 is not established for the hypotheses as stated. The proof in Section 5.2 begins by invoking Theorem 4.5 to obtain E[u] in C([0,∞),H^5) and hence C^{3,α} regularity of the characteristic flow, which is essential for the explicit solution formula (5.21) and for the claim that the pushforward (φ_t)^*u0 remains in H^2. However, Theorem 4.5 requires E[u0] in H^5, and for deterministic initial data E[u0] = u0, so the stated assumption u0 in H^2 is insufficient. Remark 4.6 concedes this additional condition and notes that the solution then loses regularity to H^2 immediately for t>0. The theorem must either be restated with the extra assumption E[u0] in H^5, or supplied with a new existence proof for merely H^2 mean data; as written, the central well-posedness claim is conditional on regularity not present in the theorem statement.
[Section 5.1, Eq. (5.4) and Theorem 4.5] The proof of Theorem 4.5 is incomplete in a load-bearing point. The a priori estimate displayed in (5.4) controls sup_{t∈[0,T]}(||Ω||^2_{H^4} + ||Θ||^2_{H^2}), while Theorem 4.5 concludes (E[u],E[θ]) ∈ C([0,∞),H^5×H^3). The missing H^3 estimate for Θ is not supplied; the text says only that the higher-order H^4 and H^3 estimates for Ω and Θ 'can be established in a similar way.' Since U∈H^5 follows from Ω∈H^4 via the Biot-Savart inequality (4.1), the missing temperature estimate is exactly the piece needed for the stated H^5×H^3 conclusion, and it feeds directly into the characteristic regularity used in the proof of Theorem 4.4. These estimates need to be written out or a precise reference provided.

minor comments (5)

[Section 5.1, Eq. (5.4)] The estimate is written with ||θ||^2_{H^2} although the dependent variable is Θ; please use consistent notation to avoid confusion with the stochastic temperature θ.
[Example 3.7, Eq. (3.35)] The last line of the Leibniz expansion contains L^2_{ξ_k}ω'⊗dθ', which should read L^2_{ξ_k}u'⊗dθ'; as written it introduces the vorticity ω' without definition in this context.
[Remark 3.8] The remark refers to 'equation (3.18)' when discussing the evolution of Θ^(2); the intended reference is likely (3.28) or (3.32).
[Remark 4.7] The statement that equations (2.12) 'lose their parabolic character and become pure transport equations' is easy to misread as applying to the full system (2.12), which retains the Itô correction term; please rephrase to make clear that the parabolicity is in the expectation equations.
[Section 5.1, Step 3] In the uniqueness proof, the Gronwall estimate is written with a proportionality constant that is not explicitly tracked; please make the dependence of the final constant on K(T) explicit for reproducibility.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the well-posedness and moment-closure arguments are derived in the paper, not imported from the cited framework; the only notable defect is a regularity gap, which is a correctness issue rather than a circular reduction.

full rationale

The derivation chain is not circular. The LA SALT Boussinesq model (2.10)-(2.12) is introduced as an explicit modeling choice following [DH19,DHL19], but the paper's central mathematical claims are proven from that model rather than obtained by substituting the desired conclusion into its assumptions. The expectation equations (2.13) are derived by taking expectations of (2.12); the fluctuation equations (3.31), the covariance identities (3.32), (3.34) and (3.36), and Proposition 3.6 are all derived by Itô calculus and standard Lie-derivative manipulations, with source terms expressed through expectation quantities already determined by the closed system (3.26)/(5.1). The well-posedness proof in Section 5 first solves the deterministic vorticity-temperature system (5.2) for E[u] and E[θ], then constructs (u,θ) explicitly through the stochastic characteristics (5.21), and establishes uniqueness by energy estimates. No fitted parameter is renamed as a prediction: the noise fields ξ_k are declared inputs to be fixed by data analysis elsewhere, and the 'climate' and 'weather' are defined, not fitted. The main weakness is that Theorem 4.4 states H^2 initial data while the proof invokes Theorem 4.5, which requires E[u0] ∈ H^5; Remark 4.6 concedes this extra condition. That is a proof gap or overstatement, not a circular step, because the theorem is not reduced to its own statement or to a fit. Self-citations to [DH19], [DHL19], [dLHLT19] provide the framework and a standard Kunita–Itô–Wentzell formula, but the central global-well-posedness and moment-closure results do not reduce to those references; they are established in the present paper by self-contained estimates and explicit constructions.

Assumptions & free parameters 0 free parameters · 4 assumptions · 0 invented entities

The central results rest on the LA SALT modeling postulate, the ellipticity of the noise vector fields, and standard stochastic analysis tools. No free parameters are fitted; the noise fields are assumed given. No new physical entities are introduced.

assumptions (4)

standard math Stochastic basis with i.i.d. Brownian motions and standard Itô calculus tools (Lemmas 4.1, 4.2, BDG inequality).
Used throughout Section 5 for the well-posedness proofs; standard in SPDE theory.
domain assumption Noise vector fields ξ_k are in L^∞([0,T], C^{4+α}(T^2;R^2)) and satisfy the uniform ellipticity condition (4.8).
Section 4.3; the ellipticity is essential for the parabolic energy estimates in Theorem 4.5, giving global regularity of the expectation equations.
domain assumption The LA SALT model is posited by replacing the drift velocity in the SALT transport process by its expectation, as in equation (2.7).
This is the defining modeling choice from [DH19]; the paper assumes this is the correct stochastic model for climate and weather separation without empirical validation.
domain assumption Incompressibility is imposed on the expectation E[u], not on the instantaneous velocity u (Remark 2.2).
Needed for well-posedness; it changes the physical interpretation of the model, making the instantaneous velocity not divergence-free.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Modelling the climate and weather of a 2D Lagrangian-averaged Euler-Boussinesq equation with transport noise." pith.science (2026). https://pith.science/paper/ZOFEQYTG

@misc{pith2026190900388,
  author       = {Pith},
  title        = {Pith review of: Modelling the climate and weather of a 2D Lagrangian-averaged Euler-Boussinesq equation with transport noise},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ZOFEQYTG}},
  note         = {Machine review of arXiv:1909.00388}
}

read the original abstract

The prediction of climate change and its impact on extreme weather events is one of the great societal and intellectual challenges of our time. The first part of the problem is to make the distinction between weather and climate. The second part is to understand the dynamics of the fluctuations of the physical variables. The third part is to predict how the variances of the fluctuations are affected by statistical correlations in their fluctuating dynamics. This paper investigates a framework called LA SALT which can meet all three parts of the challenge for the problem of climate change. As a tractable example of this framework, we consider the Euler--Boussinesq (EB) equations for an incompressible stratified fluid flowing under gravity in a vertical plane with no other external forcing. All three parts of the problem are solved for this case. In fact, for this problem, the framework also delivers global well-posedness of the dynamics of the physical variables and closed dynamical equations for the moments of their fluctuations. Thus, in a well-posed mathematical setting, the framework developed in this paper shows that the mean field dynamics combines with an intricate array of correlations in the fluctuation dynamics to drive the evolution of the mean statistics. The results of the framework for 2D EB model analysis define its climate, as well as climate change, weather dynamics, and change of weather statistics, all in the context of a model system of SPDEs with unique global strong solutions.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

14 extracted references · 11 canonical work pages

[1]

The Burgers' equation with stochastic transport: shock formation, local and global existence of smooth solutions

[AOBdLT18] Diego Alonso-Or´ an, Aythami Bethencourt de Le´on, and So Takao. The Burger’s equation with stochastic tran sport: shock formation, local and global existence of smooth solut ions. arXiv:1808.07821,

work page Pith review arXiv
[4]

[CD80] J. R. Cannon and Emmanuele DiBenedetto. The initial v alue problem for the Boussinesq equations with data in Lp. In Approximation methods for Navier-Stokes problems (Proc. Sympos., Univ. Paderborn, Paderborn, 1979), volume 771 of Lecture Notes in Math. , pages 129–144. Springer, Berlin,

work page 1979
[5]

[DHL19] Theodore D Drivas, Darryl D Holm, and James-Michael Leahy

https://doi.org/10.1017/prm.2019.43. [DHL19] Theodore D Drivas, Darryl D Holm, and James-Michael Leahy. Lagrangian averaged stochastic advection by Lie transport for ﬂuids. Submitted,

doi:10.1017/prm.2019.43 2019
[6]

Implications of Kunita-It\^o-Wentzell formula for $k$-forms in stochastic fluid dynamics

28 D. ALONSO-OR ´AN, A. BETHENCOURT DE LE ´ON, D. HOLM, AND S. TAKAO [dLHLT19] Aythami Bethencourt de Leon, Darryl Holm, Erwin L uesink, and So Takao. Implications of Kunita-Itˆ o-Wentzel l formula for k-forms in stochastic ﬂuid dynamics. arXiv preprint arXiv:1903.07201,

work page Pith review arXiv 1903
[7]

Well-posedness by noise for linear advection of $k$-forms

[dLT19] Aythami Bethencourt de Leon and So Takao. Well-pose dness by noise of linear advection of k-forms. arXiv preprint arXiv:1904.13319,

work page Pith review arXiv 1904
[10]

Stochastic Climate Theory

https://doi.org/10.3934/dcdss.2020071. [GCF16] Georg A Gottwald, Daan T Crommelin, and Christian LE Franzke. Stochastic climate theory. arXiv preprint arXiv:1612.07474,

work page Pith review arXiv
[12]

[Hol15] Darryl D Holm

https://doi.org/10.1007/s00332-017-9425-y . [Hol15] Darryl D Holm. V ariational principles for stochast ic ﬂuid dynamics. Proceedings of the Royal Society A: Mathemat- ical, Physical and Engineering Sciences , 471(2176):20140963,

work page doi:10.1007/s00332-017-9425-y
[13]

[Hol19b] Darryl D Holm

https://doi.org/10.1007/s00332-019-09565-0 . [Hol19b] Darryl D Holm. Stochastic parametrization of the r ichardson triple. Journal of Nonlinear Science. ,

work page doi:10.1007/s00332-019-09565-0

Show all 14 references

[14]

[HS09] Schmah T

https://doi.org/10.1007/s00332-018-9478-6 . [HS09] Schmah T. Holm, D.D. and C. Stoica. Geometric mechanics and symmetry: from ﬁnite to inﬁnite dim ensions, vol- ume

doi
[44]

[DS13] Joseph D Daron and David A Stainforth

Cambridge University Press., 1972 . [DS13] Joseph D Daron and David A Stainforth. On predicting c limate under climate change. Environmental Research Letters, 8(3):034021,

1972
[2005]

[HM66] Jr. H.P . McKean. A class of markov processes associat ed with nonlinear parabolic equations. Proc. Nat. Acad. Sci. U.S.A., 56:1907–1911,

1907
[2013]

Elgindi and I.J

[EJ18] Tarek M. Elgindi and I.J. Jeong. Finite-time singula rity formation for strong solutions to the Boussinesq equat ions. arXiv:1802.09936,

arXiv
[2018]

[CCH+18] Colin Cotter, Dan Crisan, Darryl D Holm, Wei Pan, and Igor Shevchenko

https://doi.org/10.5194/gc-2018-11 . [CCH+18] Colin Cotter, Dan Crisan, Darryl D Holm, Wei Pan, and Igor Shevchenko. Modelling uncertainty using circulation- preserving stochastic transport noise in a 2-layer quasi-g eostrophic model. arXiv preprint arXiv:1802.05711,

2018 arXiv
[2019]

https://doi.org/10.1007/s00332-019-09571-2 . [BAB+17] Judith Berner, Ulrich Achatz, Lauriane Batt´ e, Lisa Ben gtsson, Alvaro de la C´ amara, Hannah M Christensen, Matteo Colangeli, Danielle RB Coleman, Daan Crommelin, Stamen I Do laptchiev, et al. Stochastic parameterization:...

doi

Pith tools

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

[1] [1]

The Burgers' equation with stochastic transport: shock formation, local and global existence of smooth solutions

[AOBdLT18] Diego Alonso-Or´ an, Aythami Bethencourt de Le´on, and So Takao. The Burger’s equation with stochastic tran sport: shock formation, local and global existence of smooth solut ions. arXiv:1808.07821,

work page Pith review arXiv

[2] [4]

[CD80] J. R. Cannon and Emmanuele DiBenedetto. The initial v alue problem for the Boussinesq equations with data in Lp. In Approximation methods for Navier-Stokes problems (Proc. Sympos., Univ. Paderborn, Paderborn, 1979), volume 771 of Lecture Notes in Math. , pages 129–144. Springer, Berlin,

work page 1979

[3] [5]

[DHL19] Theodore D Drivas, Darryl D Holm, and James-Michael Leahy

https://doi.org/10.1017/prm.2019.43. [DHL19] Theodore D Drivas, Darryl D Holm, and James-Michael Leahy. Lagrangian averaged stochastic advection by Lie transport for ﬂuids. Submitted,

doi:10.1017/prm.2019.43 2019

[4] [6]

Implications of Kunita-It\^o-Wentzell formula for $k$-forms in stochastic fluid dynamics

28 D. ALONSO-OR ´AN, A. BETHENCOURT DE LE ´ON, D. HOLM, AND S. TAKAO [dLHLT19] Aythami Bethencourt de Leon, Darryl Holm, Erwin L uesink, and So Takao. Implications of Kunita-Itˆ o-Wentzel l formula for k-forms in stochastic ﬂuid dynamics. arXiv preprint arXiv:1903.07201,

work page Pith review arXiv 1903

[5] [7]

Well-posedness by noise for linear advection of $k$-forms

[dLT19] Aythami Bethencourt de Leon and So Takao. Well-pose dness by noise of linear advection of k-forms. arXiv preprint arXiv:1904.13319,

work page Pith review arXiv 1904

[6] [10]

Stochastic Climate Theory

https://doi.org/10.3934/dcdss.2020071. [GCF16] Georg A Gottwald, Daan T Crommelin, and Christian LE Franzke. Stochastic climate theory. arXiv preprint arXiv:1612.07474,

work page Pith review arXiv

[7] [12]

[Hol15] Darryl D Holm

https://doi.org/10.1007/s00332-017-9425-y . [Hol15] Darryl D Holm. V ariational principles for stochast ic ﬂuid dynamics. Proceedings of the Royal Society A: Mathemat- ical, Physical and Engineering Sciences , 471(2176):20140963,

work page doi:10.1007/s00332-017-9425-y

[8] [13]

[Hol19b] Darryl D Holm

https://doi.org/10.1007/s00332-019-09565-0 . [Hol19b] Darryl D Holm. Stochastic parametrization of the r ichardson triple. Journal of Nonlinear Science. ,

work page doi:10.1007/s00332-019-09565-0