Interpretation of stochastic primitive equations with relaxed hydrostatic assumption as a higher order approximation of 3D stochastic Navier-Stokes

This paper establishes the convergence of solutions from a stochastic 3D Navier-Stokes system to a generalized stochastic primitive equation model that incorporates relaxed hydrostatic assumptions via martingale terms, demonstrating that this modified framework serves as a well-posed, higher-order approximation of the original equations under specific asymptotic scalings and boundary conditions.

The Gist

The Big Picture: Predicting the Ocean's Mood

Imagine trying to predict the weather or the movement of ocean currents. The ocean is a massive, chaotic system. It has huge, slow-moving waves (like a giant conveyor belt) and tiny, frantic ripples (like a swarm of angry bees).

To simulate this on a computer, scientists usually have to make a choice:

1. The "Perfect" Model: Try to track every single bee and every giant wave. This requires so much computing power that it's impossible to run for long periods.
2. The "Simplified" Model: Ignore the tiny bees and only track the giant waves. This is fast, but it misses important details, like deep underwater storms or sudden up-and-down movements of water.

This paper is about building a better simplified model. The authors are trying to prove mathematically that their new, slightly more complex simplified model is actually a much closer copy of the "perfect" (but impossible) model than the old simplified models were.

The Cast of Characters

To understand the paper, let's meet the three main "models" they are comparing:

1. The 3D Navier-Stokes Equations (The "Perfect" Reality):
   Think of this as the High-Definition 4K Movie of the ocean. It captures every swirl, every drop, and every interaction in three dimensions. It is the "truth," but it is too heavy to run on a computer for long.

2. The Strong Hydrostatic Primitive Equations (The "Old" Simplified Model):
   This is the Black-and-White Cartoon. It makes a big assumption: that water pressure at the bottom is just the weight of the water above it, like a stack of pancakes. It assumes water never moves up or down quickly.

   • The Flaw: In the real ocean, water does move up and down violently (like in deep convection or storms). The "pancake" assumption breaks down here, making the cartoon inaccurate for certain events.

3. The "Relaxed" Stochastic Primitive Equations (The "New" Improved Model):
   This is the Color Cartoon with Special Effects. It keeps the simplicity of the pancake stack but adds a "wiggle room" factor. It acknowledges that the water isn't perfectly still; it allows for random, chaotic up-and-down movements (noise) that the old model ignored.

The Core Discovery: The "Goldilocks" Zone

The authors asked: When does the "Color Cartoon" (Model 3) actually look like the "4K Movie" (Model 1)?

They found a specific "Goldilocks" zone where the new model works perfectly, but the old model fails.

• The Old Model (Strong Hydrostatic): Works well only when the ocean is very calm and flat. If the vertical movements get too strong, the error explodes.
• The New Model (Relaxed Hydrostatic): Works well even when there is significant vertical movement, provided the "noise" (the random wiggle) is scaled correctly.

The Analogy of the Tightrope:
Imagine walking on a tightrope (the ocean).

• The Old Model assumes you are walking on a flat, solid bridge. If you try to walk on a tightrope with this model, you will fall because it doesn't account for the swaying.
• The New Model assumes you are on a swaying tightrope. It adds a "balance pole" (the stochastic noise) that helps you stay upright.
• The paper proves that if you adjust the length of that balance pole correctly (a specific mathematical scaling involving the aspect ratio $\epsilon$ and a noise coefficient $\alpha_\sigma$), the swaying tightrope model predicts your path almost exactly as well as the 4K movie of the real world.

How They Did It (The Math Magic)

The authors didn't just guess; they used a rigorous mathematical framework called Location Uncertainty (LU).

1. The "Blur" Technique: Instead of trying to resolve every tiny detail, they treat the small, unresolved movements as "random noise" (like static on an old TV).
2. The "Filter": They introduced a mathematical filter (like a sieve) to smooth out the most chaotic parts of this noise so the equations don't break.
3. The Comparison: They ran a mathematical race between the "Perfect 4K Movie" and their "Color Cartoon." They measured the distance (error) between the two.
   • They found that the error of the New Model is much smaller than the Old Model when the vertical movements are significant.
   • Specifically, they proved that the New Model is a "higher-order approximation." In plain English: It's not just okay; it's significantly better at capturing the complex, 3D nature of the ocean without needing the impossible computing power of the full 3D model.

The Bottom Line

The paper claims that by relaxing the strict rule that "water pressure must act like a stack of pancakes," and instead allowing for random, chaotic vertical movements (stochastic noise), we can create a model that is:

1. Computationally feasible (it runs on computers).
2. Mathematically proven to be much closer to the true physics of the ocean than previous simplified models.

It's like upgrading from a flat map of the ocean to a 3D hologram that still fits in your pocket, allowing scientists to predict complex weather and ocean events with much greater confidence.

Technical Summary

Technical Summary: Interpretation of Stochastic Primitive Equations with Relaxed Hydrostatic Assumption

Problem Statement
This paper investigates the convergence of solutions to the three-dimensional stochastic Navier-Stokes equations (specifically within the Location Uncertainty or LU framework) toward their primitive equations counterparts. The study addresses the limitations of the classical "strong hydrostatic assumption," which posits that vertical acceleration is negligible. While valid for large-scale ocean dynamics, this assumption fails to capture phenomena such as deep convection and strong vertical upwelling/downwelling. The authors aim to determine whether a "relaxed" (or weak) hydrostatic assumption, which retains specific martingale terms representing deviations from hydrostatic equilibrium, provides a superior higher-order approximation of the 3D stochastic Navier-Stokes equations compared to the standard strong hydrostatic model.

Methodology
The authors employ the Location Uncertainty (LU) framework, which models fluid flow by decomposing the Lagrangian displacement into a large-scale Eulerian velocity and a small-scale, highly oscillating noise term. The methodology involves the following steps:

1. Scaling and Asymptotic Analysis: The authors consider a thin domain $S_\epsilon = S_H \times (-\epsilon, \epsilon)$ where $\epsilon$ is the aspect ratio. They derive a scaled version of the LU Navier-Stokes equations (SNS) and introduce a vertical noise scaling coefficient $\alpha_\sigma$ to distinguish between horizontal and vertical noise contributions.
2. Model Derivation:
   • Strong Hydrostatic Model (PE): Derived by formally discarding terms of order $\epsilon^2$ in the vertical momentum equation.
   • Weak Hydrostatic Model (PE$^\epsilon_{\alpha_\sigma}$): Derived by retaining terms of order $\alpha_\sigma \epsilon^2$ and $\alpha_\sigma^2 \epsilon^2$ in the vertical momentum equation. This model includes martingale terms representing the vertical acceleration deviations. To ensure well-posedness, the authors introduce a regularization via a convolution kernel $K$ to filter high-frequency noise contributions in the vertical momentum equation.
3. Mathematical Tools:
   • Modified Projectors: To handle the pressure terms in the stochastic setting, the authors define "modified" Leray projectors ($P$ and $P_\epsilon$). These projectors cancel the scaled gradient terms arising from the stochastic pressure, allowing the application of classical Leray theory tools despite the presence of Itô calculus covariation terms.
   • Convergence Regimes: The analysis covers two boundary condition regimes:
     • Rigid-lid: Used to study $L^2-H^1$ convergence of weak solutions.
     • Fully Periodic: Used to study $H^1-H^2$ convergence of strong solutions.
4. Error Estimation: The authors establish energy estimates for the difference between the solutions of the regularized scaled Navier-Stokes equations ($rSNS^\epsilon_{\alpha_\sigma}$) and the regularized primitive equations ($PE^\epsilon_{\alpha_\sigma}$). They utilize stochastic Grönwall lemmas and Burkholder-Davis-Gundy inequalities to bound the second moments of the error.

Key Contributions and Results

1. Higher-Order Approximation: The paper demonstrates that the weakly hydrostatic primitive equations provide a higher-order approximation of the 3D LU Navier-Stokes equations than the strongly hydrostatic equations, provided the vertical noise coefficient $\alpha_\sigma$ satisfies specific scaling conditions.
2. Convergence Rates:
   • For the weak hydrostatic model, the asymptotic error is bounded by a term of order $O(\alpha_\sigma^2 \epsilon^2)$. Convergence in probability is established when $\alpha_\sigma = o(\epsilon^{-1})$.
   • For the strong hydrostatic model (or the non-regularized Navier-Stokes compared to the weak model), the error bound involves a term of order $O(\alpha_\sigma^4 \epsilon^2)$. Consequently, the weak hydrostatic model is shown to be a significantly better approximation in regimes where $\alpha_\sigma$ is large but $\alpha_\sigma \epsilon$ remains small.
3. Identification of a "Grey Zone": The authors identify a specific regime where $\alpha_\sigma \sim \epsilon^{-\gamma}$ with $\gamma \in [1/2, 1)$. In this zone, the weakly hydrostatic primitive equations constitute a suitable approximation of the regularized Navier-Stokes system, whereas the strongly hydrostatic equations do not.
4. Well-Posedness and Blow-up Time: The study generalizes deterministic well-posedness results for thin domains to the stochastic context. It is shown that the blow-up time of the scaled Navier-Stokes equations tends to infinity in probability as the aspect ratio $\epsilon \to 0$.
5. Technical Innovation: The introduction of a "modified" Leray projector is highlighted as a crucial technical contribution. This allows for the cancellation of stochastic pressure gradient terms that arise from Itô's lemma, facilitating the use of standard functional analysis tools in a stochastic setting where covariation terms would otherwise complicate the estimates.

Significance
The paper claims that its primary significance lies in providing a rigorous mathematical justification for using relaxed hydrostatic assumptions in stochastic geophysical fluid dynamics. By retaining martingale terms as deviations from hydrostatic equilibrium, the proposed model captures non-hydrostatic effects (such as deep convection) while remaining within the computationally tractable primitive equations formalism. The results suggest that for appropriately scaled noises, the weakly hydrostatic model offers a more accurate representation of 3D turbulent flows than the traditional strong hydrostatic approximation, particularly in regimes where vertical accelerations are non-negligible. The work extends deterministic convergence results (specifically those of [34]) to the stochastic setting, addressing the challenges posed by the non-uniqueness of solutions and the need for probabilistic forecasting in climate and ocean modeling.