Entropy-Stable Discontinuous Spectral-Element Methods for the Spherical Shallow Water Equations in Covariant Form

This paper introduces high-order, entropy-stable discontinuous spectral-element methods for the rotating shallow water equations on curved manifolds that achieve mass and energy conservation, well-balancing for variable topography, and exact geometric representation through a skew-symmetric covariant formulation.

The Gist

Imagine you are trying to create a digital simulation of the Earth’s weather. To do this, you need to write a set of mathematical "rules" that tell a computer how water (representing the ocean) or air (representing the atmosphere) moves across the surface of a sphere.

The problem is that the Earth isn't a flat map; it’s a curved, bumpy ball. If your math is even slightly "off" because of that curvature, your simulation will eventually "explode"—the numbers will go haywire, the wind speeds will become impossible, and the computer program will crash.

This paper introduces a new, much more stable way to write these rules. Here is the breakdown of how they did it, using some everyday analogies.

1. The Problem: The "Curved Map" Headache

Imagine you are trying to draw a perfectly straight line on a balloon. As you draw, the surface of the balloon stretches and curves under your pen. If you use a standard flat-map ruler, your line will look crooked and weird once you blow the balloon up.

In weather modeling, scientists use "equations" to track things like momentum and pressure. Most methods try to pretend the world is flat and then "fix" the errors caused by the curve. This paper does the opposite: it uses "Covariant" math, which means the math is built specifically to live on a curve. It’s like using a flexible, stretchy ruler designed specifically for balloons, rather than a stiff wooden one.

2. The Solution: The "Energy Budget" (Entropy Stability)

In a perfect world, if you stir a cup of coffee, the energy you put in stays in the coffee. In a computer simulation, however, "math errors" can act like a ghost that adds extra energy to the system. Suddenly, your digital ocean has a massive, impossible wave that appears out of nowhere because of a rounding error. This is called an instability.

The authors created a method called "Entropy-Stable" modeling.

• The Analogy: Think of it like a strict Financial Accountant. In a normal simulation, money (energy) might just appear or disappear due to bad bookkeeping. In this new method, the "Accountant" (the math) is so strict that it ensures energy can only be conserved or lost (dissipated) due to friction—it can never be created out of thin air. This prevents the "ghost energy" from causing the simulation to crash.

3. The Technique: The "Smooth Handshake" (Summation-by-Parts)

When you divide the Earth into a grid (like a giant patchwork quilt of squares), the math has to pass information from one square to the next. If the "handshake" between two squares is clumsy, you get jagged, unrealistic jumps in the data.

The authors use a technique called "Summation-by-Parts" (SBP).

• The Analogy: Imagine a relay race where runners pass a baton. If the handoff is messy, the baton drops, and the race is ruined. SBP ensures that the handoff between every single grid square is mathematically "perfect." It ensures that what leaves one square enters the next square exactly, keeping the "mass" (the amount of water/air) perfectly balanced.

4. The "Lake at Rest" Test (Well-Balancing)

One of the hardest things for a computer to do is to realize when nothing is happening. If you have a perfectly still lake, the computer might see the tiny bumps in the lake floor and think, "Oh! The water is moving uphill!" and start creating fake waves. This is called a failure of "Well-Balancing."

The authors' method is "Well-Balanced," meaning if the water is supposed to be still, the math is smart enough to recognize that the pressure and the gravity are perfectly canceling each other out. It keeps the "lake" perfectly still, just like in real life.

Summary: Why does this matter?

By combining these three things—math that understands curves, a strict energy accountant, and perfect handshakes between grid squares—the researchers have created a "Digital Earth" that is:

1. More Robust: It won't crash when things get violent (like a massive storm).
2. More Accurate: It can use larger, simpler grids without losing the fine details.
3. More Reliable: It can simulate weather patterns for much longer periods without the math "breaking."

In short, they’ve built a much better, more stable "engine" for the next generation of global weather and climate models.

Technical Summary

This paper, titled "Entropy-stable discontinuous spectral-element methods for the spherical shallow water equations in covariant form," presents a novel numerical framework for solving geophysical fluid dynamics equations on curved manifolds.

Below is a detailed technical summary of the work.

1. The Problem

The authors address the challenge of developing high-order numerical methods for the rotating shallow water equations (RSWE) on a sphere. While high-order methods like Discontinuous Galerkin (DG) and Spectral Element Methods (SEM) offer excellent accuracy and scalability, they often suffer from robustness issues. Specifically, their low inherent numerical dissipation makes them susceptible to:

• Aliasing-driven instabilities: Non-physical oscillations caused by under-resolved high-frequency components.
• Geometric complexities: The difficulty of maintaining conservation laws (mass, momentum, energy) and "well-balancing" (preserving steady states like a "lake at rest") when the equations are formulated on curved surfaces with variable topography.

Existing methods often rely on ad hoc stabilization (artificial viscosity) or computationally expensive overintegration. The authors seek a method that is structure-preserving by design.

2. Methodology

The authors propose a discontinuous spectral-element method based on a covariant flux formulation. The core of their methodology includes:

• Covariant Formulation: Instead of using Cartesian coordinates in a higher-dimensional space, the equations are formulated directly on the manifold using covariant derivatives and metric tensors. This allows for a more natural representation of the geometry.
• Skew-Symmetric Splitting: To circumvent the need for discrete versions of the product and chain rules (which are difficult to satisfy in high-order methods), the authors derive a novel skew-symmetric split form of the covariant divergence of the momentum flux. This mathematical trick allows the scheme to satisfy entropy stability and conservation properties more easily.
• Summation-By-Parts (SBP) Operators: The spatial discretization uses tensor-product SBP operators on unstructured quadrilateral grids (specifically a cubed-sphere mesh). These operators mimic the integration-by-parts property at the discrete level.
• Flux-Differencing Framework: They implement a non-conservative, variable-coefficient flux-differencing discretization. This includes:
  • Entropy-Conservative (EC) Volume Fluxes: Two-point flux functions designed to satisfy a "shuffle condition," ensuring no mathematical entropy is produced internally.
  • Entropy-Stable (ES) Interface Fluxes: A local Lax–Friedrichs dissipation term added at element interfaces to damp oscillations while maintaining stability.
• Analytical Geometry: The metric terms and Christoffel symbols are computed analytically from a bilinear mapping, avoiding the errors introduced by approximating metric identities.

3. Key Contributions

1. First Covariant Split-Form/Flux-Differencing Scheme: To the authors' knowledge, this is the first application of a split-form/flux-differencing framework to the covariant formulation of the RSWE.
2. Proven Mathematical Properties: The authors provide theoretical proofs that their scheme is:
   • Mass Conservative: Total mass is preserved globally.
   • Entropy Stable/Conservative: The scheme either preserves total energy (EC) or ensures it is dissipated (ES) rather than produced, preventing non-physical energy growth.
   • Well-Balanced: The scheme can exactly preserve the "lake at rest" equilibrium (constant surface height and zero velocity) even with variable bottom topography.
3. Robustness without Ad Hoc Tuning: The method achieves stability through the mathematical structure of the flux rather than through manually tuned artificial viscosity.

4. Results

The authors validated the scheme using several standard atmospheric test cases on a cubed-sphere grid:

• Unsteady Solid-Body Rotation: Confirmed optimal convergence rates ($N+1$ for the ES scheme) and exponential convergence with respect to polynomial degree.
• Flow Over an Isolated Mountain: Demonstrated well-balancing (the error remained at machine precision for the $V=0$ case) and showed that the ES scheme effectively manages the pressure-topography balance.
• Barotropic Instability: Showed that while the EC scheme can exhibit some oscillations, the ES scheme remains highly robust and accurately captures the growth of instabilities.
• Rossby–Haurwitz Wave: This served as a "stress test" for long-term stability. While standard DG and EC schemes eventually crashed due to uncontrolled energy growth and negative water depths, the ES scheme successfully completed the 28-day simulation, proving its superior robustness for long-term climate/weather modeling.

5. Significance

This work is highly significant for the field of Numerical Weather Prediction (NWP) and Climate Modeling. By providing a high-order, entropy-stable, and well-balanced framework, the authors have laid the groundwork for a new generation of dynamical cores. These cores are essential for global models that require high accuracy, geometric flexibility (to avoid polar singularities), and the robustness to handle complex, non-linear atmospheric flows over long time scales without crashing.