[Paper Review] High-level, high-resolution ocean modeling at all scales with Oceananigans

Gregory L. Wagner, Simone Silvestri+13 more|ArXiv.org|Feb 19, 2025

Methane Hydrates and Related Phenomena3 citations

TL;DR

Oceananigans provides a GPU-optimized, Julia-based, programmable interface to simulate ocean motion from millimeter-scale turbulence to global circulation, enabling high-fidelity, scalable modeling across all scales.

ABSTRACT

We describe the user interface, governing equations, and numerical methods underpinning the community ocean modeling software called "Oceananigans". Oceananigans development has been lead by the Climate Modeling Alliance to build a trainable climate model with quantifiable uncertainty. Oceananigans is written in the Julia programming language, which, like similar recent efforts based on modern programming languages, distinguishes it from usual software based on Fortran. Oceananigans can efficiently simulate all scales of ocean motion, ranging from millimeter-scale turbulence in a small box to planetary-scale ocean circulation. Oceananigans design combines (i) a basic structured finite volume algorithm (ii) optimized for high-resolution simulations on GPUs which is (iii) exposed behind a high-level, programmable user interface. This design negotiates a dual mandate for highest-possible performance (to support state-of-the-art applications) and enhanced accessibility (to facilitate adoption and development). The dual mandate aims ultimately to accelerate the progress of Earth system science. Achieving this aim, however, requires a substantial and sustained increase in the collective effort of Oceananigans development.

Motivation & Objective

• Describe the Oceananigans user interface, governing equations, and numerical methods for all-scale ocean modeling.
• Demonstrate how a high-level interface plus GPU-optimized numerics enables high-resolution simulations across scales.
• Showcase the ability to couple hydrostatic and nonhydrostatic dynamics within a common framework.
• Argue how modern programming languages can accelerate model development and uncertainty quantification in Earth system modeling.

Proposed method

• Present Oceananigans as a Julia-based library with a four-section script-based workflow (grid, model, simulation, post-processing).
• Adopt a basic structured finite-volume approach optimized for GPUs to enable high-resolution simulations.
• Provide both nonhydrostatic and hydrostatic models with flexible grids (rectilinear, curvilinear, cubed-sphere) and multiple vertical coordinates.
• Offer a programmable interface that allows user-defined forcing, boundary conditions, tracers, and closures to be embedded in simulations.
• Incorporate advanced numerical options (WENO advection, various diffusion/closure schemes, pressure solvers) and support for coupled components via ClimaOcean.
• Illustrate extensibility through examples (2D turbulence, passive tracers, flow around a cylinder) and lazy expression trees for diagnostics.

Experimental results

Research questions

• RQ1 How can a modern, script-first interface enable rapid prototyping and execution of ocean models across all scales?
• RQ2 What numerical strategies (e.g., WENO advection, pressure correction, closures) are required to accurately simulate both nonhydrostatic and hydrostatic dynamics on GPU-accelerated hardware?
• RQ3 To what extent can a Julia-based framework provide performance comparable to or surpassing traditional Fortran-based codes while maintaining usability?
• RQ4 What is the potential for coupling Oceananigans with sea ice and atmosphere components for end-to-end Earth system simulations?
• RQ5 What benchmarks and performance metrics demonstrate Oceananigans’ capability to accelerate climate-relevant, high-resolution simulations?

Key findings

• Oceananigans delivers a programmable, library-style interface enabling one-script simulations of complex flows (e.g., 2D turbulence) on GPUs.
• The system supports nonhydrostatic and hydrostatic modeling across rectilinear and curvilinear grids, with various vertical coordinate options and closures.
• GPU-optimized, structured finite-volume discretization yields high performance, including substantial speedups over traditional CPU-based models and enabling near-globally resolved simulations.
• The framework integrates forcing, tracers, and diagnostics within a flexible, script-driven workflow that supports on-the-fly analysis and configurable boundaries.
• Preliminary performance benchmarks indicate substantial efficiency gains, enabling high-resolution simulations and potential routine climate-scale studies with mesoscale-resolving fidelity.
• Oceananigans aims to enable ensemble calibration and uncertainty quantification driven by trainable parameterizations and differentiable workflows.