🌎 A framework for simulating the Earth system at all scales with prescribed or prognostic atmosphere, ocean, sea ice and land components, based on Oceananigans.

NumericalEarth.jl provides infrastructure for running Earth system model components—ocean, atmosphere, sea ice, and others—coupled together or driven by prescribed datasets. The coupling interface is generic: plug in Oceananigans for ocean dynamics, ClimaSeaIce for sea ice, SpeedyWeather or other atmospheric models, or use reanalysis products like JRA55 and ERA5 as prescribed forcing.

The package handles the complexity of component communication: interpolating between grids, computing air-sea fluxes via similarity theory, managing radiative transfer, and synchronizing time-stepping across components with different temporal resolutions.

NumericalEarth.jl also serves as a sandbox for developing and testing interface parameterizations—bulk flux formulations, roughness length models, albedo schemes, and other boundary layer physics—in a modular environment where they can be validated against observations before deployment in production climate models.

NumericalEarth is a registered Julia package. So to install it,

1. Download Julia (version 1.10 or later).

2. Launch Julia and type

julia> using Pkg

julia> Pkg.add("NumericalEarth")

This installs the latest version that's compatible with your current environment.

Running realistic Earth system simulations requires wrangling gigabytes of observational and reanalysis data into formats your model can ingest. NumericalEarth.jl abstracts away this pain. Point the package at a dataset and a target grid, and it handles the downloading, caching, and regridding automatically.

The Metadatum abstraction provides a unified interface: whether you're initializing ocean temperature from reanalysis or prescribing atmospheric boundary conditions, the workflow is the same.

using NumericalEarth
using Dates

# Load temperature from reanalysis on a specific date
T_init = Metadatum(:temperature; date=DateTime(1993, 1, 1), dataset=ECCO2Daily())

# Build a prescribed atmosphere from JRA55
atmosphere = JRA55PrescribedAtmosphere(arch)

The coupling infrastructure is anchored by EarthSystemModel, which encapsulates the component models—ocean, sea ice, atmosphere—and specifies how they communicate. Each component can be either prognostic (time-stepped by its own dynamics) or prescribed (interpolated from data). The model handles flux computations at interfaces, grid interpolation between components, and synchronized time-stepping.

We conceive of EarthSystemModel as a model in its own right, not just a container for components. This means it works with all the Oceananigans tools you'd use for any other model—run!(simulation), Callback, Checkpointer, output writers, and the rest.

To illustrate, here's a global ocean simulation driven by prescribed atmospheric reanalysis:

using Oceananigans
using Oceananigans.Units
using Dates
using CUDA
import NumericalEarth

arch = GPU()
grid = LatitudeLongitudeGrid(arch,
                             size = (1440, 560, 10),
                             halo = (7, 7, 7),
                             longitude = (0, 360),
                             latitude = (-70, 70),
                             z = (-3000, 0))

bathymetry = NumericalEarth.regrid_bathymetry(grid)
grid = ImmersedBoundaryGrid(grid, GridFittedBottom(bathymetry))

# Build an ocean simulation initialized to the ECCO state estimate
ocean = NumericalEarth.ocean_simulation(grid)
start_date = DateTime(1993, 1, 1)
set!(ocean.model,
     T=NumericalEarth.Metadatum(:temperature; date=start_date, dataset=NumericalEarth.ECCO2Daily()),
     S=NumericalEarth.Metadatum(:salinity;    date=start_date, dataset=NumericalEarth.ECCO2Daily()))

# Couple the ocean to JRA55 atmospheric forcing
atmosphere = NumericalEarth.JRA55PrescribedAtmosphere(arch)
coupled_model = NumericalEarth.OceanOnlyModel(ocean; atmosphere)
simulation = Simulation(coupled_model, Δt=20minutes, stop_time=30days)
run!(simulation)

This simulation achieves approximately 8 simulated years per day of wall time on an Nvidia H100 GPU.

Since ocean.model is an Oceananigans.HydrostaticFreeSurfaceModel, we can leverage Oceananigans features directly. For example, to plot the surface speed at the end of the simulation:

u, v, w = ocean.model.velocities
speed = Field(sqrt(u^2 + v^2))
compute!(speed)

using GLMakie
heatmap(view(speed, :, :, ocean.model.grid.Nz), colorrange=(0, 0.5), colormap=:magma, nan_color=:lightgray)

which produces

NumericalEarth is a registered Julia package. To install from a Julia REPL:

julia> using Pkg

julia> Pkg.add("NumericalEarth")

Use Pkg.add(url="https://github.com/NumericalEarth/NumericalEarth.jl.git", rev="main") to install the latest development version.

If you use NumericalEarth for your research, teaching, or fun 🤩, everyone in our community will be grateful if you give credit by citing the corresponding Zenodo record, e.g.,

Wagner, G. L. et al. (2026). NumericalEarth/NumericalEarth.jl: v0.1.1 (v0.1.1). Zenodo. https://doi.org/10.5281/zenodo.18665405

and also the recent preprint submitted to the Journal of Advances in Modeling Earth Systems that presents an overview of all the things that make Oceananigans unique:

"High-level, high-resolution ocean modeling at all scales with Oceananigans"

by Gregory L. Wagner, Simone Silvestri, Navid C. Constantinou, Ali Ramadhan, Jean-Michel Campin, Chris Hill, Tomas Chor, Jago Strong-Wright, Xin Kai Lee, Francis Poulin, Andre Souza, Keaton J. Burns, Siddhartha Bishnu, John Marshall, and Raffaele Ferrari

arXiv:2502.14148

@article{Oceananigans-overview-paper-2025,
  title = {{High-level, high-resolution ocean modeling at all scales with Oceananigans}},
  author = {G. L. Wagner and S. Silvestri and N. C. Constantinou and A. Ramadhan and J.-M. Campin and C. Hill and T. Chor and J. Strong-Wright and X. K. Lee and F. Poulin and A. Souza and K. J. Burns and S. Bishnu and J. Marshall and R. Ferrari},
  journal = {arXiv preprint},
  year = {2025},
  archivePrefix = {arXiv},
  eprint = {2502.14148},
  doi = {10.48550/arXiv.2502.14148},
  notes = {submitted to the Journal of Advances in Modeling Earth Systems},
}