FV3 | Joseph Mouallem

The Duo-Grid and Cubed-Sphere Grid Imprinting

Fri, 01 Dec 2023 00:00:00 +0000

Motivation

Addressing the challenge of running localized high-resolution simulations within a global model, the duo-grid imprints a finer mesh over a region of interest, reducing computational cost by an order of magnitude compared to traditional regridding approaches.

The problem: grid imprinting

The gnomonic cubed-sphere grid has excellent accuracy and uniformity, but the coordinates have a kink at the cube edges. In the halo region this kink leaves a visible imprint of the cube in the solution and requires special edge handling throughout the solver.

The Duo-Grid

To reduce grid imprinting, we implemented the novel Duo-Grid within FV3. The Duo-Grid remaps a cube face’s data from the neighboring face, moving it from the kinked locations to natural locations along great circle lines using 1D piecewise linear interpolation. A separate 2D interpolation algorithm fills the correct data at the eight corners of the cubed-sphere, which FV3’s 2D advection scheme requires.

C8 cubed-sphere grid with a three-cell halo. Left: “kinked” grid showing halo updated directly from the neighboring face. Right: “extended” grid for the forward face showing data remapped onto the extended grid. Note that the great circle coordinate lines extend from the compute domain into the grid halo without interruption..

Validation

The Duo-Grid was evaluated across a comprehensive suite of idealized test cases spanning both two-dimensional shallow-water dynamics and three-dimensional hydrostatic and non-hydrostatic flows. These tests were designed to assess the impact of the Duo-Grid on grid imprinting, numerical errors, and the overall behavior of the FV3 dynamical core.

The steady-state geostrophic balance test provides a direct assessment of cubed-sphere grid imprinting. When the flow is oriented perpendicular to the cubed-sphere edges, the conventional kinked grid produces errors aligned with the cube geometry. With the Duo-Grid, these grid-aligned errors are substantially reduced.

Your browser does not support the video tag.

Meridional velocity errors of the C48 steady state geostrophic balance flow with a flow oriented perpendicular to the cubed-sphere edges. Duo-Grid significantly reduces these errors.

The improvement extends to fully three-dimensional dynamics. In the baroclinic wave test, the Duo-Grid suppresses the development of cubed-sphere imprinting and errors in the southern hemisphere while maintaining the evolution of the solution over time.

Your browser does not support the video tag.

Time evolution of meridional winds in the three-dimensional baroclinic wave test, demonstrating the reduced cubed-sphere errors in the southern hemisphere with the Duo-Grid.

The Duo-Grid was further evaluated using a broad suite of standard idealized tests, including shallow-water steady-state geostrophic flow, the splash test, Rossby–Haurwitz wave, colliding modons, cosine-bell advection, and the three-dimensional non-hydrostatic baroclinic wave.

Your browser does not support the video tag.

Idealized test suite run on the Duo-Grid: shallow-water steady-state geostrophic flow, splash test, Rossby-Haurwitz wave, colliding modons, cosine bell advection, and the 3D non-hydrostatic baroclinic wave.

Across these tests, the Duo-Grid consistently reduces grid imprinting and numerical errors while preserving the accuracy and numerical characteristics of the original FV3 formulation.

Key results

Grid imprinting of the cubed sphere is greatly reduced in idealized tests and practically eliminated.
Duo-Grid decreases the growth rate of error norms in all cases compared to the kinked grid, up to one order of magnitude.
Order of accuracy of FV3’s horizontal discretization is conserved.
Dispersion and dissipation properties are identical to those of the original FV3 algorithm.
Edge handling code is eliminated -> significant performance gain in current/future GPU development
FV3’s robustness and accuracy have increased.

These results indicate a clear improvement in FV3’s robustness.

Reference

Mouallem, J., Harris, L., and Chen, X.: Implementation of the Novel Duo-Grid in GFDL’s FV3 Dynamical Core, Journal of Advances in Modeling Earth Systems, 15(12), 2023. https://doi.org/10.1029/2023MS003712

Multiple Same-Level and Telescoping Grid Nesting

Tue, 07 Jun 2022 00:00:00 +0000

Motivation

Multi-scale modeling often requires simulations at many different resolutions. This work shows how to couple coarse and fine domains within a single dynamical core, enabling cost-effective high-resolution forecasts of localized phenomena like hurricanes.

Overview

Two-way multiple same-level and telescoping grid nesting capabilities are implemented in FV3 using GFDL’s Flexible Modeling System (FMS).

A nest is an additional grid that zooms in over a region of interest to resolve the small-scale structures needed for better forecasts of localized weather events such as severe storms and hurricanes. A telescoping nest is a nest within a nest, allowing resolution to be refined progressively over the target region.

Nests can be used in both global and regional domains, and each level of the hierarchy can refine the parent resolution by an arbitrary factor.

A telescoping configuration refining a global ~13 km grid down to ~4.3 km, ~1.4 km and ~0.5 km over the region of interest.

Computational design

The nested grids run concurrently on different sets of processors and interact two-way with their parent grids. This provides more accurate results on both the nest and the parent, and reduces load imbalance between processors.

Availability

Starting from the FV3 public release of 2021, multiple same-level and telescoping nests are fully functional and available to the broader scientific community. This drastically improves overall forecast performance and opens the door to numerous research possibilities for scientists and meteorologists alike.

Reference

Mouallem, J., Harris, L., and Benson, R.: Multiple same-level and telescoping nesting in GFDL’s dynamical core, Geoscientific Model Development, 15(11), 4355-4371, 2022. https://doi.org/10.5194/gmd-15-4355-2022

SHiELD-MOM6: High-Resolution Coupled Atmosphere-Ocean Modeling

Fri, 26 Sep 2025 00:00:00 +0000

What I developed

The coupling framework connecting SHiELD, MOM6, and SIS2 through FMS
Exchange-grid infrastructure for conservative flux exchange between components
High-resolution coupled atmosphere-ocean-ice model configurations
Evaluation of hurricane-ocean interaction, including Hurricane Helene (2024)

Motivation

Air-sea interactions drive storm intensity and ocean response. This coupled system captures two-way feedback between the atmosphere and ocean at kilometer scales, enabling accurate simulations of hurricane-ocean interactions and coastal impacts.

Overview

We present a new high-resolution coupled atmosphere-ocean model, SHiELD-MOM6, which integrates GFDL’s advanced atmospheric model, the System for High-resolution modeling for Earth-to-Local Domain (SHiELD), the Modular Ocean Model version 6 (MOM6), and the Sea Ice Simulator (SIS2).

The model leverages the Flexible Modeling System (FMS) coupler and its innovative exchange grid to enable a robust and scalable two-way interaction between the atmosphere and ocean. The atmospheric component is built on the non-hydrostatic Finite-Volume Cubed-Sphere Dynamical Core (FV3) with the latest version of the SHiELD physics parametrization suite, while the ocean component is the latest version of MOM, supporting kilometer-scale high-resolution and regional applications.

Coupling infrastructure

The SHiELD-MOM6 model employs the Flexible Modeling System (FMS) coupler, which facilitates the exchange of information between the atmospheric and oceanic components. The exchange grid ensures accurate and efficient communication, enabling the two-way interaction necessary for realistic coupled simulations.

Schematic of a one-dimensional exchange grid and communication map between the atmosphere and ice components at different resolutions. The red sides of the arrow indicate the step where variables are projected from the exchange grid. The light-blue and mauve sides of the arrows represent the projection of variables onto the exchange grid from the atmosphere or ice components, respectively.

Hurricane Helene (2024)

Validation is demonstrated through a suite of experiments, including idealized hurricane simulations and a realistic North Atlantic case study featuring Hurricane Helene 2024. The animation below shows the simulated sea level pressure and 10 m winds (left) alongside the sea surface temperature anomaly and ocean surface currents (right) as the storm crosses the Gulf.

Your browser does not support the video tag.

Hurricane Helene (2024): sea level pressure and surface winds (left); sea surface temperature change and ocean currents (right). The cold wake and upwelling behind the storm are captured by the two-way coupling.

Scalability

Scalability tests have been conducted to evaluate the model’s performance on massively parallel computing systems. The results demonstrate that SHiELD-MOM6 maintains high computational efficiency as the number of processors increases, ensuring that high-resolution coupled simulations can be performed within practical timeframes

Strong scaling (a) and weak scaling (b): actual speedup/efficiency (red circles) compared to ideal speedup/efficiency (black squares) as a function of the number of PEs.

Key results

Air-sea interactions are effectively captured, both in storm intensity and structure and in the ocean response.
The coupling reproduces ocean phenomena such as storm-induced upwelling, the cold wake, and sea level changes.
Scalability tests confirm the model’s computational efficiency on massively parallel systems.

This work establishes a unified, modular cornerstone for advancing high-resolution coupled modeling, with significant implications for weather forecasting and climate research.

Reference

Mouallem, J., Gao, K., Reichl, B. G., Chilutti, L., Harris, L., Benson, R., Zadeh, N., Chen, J., Chen, J.-H., and Zhang, C.: Development of a high-resolution coupled SHiELD-MOM6 model – Part 1: Model overview, coupling technique, and validation in a regional setup, Geoscientific Model Development, 18(18), 6461-6478, 2025. https://doi.org/10.5194/gmd-18-6461-2025

A Minimal, Adiabatic Example of Sudden Stratospheric Warming

Mon, 01 Sep 2025 00:00:00 +0000

Motivation

Sudden stratospheric warmings have profound impacts on surface weather weeks later. This idealized modeling study isolates the key physical mechanisms driving SSW dynamics, building intuition and validating model representations of stratospheric-tropospheric coupling.

Overview

Sudden Stratospheric Warmings (SSW) are extreme events that can significantly impact weather patterns on short, subseasonal and seasonal timescales. In this study we present a new idealized test case of an SSW event implemented in GFDL’s FV3 dynamical core.

Setup

The initial condition features a wintertime stratospheric circulation with a westerly jet in the Northern Hemisphere and an easterly jet in the Southern Hemisphere. In the absence of tropospheric wave forcing, the model preserves this stratospheric circulation for approximately 200 days, which makes it a clean baseline.

To induce an SSW, we introduce a moving mountain that generates planetary waves of a prescribed zonal wavenumber.

Your browser does not support the video tag.

Animation of the moving mountain forcing used to generate planetary waves of a prescribed zonal wavenumber.

The moving mountain is introduced through a time-dependent surface geopotential perturbation,

$$ \phi’ = g h_0 \sin\left(\frac{r,\mathrm{time}}{20}\right) \sin^2\left[ \frac{\pi(\phi-\phi_1)}{\phi_2-\phi_1} \right] \cos\left( z_w\lambda+\frac{10sr,\mathrm{time}}{360} \right), \qquad \phi_2\geq\phi\geq\phi_1. $$

where (g) is gravitational acceleration, ($h_0$) is the mountain height, ($\phi_1$) and ($\phi_2$) define its latitudinal extent, ($z_w$) is the zonal wavenumber, ($s$) controls the phase speed, and ($r$) controls the temporal forcing frequency. The westward-moving mountain generates planetary waves that propagate upward into the stratosphere and interact with the polar vortex.

Results

The Hovmöller diagram shows the temporal evolution of the zonal-wavenumber components of the 10 hPa zonal wind. The growth and propagation of the planetary-wave components illustrate how the imposed forcing develops and interacts with the stratospheric circulation leading up to the SSW.

Hovmoller diagram for decomposed zonal wind amplitudes at 10⁢ℎ⁡𝑃⁢𝑎 for a perturbed simulation, first row shows wavenumber 0, 1, 2, and 3. Time goes upward in days. Second row shows the decomposed wavenumber 1, 2, and 3 components in zonal winds in (m/s)

The animation shows the evolution of the zonal-mean Eliassen–Palm (EP) flux and its divergence. The upward propagation of EP flux demonstrates the transport of planetary-wave activity into the stratosphere, followed by enhanced wave–mean-flow interaction and deceleration of the polar-night jet.

Your browser does not support the video tag.

Zonal-mean Eliassen–Palm (EP) flux vectors (arrows) and EP flux divergence (shading, in m/s/day). Gray contours denote the zonal-mean zonal wind (in m/s), with solid lines for positive values and dashed lines for negative values.

Polar view of the two SSW regimes obtained in the idealized setup: wavenumber-1 forcing produces a vortex displacement event (left), while wavenumber-2 forcing produces a vortex split event (right).

Wavenumber-1 forcing leads to a vortex displacement SSW.
Wavenumber-2 forcing produces a vortex split SSW.

Both are consistent with observations and the published literature.

This minimal setup offers a controlled environment for studying SSW dynamics and serves as a useful testbed for evaluating the ability of dynamical cores to capture key stratospheric processes and troposphere-stratosphere interactions.

Reference

Mouallem, J., Yao, W., Harris, L., Lin, S.-J., and Chen, X.: A Minimal, Adiabatic Example of Sudden Stratospheric Warming, Journal of Advances in Modeling Earth Systems, 17(9), 2025. https://doi.org/10.1029/2024MS004760

Integrating a Three-Dimensional TKE-Based Subgrid Diffusion Scheme Into GFDL FV3

Mon, 01 Jun 2026 00:00:00 +0000

Development of a high-resolution coupled SHiELD-MOM6 model – Part 1: Model overview, coupling technique, and validation in a regional setup

Fri, 26 Sep 2025 00:00:00 +0000

A Minimal, Adiabatic Example of Sudden Stratospheric Warming

Mon, 01 Sep 2025 00:00:00 +0000

Implementation of the Novel Duo-Grid in GFDL's FV3 Dynamical Core

Fri, 01 Dec 2023 00:00:00 +0000

Multiple same-level and telescoping nesting in GFDL's dynamical core

Tue, 07 Jun 2022 00:00:00 +0000

FV3 | Joseph Mouallem

The Duo-Grid and Cubed-Sphere Grid Imprinting

Motivation

The problem: grid imprinting

The Duo-Grid

Validation

Key results

Reference

Multiple Same-Level and Telescoping Grid Nesting

Motivation

Overview

Progressive refinement

Computational design

Availability

Reference

SHiELD-MOM6: High-Resolution Coupled Atmosphere-Ocean Modeling

What I developed

Motivation

Overview

Coupling infrastructure

Hurricane Helene (2024)

Scalability

Key results

Reference

A Minimal, Adiabatic Example of Sudden Stratospheric Warming

Motivation

Overview

Setup

Results

Reference

Integrating a Three-Dimensional TKE-Based Subgrid Diffusion Scheme Into GFDL FV3

Development of a high-resolution coupled SHiELD-MOM6 model – Part 1: Model overview, coupling technique, and validation in a regional setup

A Minimal, Adiabatic Example of Sudden Stratospheric Warming

Implementation of the Novel Duo-Grid in GFDL's FV3 Dynamical Core

Multiple same-level and telescoping nesting in GFDL's dynamical core