Abstract:Data‐driven models can emulate the gravity‐wave drags in a one‐dimensional quasibiennial oscillation model on which they were trained accurately, yielding the correct winds when coupled back to the one‐dimensional model. However, they are sensitive to perturbations in the gravity‐wave sources, for example, due to climate model biases or climate change. An effective solution for model biases is to remap the wave sources before feeding them to the data‐driven methods. However, the response to climate change is an ongoing challenge. Two key challenges in the development of data‐driven gravity‐wave parameterizations are generalization, how to ensure that a data‐driven scheme trained on the present‐day climate will continue to work in a new climate regime, and calibration, how to account for biases in the "host" climate model. Both problems depend fundamentally on the response to out‐of‐sample inputs compared with the training dataset, and are often conflicting. The ability to generalize to new climate regimes often goes hand in hand with sensitivity to model biases. To probe these challenges, we employ a one‐dimensional (1D) quasibiennial oscillation (QBO) model with a stochastic source term that represents convectively generated gravity waves in the Tropics with randomly varying strengths and spectra. We employ an array of machine‐learning models consisting of a fully connected feed‐forward neural network, a dilated convolutional neural network, an encoder–decoder, a boosted forest, and a support‐vector regression model. Our results demonstrate that data‐driven schemes trained on "observations" can be critically sensitive to model biases in the wave sources. While able to emulate accurately the stochastic source term on which they were trained, all of our schemes fail to simulate fully the expected QBO period or amplitude, even with the slightest perturbation to the wave sources. The main takeaway is that some measures will always be required to ensure the proper response to climate change and to account for model biases. We examine one approach based on the ideas of optimal transport, where the wave sources in the model are first remapped to the observed one before applying the data‐driven scheme. This approach is agnostic to the data‐driven method and guarantees that the model adheres to the observational constraints, making sure the model yields the right results for the right reasons.

Explainable Offline‐Online Training of Neural Networks for Parameterizations: A 1D Gravity Wave‐QBO Testbed in the Small‐Data Regime

Explainable Offline-Online Training of Neural Networks for Parameterizations: A 1D Gravity Wave-QBO Testbed in the Small-data Regime

On the importance of learning non-local dynamics for stable data-driven climate modeling: A 1D gravity wave-QBO testbed

The graft‐versus‐host problem for data‐driven gravity‐wave parameterizations in a one‐dimensional quasibiennial oscillation model

Data Imbalance, Uncertainty Quantification, and Generalization via Transfer Learning in Data-driven Parameterizations: Lessons from the Emulation of Gravity Wave Momentum Transport in WACCM

Spatially Extended Tests of a Neural Network Parametrization Trained by Coarse-graining

Non‐local parameterization of atmospheric subgrid processes with neural networks

Uncertainty Quantification of a Machine Learning Subgrid‐Scale Parameterization for Atmospheric Gravity Waves

Neural Network Parameterization of Subgrid‐Scale Physics From a Realistic Geography Global Storm‐Resolving Simulation

Use of neural networks for stable, accurate and physically consistent parameterization of subgrid atmospheric processes with good performance at reduced precision

Interpreting and Stabilizing Machine-learning Parametrizations of Convection

Recreating Observed Convection‐Generated Gravity Waves From Weather Radar Observations via a Neural Network and a Dynamical Atmospheric Model

Sensitivity of the Quasi‐biennial Oscillation Simulated in WACCM to the Phase Speed Spectrum and the Settings in an Inertial Gravity Wave Parameterization

Data-driven multiscale modeling of subgrid parameterizations in climate models

Regression Forest Approaches to Gravity Wave Parameterization for Climate Projection

Gradient-free online learning of subgrid-scale dynamics with neural emulators

Adjoint-based online learning of two-layer quasi-geostrophic baroclinic turbulence

Overcoming set imbalance in data driven parameterization: A case study of gravity wave momentum transport

Machine Learning Global Simulation of Nonlocal Gravity Wave Propagation

Multi-Scale Dynamics of the Interaction Between Waves and Mean Flows: From Nonlinear WKB Theory to Gravity-Wave Parameterizations in Weather and Climate Models