Machine Learning for Stochastic Parameterization: Generative Adversarial Networks in the Lorenz '96 Model

TLDR

Stochastic parameterizations model uncertainty in unresolved subgrid processes, and while data‑driven approaches exist, they often rely on restrictive structural assumptions that limit scalability. This study develops a stochastic parameterization based on a generative adversarial network (GAN) framework. The GAN is trained on Lorenz '96 model output, with experiments exploring different input‑noise characterizations and running the parameterization at both weather and climate time scales. Certain GAN configurations outperform a bespoke baseline at both time scales, accurately reproducing the Lorenz '96 system’s spatiotemporal correlations and regimes, and skillful forecasts correlate with superior climate simulations.

Abstract

Abstract Stochastic parameterizations account for uncertainty in the representation of unresolved subgrid processes by sampling from the distribution of possible subgrid forcings. Some existing stochastic parameterizations utilize data‐driven approaches to characterize uncertainty, but these approaches require significant structural assumptions that can limit their scalability. Machine learning models, including neural networks, are able to represent a wide range of distributions and build optimized mappings between a large number of inputs and subgrid forcings. Recent research on machine learning parameterizations has focused only on deterministic parameterizations. In this study, we develop a stochastic parameterization using the generative adversarial network (GAN) machine learning framework. The GAN stochastic parameterization is trained and evaluated on output from the Lorenz '96 model, which is a common baseline model for evaluating both parameterization and data assimilation techniques. We evaluate different ways of characterizing the input noise for the model and perform model runs with the GAN parameterization at weather and climate time scales. Some of the GAN configurations perform better than a baseline bespoke parameterization at both time scales, and the networks closely reproduce the spatiotemporal correlations and regimes of the Lorenz '96 system. We also find that, in general, those models which produce skillful forecasts are also associated with the best climate simulations.