Machine Learning Climate Model Dynamics: Offline versus Online Performance

TL;DR

This study compares machine learning models, neural networks and random forests, for improving climate model parametrizations, highlighting differences in offline accuracy and online stability in coupled atmospheric simulations.

Contribution

It demonstrates the contrasting offline and online performance of ML models in climate simulations, emphasizing the importance of stability in coupled models.

Findings

Neural networks outperform random forests offline.

Neural networks cause simulation crashes online.

Both ML models improve weather forecast accuracy.

Abstract

Climate models are complicated software systems that approximate atmospheric and oceanic fluid mechanics at a coarse spatial resolution. Typical climate forecasts only explicitly resolve processes larger than 100 km and approximate any process occurring below this scale (e.g. thunderstorms) using so-called parametrizations. Machine learning could improve upon the accuracy of some traditional physical parametrizations by learning from so-called global cloud-resolving models. We compare the performance of two machine learning models, random forests (RF) and neural networks (NNs), at parametrizing the aggregate effect of moist physics in a 3 km resolution global simulation with an atmospheric model. The NN outperforms the RF when evaluated offline on a testing dataset. However, when the ML models are coupled to an atmospheric model run at 200 km resolution, the NN-assisted simulation crashes with 7 days, while the RF-assisted simulations remain stable. Both runs produce more accurate weather forecasts than a baseline configuration, but globally averaged climate variables drift over longer timescales.