EGU General Assembly 2022
© Author(s) 2022. This work is distributed under
the Creative Commons Attribution 4.0 License.

Climate-Invariant, Causally Consistent Neural Networks as Robust Emulators of Subgrid Processes across Climates

Tom Beucler1, Fernando Iglesias-Suarez2, Veronika Eyring2, Michael Pritchard3, Jakob Runge4, and Pierre Gentine5
Tom Beucler et al.
  • 1Institute of Earth Surface Dynamics, University of Lausanne, Lausanne, Switzerland
  • 2Institut für Physik der Atmosphäre, German Aerospace Center (DLR), Oberpfaffenhofen, Germany
  • 3Department of Earth System Science, University of California, Irvine, USA
  • 4Institute of Data Science, German Aerospace Center (DLR), Jena, Germany
  • 5Department of Earth and Environmental Engineering, Columbia University, New York, USA

Data-driven algorithms, in particular neural networks, can emulate the effects of unresolved processes in coarse-resolution Earth system models (ESMs) if trained on high-resolution simulation or observational data. However, they can (1) make large generalization errors when evaluated in conditions they were not trained on; and (2) trigger instabilities when coupled back to ESMs.

First, we propose to physically rescale the inputs and outputs of neural networks to help them generalize to unseen climates. Applied to the offline parameterization of subgrid-scale thermodynamics (convection and radiation) in three distinct climate models, we show that rescaled or "climate-invariant" neural networks make accurate predictions in test climates that are 8K warmer than their training climates. Second, we propose to eliminate spurious causal relations between inputs and outputs by using a recently developed causal discovery framework (PCMCI). For each output, we run PCMCI on the inputs time series to identify the reduced set of inputs that have the strongest causal relationship with the output. Preliminary results show that we can reach similar levels of accuracy by training one neural network per output with the reduced set of inputs; stability implications when coupled back to the ESM are explored.

Overall, our results suggest that explicitly incorporating physical knowledge into data-driven models of Earth system processes may improve their ability to generalize across climate regimes, while quantifying causal associations to select the optimal set of inputs may improve their consistency and stability.

How to cite: Beucler, T., Iglesias-Suarez, F., Eyring, V., Pritchard, M., Runge, J., and Gentine, P.: Climate-Invariant, Causally Consistent Neural Networks as Robust Emulators of Subgrid Processes across Climates, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-722,, 2022.