Online Test of a Neural Network Deep Convection Parameterization in ARP-GEM1
This work demonstrates a proof-of-concept for neural network parameterizations in climate models, which could improve computational efficiency, but it is incremental as it emulates an existing scheme.
The study integrated a neural network emulator for deep convection into the ARP-GEM1 atmospheric model, achieving good agreement with the original physics-based scheme in a 5-year simulation.
In this study, we present the integration of a neural network-based parameterization into the global atmospheric model ARP-GEM1, leveraging the Python interface of the OASIS coupler. This approach facilitates the exchange of fields between the Fortran-based ARP-GEM1 model and a Python component responsible for neural network inference. As a proof-of-concept experiment, we trained a neural network to emulate the deep convection parameterization of ARP-GEM1. Using the flexible Fortran/Python interface, we have successfully replaced ARP-GEM1's deep convection scheme with a neural network emulator. To assess the performance of the neural network deep convection scheme, we have run a 5-years ARP-GEM1 simulation using the neural network emulator. The evaluation of averaged fields showed good agreement with output from an ARP-GEM1 simulation using the physics-based deep convection scheme. The Python component was deployed on a separate partition from the general circulation model, using GPUs to increase inference speed of the neural network.