Online Test of a Neural Network Deep Convection Parameterization in ARP-GEM1

TL;DR

This paper demonstrates integrating a neural network-based deep convection parameterization into the ARP-GEM1 atmospheric model via a Python interface, enabling faster inference and comparable simulation results.

Contribution

It introduces a novel method for replacing traditional convection schemes with neural networks in a global atmospheric model using a Python-Fortran interface.

Findings

Neural network emulator achieved good agreement with traditional scheme.

GPU acceleration improved neural network inference speed.

Five-year simulation showed comparable results to physics-based scheme.

Abstract

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.