OneScience-Group/DLESyM

DLESyM

Model Introduction

DLESyM asynchronously couples deep-learning atmosphere and ocean modules for long free-running climate simulations and diagnostic precipitation.

Paper: A Deep Learning Earth System Model for Efficient Simulation of the Observed Climate
https://arxiv.org/abs/2409.16247

Model Description

The model was proposed by an atmospheric-science and machine-learning research team. It was trained with 1983–2017 ERA5 fields, ISCCP OLR, and SST. Coupled DLWP, DLOM, and precipitation modules support current-climate simulation and internal-variability analysis.

Use Cases

Use Case	Description
Long climate simulation	Run stable atmosphere-ocean rollouts.
Climate variability	Analyze ENSO, monsoons, and annular modes.
Precipitation diagnosis	Diagnose accumulated precipitation from atmospheric states.
ModelScope/OneCode execution	Validate data, training, inference, climate metrics, and visualization.
Multi-GPU training	Start multi-process training through torchrun.

Usage Instructions

Use a GPU or DCU when available; CPU supports the default smoke configuration.

hf download OneScience-Group/DLESyM --local-dir ./DLESyM
cd DLESyM

Environment Dependencies

Hardware Requirements

• A GPU or DCU is recommended.
• A CPU can be used for connectivity validation with the default small-sample configuration.
• DCU users should install DTK 25.04.2 or a compatible OneScience-recommended version first.

DCU Environment

# Activate DTK and Conda first
conda create -n onescience311 python=3.11 -y
conda activate onescience311
pip install onescience[earth-dcu] -i http://mirrors.onescience.ai:3141/pypi/simple/ --trusted-host mirrors.onescience.ai

GPU Environment

# Activate Conda first
conda create -n onescience311 python=3.11 -y libstdcxx-ng=12 libgcc-ng=12 gcc_linux-64=12 gxx_linux-64=12
conda activate onescience311
pip install onescience[earth-gpu] -i http://mirrors.onescience.ai:3141/pypi/simple/ --trusted-host mirrors.onescience.ai

python scripts/fake_data.py
python scripts/train.py
torchrun --standalone --nproc_per_node=2 scripts/train.py
python scripts/inference.py
python scripts/result.py

Training jointly optimizes atmosphere, ocean, and precipitation modules. Inference runs four coupled cycles and evaluation reports finite drift diagnostics.

Trained Weights

No weights are bundled under weight/. The authors provide configurations and weights at https://github.com/AtmosSci-DLESM/DLESyM.

Citation and License

This repository is an independent engineering reproduction of the public DLESyM specifications.

The original preprint, official code, model weights, and related data remain subject to their respective licenses and terms.