ACE2-ERA5

Ai2 Climate Emulator (ACE) is a family of models designed to simulate atmospheric variability from the time scale of days to centuries.

Disclaimer: ACE models are research tools and should not be used for operational climate predictions.

ACE2-ERA5 is trained on the ERA5 dataset and is described in ACE2: Accurately learning subseasonal to decadal atmospheric variability and forced responses. As part of that paper, the repository containing training and evaluation scripts and configuration files used for this model is located here.

Quick links

• 📃 Paper
• 💻 Code
• 💬 Docs
• 📂 All Models

Inference quickstart

1. Download this repository. Optionally, you can just download a subset of the forcing_data and initial_conditions for the period you are interested in.

2. Update paths in the inference_config.yaml. Specifically, update experiment_dir, checkpoint_path, initial_condition.path and forcing_loader.dataset.path.

3. Install code dependencies with pip install fme.

4. Run inference with python -m fme.ace.inference inference_config.yaml.

Strengths and weaknesses

Briefly, the strengths of ACE2-ERA5 are:

• accurate atmospheric warming response to combined increase of sea surface temperature and CO2 over last 80 years
• highly accurate atmospheric response to El Niño sea surface temperature variability
• good representation of the geographic distribution of tropical cyclones
• accurate Madden Julian Oscillation variability
• realistic stratospheric polar vortex strength and variability
• exact conservation of global dry air mass and moisture

Some known weaknesses are:

• the individual sensitivities to changing sea surface temperature and CO2 are not entirely realistic
• the medium-range (3-10 day) weather forecast skill is not state of the art
• not expected to generalize accurately for large perturbations of certain inputs (e.g. doubling of CO2)