Baroclinic Wave Simulation Ensemble: a Machine Learning ready dataset.

Area of Science:

• Atmospheric Science
• Climate Science
• Meteorology

Background:

• Extratropical cyclones (ETCs) are crucial for mid-latitude weather dynamics.
• Large ensembles are essential for robust climate and weather system analysis.
• Open science initiatives facilitate data sharing and reproducibility.

Purpose of the Study:

• To generate and provide a comprehensive dataset of baroclinic wave simulations for studying ETCs.
• To analyze mid-latitude dynamics using a large ensemble approach.
• To make the simulation data, extracted features, and code publicly accessible.

Main Methods:

• Ran 6,500 baroclinic wave simulations using the OpenIFS 43R3v2 model via the OpenIFS@home project.
• Tracked cyclones within each simulation and extracted 89 features, including 16 intensity measures.
• Statistically assessed and explained computational failures for 112 ensemble members.

Main Results:

• A dataset comprising raw model output for 6,388 ensemble members and extracted cyclone features was created.
• Distributions of minimum sea level pressure and maximum 850 hPa relative vorticity were plotted for comparison with reanalysis data.
• The dataset and associated code are available for external use.

Conclusions:

• The generated dataset provides a valuable resource for ETC and mid-latitude dynamics research.
• The open accessibility of data and code promotes further scientific investigation and validation.
• The study highlights the utility of distributed computing for large-scale climate simulations.