Dataset for 'Stable Machine Learning based Radiation Emulation for ICON'

Folder inlcudes data to reproduce 'Stable Machine Learning based Radiation Emulation for ICON' ICON v2.6.4 (ICON Partnership, 2025) was run using three different setups: - radiation with high frequent calls (every 6 min) (hf) - radiation with low frequent calls (every 60 min) (lf) - ML radiation calls (ml) For every setup, we made - 10 ensemble runs (different inital conditions) 1 year each - 1 10 year long simulation - SST+4K for 3 years - SST+8K for 3 years ML-based radiation emulator is based on Hafner et al. 2025 and was coupled to ICON using FTorch (Atkinson et al. 2025). References: Atkinson, J., Elafrou, A., Kasoar, E., Wallwork, J. G., Meltzer, T., Clifford, S., et al. (2025). Ftorch: a library for coupling pytorch models to fortran. Journal of Open Source Software, 10 (107), 7602. doi: 10.21105/joss.07602 Hafner, K., Iglesias-Suarez, F., Shamekh, S., Gentine, P., Giorgetta, M. A., Pincus, R., & Eyring, V. (2025). Interpretable machine learning-based radiation emulation for ICON. Journal of Geophysical Research: Machine Learning and Computation, 2, e2024JH000501. https://doi.org/10.1029/2024JH000501 ICON Partnership (DWD, MPI-M, DKRZ, KIT, and C2SM): ICON release 2024.01, ICON Partnership [code], https://doi.org/10.35089/WDCC/IconRelease01, 2024.