STRATA-SCREAM-sdecadal

Dataset Description: Global, storm-resolving atmospheric simulation output generated by the SCREAM (Simple Cloud-Resolving E3SM Atmosphere Model) physics model. The dataset provides high-resolution global atmospheric state fields suitable for training and evaluating machine-learning weather and climate emulators. The data was produced by an external collaborator and is publicly available at NERSC under CC-BY-4.0. This release is a reformatted, ML-ready version: files have been converted from NetCDF (.nc) to Zarr (.zarr) and subsampled. No underlying data values were altered. This dataset is ready for commercial and non-commercial use. This repository contains the [sdecadal] simulation store. It is part of a three-part release; the other simulation periods are available in the companion repositories (STRATA-SCREAM-sdy1, STRATA-SCREAM-sdy2). Dataset Owner(s): E3SM Project (U.S. Department of Energy) Dataset Creation Date: 2025-09-22 Version: v1.0 Previous Version(s): Original raw SCREAM simulation output data in .nc format: https://portal.nersc.gov/archive/home/n/ndk/www/STRATA2026 License/Terms of Use: Creative Commons Attribution 4.0 International (CC-BY-4.0) Intended Usage: Training and evaluating machine-learning weather and climate emulators using high-resolution global atmospheric simulation data. Dataset Characterization Data Collection Method** Synthetic Labeling Method** Not Applicable Dataset Format Modality: geospatial, timeseries Format: Zarr (.zarr) Dataset Quantification Record Count: 2,017 timesteps (10-minute temporal resolution) Spatial Grid: cubed-sphere ne1024pg2, 25,165,824 columns (~4.9 km), 32 vertical levels Feature Count: 29 variables Simulation: F20TR-SCREAMv1 Ethical Considerations: NVIDIA believes Trustworthy AI is a shared responsibility and we have established policies and practices to enable development for a wide array of AI applications. Developers should work with their internal developer teams to ensure this dataset meets requirements