MANGO Dataset : Magnetosphere Atlas from Normalized Geospace Observations

The Magnetosphere Atlas from Normalized Geospace Observations (MANGO) dataset provides a comprehensive, data-driven reconstruction of the Earth’s plasma environment with unprecedented spatial resolution. It aggregates in situ magnetic field and plasma moment (density, temperature, and bulk velocity) measurements collected over 20 years by nine spacecraft across the Cluster, Double Star, THEMIS, and MMS missions. The MANGO processing pipeline employs machine learning algorithms at different stages to: (1) classifies each sample as magnetosphere, magnetosheath, or solar wind, (2) pairs magnetospheric and magnetosheath observations with upstream solar-wind and IMF conditions, (3) normalizes spacecraft positions relative to standardized magnetopause and bow shock surfaces to account for boundary dynamics, and (4) introduces tailored coordinate systems designed to maximize the spatial coverage of the environment. Sampled at a high 5-second resolution, the repository includes both intermediate processing outputs and final curated datasets, culminating in approximately 55 million magnetospheric, 45 million magnetosheath, and 17 million solar wind measurements. The datasets will be made publicly available concurrently with the release of the article preprint.