The MERRA-2 Gridded Innovations and Observations (GIO) dataset provides assimilated observations and key statistics generated during observational data assimilation. GIO data are created by binning observations and innovations onto grids consistent with MERRA-2 specifications and storing them in NetCDF format for convenient access. This dataset serves multiple purposes: it shows which data was assimilated (where and when), enables systematic evaluation of observing system impacts in reanalysis, provides a resource for teaching data assimilation concepts, and offers training data for machine learning and artificial intelligence applications.

This collection includes hourly values for various measurements (temperature profile [tv], water vapor profile [qv], zonal wind [u], meridional wind [v], surface pressure [ps], and sea surface temperature [sst]) from several conventional observation sources, including radiosondes, land surface stations, aircraft, scatterometers, and drifting buoys. The dataset also contains observation data counts (nobs) and assimilation statistics, including: (a) forecast departure, or observation minus forecast (OmF), and (b) analysis departure, or observation minus analysis (OmA).