spateGAN-ERA5 precipitation downscaling - Datasets

Datasets of "Global spatio-temporal ERA5 precipitation downscaling to km and sub-hourly scale using generative AI" (article) ============================ spateGAN-ERA5 is a deep learning framework designed for spatio-temporal downscaling of ERA5 precipitation data (github) ============================ Here we present data used for training the model in Germany, data for model selection from an independent time frame and evaluation data for Germany, USA and Australia. ============================ Licenses: Data from Australia consists of processed Rainfields 3 operational radar QPE under Creative Commons Attribution NonCommercial 4.0 International license, accessible at https://thredds.nci.org.au/thredds/catalog/rq0/rainfields3/catalog.html. Data from Germany consists of processed RADKLIM-YW rain gauge adjusted radar estimates under Creative Commons BY 4.0 - Licence "CC BY 4.0" license, accessible at https://opendata.dwd.de/climate_environment/CDC/grids_germany/hourly/radolan/reproc/2017_002/netCDF/. Data from the USA consists of processed MRMS weather radar data from the NOAA National Severe Storms Laboratory (NSSL) MRMS system. The MRMS dataset is openly accessible and free to use via the NOAA Open Data Program. It is also accessible at https://mtarchive.geol.iastate.edu/. All data processing and modification steps are explained in https://www.nature.com/articles/s41612-025-01103-y.