The extrAIM dataset: A merged satellite-based daily precipitation dataset for the Mediterranean region (including an ensemble of 20 synthetic realisations)

extrAIM dataset is a new merged daily precipitation product (extraim_merged_data.nc) for the Mediterranean region with the following characteristics:

• Dataset format: NetCDF
• Spatial resolution: 25 x 25 km
• Temporal resolution: 1 day
• Spatial coverage: Longitude: from -6.25 to 38.25, Latitude: 27.75 to 49
• Temporal coverage: 01-01-2007 to 30-09-2021
• Merging approach: Two-step merging (classification and regression)
  • Algorithm: Random Forest for both classification and regression
  • Training strategy: Full training strategy
• Merged precipitation products: SM2Rain-ASCAT and GPM Late Run
• Reference precipitation product: EMO5
• Static covariates: Longitude, Latitude and Elevation, in both classification and regression step
  • Classification step: probability dry and probability dry of the 5 neighboring points around the target locations
  • Regression step: mean, standard deviation and skewness of daily precipitation, of the entire series and non-zero amounts, as well as mean precipitation of the 5 neighboring points around the target locations

In addition, an ensemble of 20 synthetic realizations (equiprobable and bias-adjusted) of the merged dataset is provided (files named: “extraim_realisation_XX.nc”). The synthetic realisations were produced using the extrAIM’s uncertainty-quantification approach and the associated conditional sampling method.