Verification of the data-driven weather model Bris

[tnipen]

The dataset contains verif files for the data-driven weather model called Bris, which is developed by MET Norway. It is based on the Anemoi framework for building machine learning weather models. The data can be downloaded from: https://zenodo.org/records/13644810

The dataset contains 2m temperature (t2m), 6-h precipitation accumulation (precip6h), 10m wind speed (ws10m), and mean sea level pressure (mslp). Data for each variable is stored in separate zip files. Each zip file includes forecasts from 4 sources:

• MEPS_control.nc: The control run of the MetCoOp Ensemble Prediction System (MEPS), which is a high-resolution numerical weather prediction model (NWP) at 2.5 km grid spacing.
• MEPS_ens_mean.nc: The ensemble mean of MEPS
• IFS_control.nc: The control run of the Integrated Forecasting System (IFS), a global NWP model at 0.1° resolution.
• Data-driven.nc: A stretched-grid data-driven model running at 2.5 km resolution.

Forecasts are interpolated bilinearly to a set of SYNOP observation stations for Norway.

Data access
https://zenodo.org/records/13644810/files/mslp.zip?download=1
https://zenodo.org/records/13644810/files/t2m.zip?download=1
https://zenodo.org/records/13644810/files/precip6h.zip?download=1
https://zenodo.org/records/13644810/files/ws10m.zip?download=1