Bris - High-Resolution Probabilistic Data-Driven Weather Modeling
Abstract
Data-driven weather forecasting has over the last few years stunned people, showing competetive predictions at the fraction of the cost of conventional state-of-the-art models. However, most data-driven models are trained on ERA5 with ~31 km horizontal resolution. Is it possible to replicate the results in a high-resolution setting? We introduce Bris - a high-resolution probabilistic data-driven model trained on the MetCoOp Ensemble Prediction System (MEPS) analysis at 2.5 km. Bris means *breeze* in Norwegian, and is also pronounced similarly (no abbreviation/acronym). The work is carried out at the Norwegian Meteorological Institute in collaboration with the European Centre for Medium-range Weather Forecasting (ECMWF).
Publication details
- Journal / source
- Zenodo (CERN European Organization for Nuclear Research)
- Publication date
- 19 January 2026