Microsoft Research published results on October 15 for a Probabilistic Bias Correction method that targets systematic errors in existing forecast models to push reliable predictions past the two-week limit.
In tests, the tool improved a European Centre for Medium-Range Weather Forecasts system on 98% of precipitation targets, and the method won top honors in a global forecasting competition. Microsoft developed the work with the Cooperative Institute for Research in the Atmosphere at Colorado State University and NOAA, and cited agriculture as one field that would benefit from longer-range forecasts.