Additional reporting noted the tool improved an ECMWF system on 98% of precipitation targets in Microsoft's tests.
Following·Day 2·first covered 14 Oct 2026·2 sources
Microsoft AI Bias-Correction Tool Extends Weather Forecast Reliability Past Two Weeks
The gist
Weather forecasts become unreliable beyond two weeks, creating a gap for farmers and others who need longer-range predictions. A method that addresses this through bias correction in existing models could have practical value in those sectors.
The full picture
Microsoft Research developed a Probabilistic Bias Correction technique that corrects systematic errors in existing weather forecast models, improving forecast accuracy beyond the two-week reliability barrier. In tests, the tool improved a European Centre for Medium-Range Weather Forecasts system on 98% of precipitation targets. The method won top honors in a global forecasting competition and is described as showing promise for real-world deployment in sectors such as agriculture that depend on longer-range predictions.
How it developed
Microsoft Research announced Probabilistic Bias Correction method for extended weather forecasting, citing global competition win and agricultural applications.
Sources
Microsoft Research Twitter
The daily email
Want this in your inbox?
I send one email each morning with the stories that moved. If you would rather just read here, that works too.
Subscribe free