There are apparently some very capable folks working at Google’s DeepMind. First the brilliant AlphaFold and now GenCast — their New AI model that they claim (see their presentation in Nature here) “advances the prediction of weather uncertainties and risks, delivering faster, more accurate forecasts up to 15 days ahead.” The Nature paper claims that GenCast has “greater skill and speed than the top operational medium-range weather forecast in the world, ENS, the ensemble forecast of the European Centre for Medium-Range Weather Forecasts…GenCast is trained on 40 years of best-estimate analysis from 1979 to 2018, taken from the publicly available ERA5 (fifth generation ECMWF reanalysis) reanalysis dataset. Reanalysis provides a reconstruction of past weather by computing analysis for historical dates and times.” Though the comparison is with the 2019 ENS modeling (and the latter has improved since then), this seems pretty impressive.
