Sakhanda Wire
NVDA $153.63 -1.61% MSFT $506.84 -2.15% GOOGL $280.27 -1.22% META $619.77 -1.06% AMZN $256.65 -0.43%
← Back to the news

WeatherNext 3 adds hourly 5 km global forecasting

Google DeepMind and Google Research have released WeatherNext 3, a global weather model designed to address limitations in resolution and update frequency. It uses a live geostationary satellite mosaic together with ECMWF HRES analysis, refreshes its initialization 24 times daily, and produces forecasts at resolutions down to 0.05 degrees, or about 5 km. Training includes raw weather-station observations alongside reanalysis, satellite and precipitation datasets.

One model run generates several levels of output: station-trained temperature and dew point at 0.05 degrees; surface wind, pressure, cloud, solar and precipitation variables at 0.1 degrees; and atmospheric fields at 13 pressure levels at 0.25 degrees. The main 00, 06, 12 and 18 UTC cycles extend to 15 days with 64 ensemble members, while interim hourly runs cover 48 hours.

Google reports precipitation CRPS improvements of up to 60% against NASA’s IMERG data at early forecast ranges. Independent live evaluations by Brightband, as cited by Google AI, rank the model as the most accurate global weather model to date. Forecast data is available through BigQuery, Earth Engine and Cloud Storage after an allowlist request. The weights are not open source, and custom on-demand inference currently uses WeatherNext 2.

Written by Sakhanda Wire AI from reporting by MarkTechPost. Facts, quotes and full context are in the original article below.

Originally published by MarkTechPost on

Read the original on MarkTechPost ↗

The reporting and the image belong to MarkTechPost and are credited and linked above; the text on this page is Sakhanda Wire's own summary.

← Back to the news

More stories

All the latest news