Huawei's Pangu Meteorological Model Assists ECMWF in Accurate Weather Forecasts
The European Centre for Medium-Range Weather Forecasts (ECMWF) released its comparison test report conducted from April to July this year that sets the Pangu Meteorological Model and the European Numerical Model against each other. The report shows that the Pangu Meteorological Model displays advantages across a series of precision indicators, and shined when handling extreme weather forecasts that were concerning meteorologists. It has proven itself in terms of competitiveness, and its potential remains great.
The report points out that the AI method adopted by Pangu will change the trajectory of numerical weather forecast precision improvement over the coming years.
Due to the limited accuracy of meteorological observation, the complexity of physical processes in the atmospheric system, and the huge scale of computing resources required by traditional numerical methods, the scope of global medium-range weather forecasts can only be improved by one day every 10 years. The data-driven AI approach will become a revolutionary force, enabling high-precision predictions quickly at lower computing costs.
ECMWF added a range of data-driven AI predictive models as part of its platform solution. Pangu forecast maps are displayed on the website in six dimensions: mean sea level atmospheric pressure and 850 hPa wind speed, 500 hPa altitude and 850 hPa temperature, mean sea level atmospheric pressure and 200 hPa wind, temperature and geopotential at different pressure levels, 2 m temperature and 10 m air volume of the sea level, and finally wind and geopotential at different pressure levels. This information is critical for predicting the development of weather systems, storm paths, air quality and climate patterns.
The Pangu meteorological model also played a role in accurately predicting the path of Dusuri — this year's largest typhoon. Since it reached the cover of Nature magazine, more and more weather enthusiasts have begun to pay attention to the results of Pangu forecasts.