by HuaweiCN
Replace hours of supercomputer physics with a learned model that forecasts global weather in seconds, at least as accurately as the best traditional forecast, so far more forecasts can be run and shared.
Published in Nature in 2023; charted by ECMWF and behind Shenzhen's 3 km Zhiji regional forecasts.
Updated 23 Mar 2024Checked 11 Oct0 updates this week
Its 3D Earth-Specific Transformer treats the stack of pressure levels as a cube, so it learns how air at different heights moves together.
Several networks each forecast a different jump ahead; taking the biggest jumps first means fewer steps, so small errors pile up less.
Trained on ERA5, ECMWF's hour-by-hour reconstruction of global weather since 1979, instead of solving physics equations.
Fine-tuned on regional data, Zhiji gives 5-day temperature, rain and wind forecasts for Shenzhen at 3 km, versus ~25 km globally.
| Spec | Pangu-Weather |
|---|---|
| Global grid | 0.25° (~25 km)R (reported) |
| Forecast range tested | 1 hour to 7 daysR (reported) |
| Training data | Hourly ERA5, 1979–2021R (reported) |
| Model size (all networks) | ~256M parametersR (reported) |
| 24-hour global forecast | ~1.4 s on one V100 GPUR (reported) |
| Speed vs. ECMWF IFS | >10,000x fasterR (reported) |
| Zhiji regional model | 3 km, 5-day forecastsR (reported) |
R reported by the company
Huawei
Ascend NPU, Atlas SuperPoD, telecommunications
Huawei designs semiconductors, cloud systems and telecoms gear in 170 countries. Founded 1987, it grew to lead telecoms infrastructure. Now competing directly with NVIDIA in AI accelerators with the Ascend chip series.
