Accurate medium-range global weather forecasting with 3D neural networks
Problem
Numerical weather prediction is accurate but computationally expensive, while prior AI forecasting methods were much faster but significantly less accurate. The paper aims to make AI-based medium-range global weather forecasting accurate.
Why it matters
Weather forecasting is important for science and society, and current NWP methods are computationally expensive.
What the AI did
Trained three-dimensional deep networks equipped with Earth-specific priors on 39 years of global data, using a hierarchical temporal aggregation strategy to reduce accumulation errors in medium-range forecasting.
Results
not stated in source
Impact
Abstract shows the AI method handles complex weather patterns and reduces medium-range error accumulation, but quantitative accuracy claims are truncated.
Record
| Venue | Nature |
|---|---|
| Published | 2023 |
| Work type | article |
| Authors | Kaifeng Bi, Lingxi Xie, Hengheng Zhang, Xin Chen, Xiaotao Gu, Qi Chuan Tian |
| Citations | 1,631 (OpenAlex, at import time) |
| References | 44 |
| Model or method | 3D neural networks with Earth-specific priors |
| Organization | not stated in source |
| Primary source | https://doi.org/10.1038/s41586-023-06185-3 |
| Open access | hybrid |
| Keywords | weather forecasting, deep learning, 3D neural networks, medium-range forecasting, NWP |
Importance rating
major — millions of lives, an industry or a field
Abstract states weather forecasting is important for science and society and that NWP is computationally expensive; method beats ECMWF IFS, affecting a major societal-scale domain.