arXiv cs.AI· Marvin Vincent Gabler, Roberto Molinaro, Niall Siegenheim, Henry Martin, Mark Frey, Niels Poulsen, Philipp Seitz, Olivier Lam·· 10 天前AI 评分30
AI 天气模型会漏报极端天气吗？研究对比十二个物理与 AI 模型未发现 AI 特有缺陷
Do AI weather models miss extremes?
AI 导读
一项研究用十个月欧洲地面观测数据，将十二个物理与 AI 预报模型对照 ECMWF IFS 评估 10 m 风速、2 m 气温、太阳辐射和降水，未发现 AI 模型在极端值尾部存在统一的特有缺陷。部分 AI 模型在极端条件下仍比 IFS 更准，另一些则明显退化，物理模型间也有类似差异。所有模型都表现出高值低估、低值高估的条件误差模式，尾部表现取决于模型、变量和评估设置，而非是否由 AI 生成。
正文 · 原文
Abstract:AI weather models are often reported to underestimate extremes, but most evidence concerns deterministic regression models verified against reanalysis. We evaluate twelve physical and AI forecast models against ECMWF IFS using ten months of European station observations. The evaluation covers 10 m wind, 2 m temperature, solar radiation, and precipitation within regimes defined from a fixed ERA5 1991-2020 climatology. We find no uniform AI-specific deficit in the tails. Several AI models remain more accurate than IFS under extreme conditions, while others deteriorate markedly; comparable variation occurs among physical models. Every model nevertheless exhibits a common conditional-error pattern, overpredicting low observations and underpredicting high observations. Attenuation of extreme values therefore does not imply a uniform loss of relative skill: tail performance depends on the model, variable, and evaluation setting rather than on whether the forecast is produced by AI or physical numerical modelling.
| Subjects: | Atmospheric and Oceanic Physics (physics.ao-ph); Artificial Intelligence (cs.AI); Machine Learning (cs.LG) |
| Cite as: | arXiv:2608.09972 [physics.ao-ph] |
| (or arXiv:2608.09972v2 [physics.ao-ph] for this version) | |
| https://doi.org/10.48550/arXiv.2608.09972 arXiv-issued DOI via DataCite |
Submission history
From: Niall Siegenheim [view email]
[v1]
Fri, 31 Jul 2026 14:18:43 UTC (11,528 KB)
[v2]
Tue, 6 Oct 2026 17:28:42 UTC (3,632 KB)
来源：arXiv cs.AI · arxiv.org