Searcharxiv⌕ Search
Home/Archive/arXiv

arXiv · 2608.09972

Do AI weather models miss extremes?

Marvin Vincent Gabler·Roberto Molinaro·Niall Siegenheim·Henry Martin·Mark Frey·Niels Poulsen·Philipp Seitz·Olivier Lam
View original source ↗Read source document ↗Export metadata ↓

Abstract

First-generation AI weather models are often reported to underperform at extremes, mostly in reanalysis-based evaluations of deterministic regression systems. We verify eleven physical and AI forecast systems against European synoptic, solar, and rain-gauge stations over ten months for 10 m wind, 2 m temperature, hourly shortwave accumulation, and hourly precipitation, scoring mean absolute error (MAE) against ECMWF IFS in ERA5 1991-2020 climatological regimes. Among these systems, AI models do not show a uniform relative-skill deficit in the tails. Jua EPT-2.1 Europa leads all-conditions wind (+8.4%), while Jua EPT-2 HRRR leads temperature overall (+12.1%) and in the heat regime (+19.6 +/- 2.2%). EPT-2.1 Europa and DWD ICON Global lead at gale-force wind. Jua EPT-2.1 Helios leads solar overall (+10.2 +/- 1.7%), in overcast conditions (+16.4 +/- 3.4%), and in the clear-sky tail (+24.8 +/- 5.4%). For precipitation, three Jua models gain 14-15% at moderate intensity and 9-11% at P75-P95; EPT-2 Reasoning remains ahead above P95 (+1.7 +/- 0.5%). Failures are model-specific: ECMWF AIFS loses 4.9 +/- 2.0% in the heat tail, while NOAA GFS loses 22.8 +/- 2.0% there. Every model, including numerical weather prediction systems, shows a shared conditional bias toward the centre of the observed distribution, with an inter-model spread several times smaller than the shared signal. Missing relative skill at extremes is therefore not a property of AI weather models as a class, but of particular AI and physical models.

Explore related subjects

physics.ao-phcs.AIcs.LG

Keep this discovery

BibTeXRIS

Marvin Vincent Gabler, Roberto Molinaro, Niall Siegenheim, Henry Martin, Mark Frey, Niels Poulsen, Philipp Seitz, Olivier Lam. 2026-07-31. Do AI weather models miss extremes?. https://arxiv.org/abs/2608.09972

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

arXiv2026

Windowed Envelope Statistics for Time-Domain Significant Wave Height Estimation From HF Radar

Significant wave height (SWH) retrieval from high-frequency (HF) radar typically relies on a weak second-order Doppler continuum that is sensitive to noise, interference, and spectral leakage. This letter presents a Windowed Envelope Statistics Estimator (WESE) that operates directly on beam-formed time-domain voltages. A second-order term obtained from a Neumann expansion of the rough-surface field equation motivates quadratic compensation of localized radar features. WESE extracts the mean, standard deviation, or variance from overlapping windows of the in-phase, quadrature, or envelope-magnitude sequence, followed by quadratic compensation, rank ordering, least-squares regression, and causal smoothing. Evaluation used 335 synchronized hourly observations from a 13.385 MHz, 12-element WERA system at Argentia, Newfoundland and Labrador. The optimal configuration used quadrature variance, a 16-sample window, 896 retained chronological samples, and 30-h smoothing, achieving an RMSE of 0.152 m and a Pearson correlation of 0.978. This represents RMSE reductions of 32.1% and 18.7% relative to previously reported linear and second-order compensated ordered-statistics models, respectively. The results demonstrate robust time-domain SWH estimation without explicit Doppler-spectrum construction.

physics.ao-ph↗
arXiv2026

KiloDA: Reconstructing kilometer-scale near-surface wind states from sparse station observations

Accurate kilometer-scale near-surface winds are important for understanding atmospheric processes over complex terrain, yet remain difficult to reconstruct from sparse and unevenly distributed observations. Here we introduce KiloDA, a diffusion framework for hourly kilometer-scale wind reconstruction from surface stations. KiloDA learns the statistical distribution and spatial structure of wind fields from historical 3-km Weather Research and Forecasting (WRF) model forecasts. At each reconstruction time, no contemporaneous WRF field is used. Instead, station observations provide the only constraints on the current atmospheric state and guide posterior sampling from the learned prior. In idealized WRF experiments, KiloDA recovers localized wind structures when only 0.24% of grid cells are observed and shows an overall advantage over conventional interpolation across terrain conditions and wind speed regimes. This capability largely transfers to real observations. In a fully withheld region, KiloDA reduces the median wind speed root mean square error (RMSE) by 19% relative to ERA5 reanalysis, using only observations outside the region, with the largest improvements over high-elevation and high-relief terrain. A random station holdout further confirms that this advantage extends across different complex-terrain locations and holdout configurations. These results show that historical model archives can provide useful structural knowledge for reconstructing kilometer-scale wind fields from sparse observations without requiring an accurate model estimate of the current atmospheric state.

physics.ao-ph↗