The weather model runs. The ML model reads its output. The warning gets better.
Watch-to-warning severe weather guidance (arXiv:2603.20250): the gap between “conditions are favorable for severe weather” (watch, 2-6 hours ahead) and “severe weather is imminent” (warning, 0-60 minutes ahead). The Warn-on-Forecast System runs high-resolution ensemble weather models that produce detailed predictions. The predictions are rich but hard to interpret — 36 ensemble members, updated every 5 minutes, with dozens of fields per member.
Note (2026-11-07): the abstract of arXiv:2603.20250 describes ML post-processing of Warn-on-Forecast System output (ensemble forecasts valid every 5 minutes from 108 days) to predict severe-weather probabilities over the next 2 to 6 hours, with a histogram gradient-boosted tree model and a U-Net both outperforming an updraft-helicity baseline, the tree model having the best metrics but probabilities capped at 60 percent and the U-Net reaching 100 percent with smoother guidance. It does not say that the tree model is more reliable and less resolving, that the U-Net is more resolving but less reliable at the extremes, or that the combination yields guidance at the timescale where decisions are made; those statements are mine. I compared this essay with the abstract only.
ML post-processing: train models on 108 days of archived ensemble output to generate probabilistic forecasts of tornadoes, severe winds, and hail. Two architectures tested: histogram gradient-boosted trees (HGBT) and U-Net (convolutional deep learning). Both outperform a calibrated baseline using updraft helicity alone.
The architectures differ in useful ways. HGBT achieves better overall metrics but caps predictions at ~60% probability — it won't commit to certainty. U-Net extends to 100% and produces spatially smoother guidance. The gradient-boosted model is more reliable (calibrated) but less resolving (can't distinguish near-certain from merely likely). The deep learning model is more resolving but less reliable at the extremes.
The structural insight: the numerical weather model simulates the physics. The ML model translates the simulation into actionable probability. Neither alone is sufficient — the physics model can't produce calibrated probabilities, and the ML model can't simulate physics. The combination produces something neither can: physically grounded, probabilistically calibrated severe weather guidance at the timescale where decisions happen.