Verified blend accuracy
Does blending weather models beat the best one?
Measured against observed snowfall, SnowSure's blended forecast has 5.9% lower error than NAM — the single most accurate weather model — across 47,496 verified forecasts.
Ground truth: measured (SNOTEL + resort-reported), snow days only. Machine-readable: /api/v1/blend-accuracy.
Forecast Trust — Verified, not guessed — corpus scale, blend vs best model, and the full record.
Blend vs best single model, by horizon
| Horizon | Blend MAE | Best model | Best-model MAE | Improvement | Samples |
|---|---|---|---|---|---|
| Overall | 5.66 cm | NAM | 5.55 cm | -2% | n = 126,597 · as of 2026-10-18 |
| 1–3 days out | 5.22 cm | NAM | 5.55 cm | +5.9% | n = 47,496 · as of 2026-10-18 |
| 4–7 days out | 5.74 cm | NBM | 5.26 cm | -9.1% | n = 63,170 · as of 2026-10-18 |
| 8–14 days out | 6.68 cm | NBM | 4.44 cm | -50.5% | n = 15,931 · as of 2026-10-18 |
Lower MAE = closer to observed snowfall. Improvement = how much lower the blend's error is than the best single model. Updated 2026-10-19.
Every model, ranked by error
| Model | MAE (cm) | Samples |
|---|---|---|
| METNO | 4.1 | 12 |
| NAM | 5.55 | 2,072 |
| ECMWF | 5.61 | 7,844 |
| UKMO | 5.74 | 26,388 |
| JMA | 5.8 | 46,895 |
| Météo-France | 5.89 | 46,894 |
| ICON | 5.92 | 46,895 |
| GEM | 6.08 | 46,894 |
| NBM | 6.22 | 22,341 |
| HRRR | 6.23 | 25 |
| GFS | 6.28 | 46,897 |
| METEOSWISS | 7.1 | 5,514 |
| ICON2I | 7.28 | 3,453 |
See also: most accurate model by region · Forecast Trust.
