What it is: The National Blend of Models (NBM) is NOAA/NWS’s statistically post-processed blend that combines dozens of deterministic and ensemble models into a single, calibrated forecast grid used operationally across the U.S.
Models included: Core inputs typically include GFS, GEFS, ECMWF (Euro), ECMWF ensembles, NAM (12 km & 3 km), HRRR, RAP, CMC/GEM, UKMET, and select regional ensembles (availability varies by cycle and variable).
Weighting approach: Dynamic, variable-specific weighting is applied—models are weighted differently by forecast hour, parameter (QPF, snow, wind, temp), season, region, and recent skill, rather than a fixed average.
Post-processing & bias correction: Raw model output is statistically bias-corrected, smoothed, and calibrated using historical verification, climatology, and consistency checks to reduce noise, outliers, and known model biases.
Operational output: Produces probabilistic and deterministic fields (means, percentiles, PoPs, snowfall ranges) and serves as the baseline guidance for NWS forecast grids, with human forecasters adjusting for mesoscale and event-specific factors.