No one disputes the European weather model (the ECMWF) is, on balance, superior to the American weather model, the GFS.  But the GFS can and does occasionally provide the better forecast.

Case in point: On Wednesday morning, the GFS model  forecast 3-6 inches of rain through the heart of the D.C. area through Thursday morning – coming close to nailing the record-setting amount which actually fell.

GFS model precipitation forecast Wednesday morning to Thursday morning (

The European model, on the other hand, forecast closer to 2-4 inches:

European model precipitation forecast Wednesday morning to Thursday morning (

For comparison, here’s the observed rainfall distribution, as computed by the National Weather Services (using multiple sources):

Computed rainfall Wednesday to Thursday (National Weather Service)

The color scales between the forecast and observed rain are different, so comparison isn’t easy.  But, generally speaking, the GFS model forecast offers a much better match with the observed rainfall.  The European model underestimated the amount of rain which would fall over the immediate D.C. area by roughly 1-2 inches.

While just one case for one location, this week’s rain forecast nicely demonstrates the European model isn’t always better than the GFS.  Considering how complex the weather set up was for this rain event, the GFS forecast was remarkably good (it was also better than the Canadian and NAM model forecasts).


Senator Mikulski touts National Weather Service computer upgrades

Game-changing improvements in the works for U.S. weather prediction