Friday, December 26, 2014

Quick Look At How The 240-Hour Forecasts Did For Christmas




A number of posts back, I looked at 240-hour forecasts that were valid at 5 pm MST on Christmas day (yesterday). So here is a brief and simple evaluation of how the ECMWF and GFS did with those 10-day forecasts. I used the NAM analyses at 5 pm yesterday as the verifying charts and these are in the middle at both top (500 mb) and bottom (surface). 

The ECMWF was amazingly accurate wrt the position of the 500 mb trough over the west. That model was a bit too deep - forecasts heights over southern Utah were about 60 m too low. The GFS was substantially off in both the shape of the trough, its location, and its intensity - 500 mb heights over southern Utah verified 200m+ lower than the GFS forecast. The trough near the northeast was forecast about the same by both models, with the verifying feature being a bit deeper and located about midway between the two forecasts. So, for the West, the GFS was beaten badly by the ECMWF, as it often (usually?) is.

The same panels are shown below, but for the surface conditions. Interestingly, neither model did very well with the location or the intensity of the surface low that was over eastern Canada - no clear winning forecast there. But back over the West, the ECMWF forecast was again clearly more accurate than the GFS. 

When all is said and done, these 10-day forecasts are just unbelievably good from my dinosaur's perspective. When I began working for the US Weather Bureau (January 1967), the PE model (Primitive Equations) was experimental and not at all reliable or very useful at 72-hours!




No comments:

Post a Comment