Climate Prediction CenterSeasonal Forecast

Seasonal forecasts have a .5-month lead time

Score will be updated quarterly as a 4-year running mean (36 forecasts)

Previous slide

Next slide

Back to first slide

View graphic version

Seasonal Forecast scores for T and P (at 0.5 month lead time) are

averaged over 36 month periods to smooth over ENSO active/inactive periods.

The scores are also only for those areas for which we make forecasts. The

early peaks (96-99) in T & P are likely due to "chance," where we had some

good hits and forecast only over a small area, thus a "false" sense of

decline. Maybe we should normalize these for area forecast? Later peaks

(97-00) and the apparent valley at the right end (98-00) are due to El Niño

and La Niña success in consecutive winters (1997-98 and 1998-99) followed

by a less "classical" La Niña winter (1999-2000)and then an ENSO near

neutral year with even less skill 2000-01) In 2000-01 we had the forecast

concept (colder, snowier) correct, but the timing of cold events hurt

mathematical scores - no tools yet for intraseasonal timing.