Improved Skill of Northern Hemisphere Winter Surface Temperature Predictions Based on Land–Atmosphere Fall Anomalies

Abstract

A statistical forecast model, referred to as the snow-cast (sCast) model, has been developed using observed October mean snow cover and sea level pressure anomalies to predict upcoming winter land surface temperatures for the extratropical Northern Hemisphere. In operational forecasts since 1999, snow cover has been used for seven winters, and sea level pressure anomalies for three winters. Presented are skill scores for these seven real-time forecasts and also for 33 winter hindcasts (1972/73?2004/05). The model demonstrates positive skill over much of the eastern United States and northern Eurasia?regions that have eluded skillful predictions among the existing major seasonal forecast centers. Comparison with three leading dynamical forecast systems shows that the statistical model produces superior skill for the same regions. Despite the increasing complexity of the dynamical models, they continue to derive their forecast skill predominantly from tropical atmosphere?ocean coupling, in particular from ENSO. Therefore, in the Northern Hemisphere extratropics, away from the influence of ENSO, the sCast model is expected to outperform the dynamical models into the foreseeable future.