Zonally averaged composite anomalies of 700 mb height for years preceding and during four warm winters in the contiguous United States show generally positive anomalies over subtropical and middle latitudes and generally negative anomalies over high latitudes. The overall composite hemispheric height anomaly was positive for the years with warm winters. For four years with cold winters, generally inverse composite patterns are noted. In both composites, pattern reversals with respect to latitude occur during the following spring seasons. The four warm and four cold winters were those whose average temperature anomalies, nationwide, were the largest in the 33-yeu period 1948–80. The occurrence of consecutive severe winters within this relatively short record places limits on interpretation.
More generally, based on 33 years of data and Monte Carlo testing procedures, a statistical relation that is marginally significant at the 95% level appears to exist between the overall pattern of zonally averaged 700 mb height anomalies and mean winter temperatures in the United States. The relation is weaker and less coherent than that previously found with respect to mean summer temperatures. However, as with summer, the individual coefficients that are statistically significant are overwhelmingly positive in sign.
The overall correlation patterns with respect to spring and fall mean temperatures in the United States are obviously not significant. In terms of area-weighted percentages, the totals of individually significant coefficients for spring and fall are even less than to be expected from chance.