was suggested by Lamb (1972), who argued that the exceptionally cold summer of 1968 in northeastern Europe and northwestern Asia was attributable in part to the previous winter's extensive snow cover. Because the snow did not retreat until June and July, the ground remained cold and wet, thereby favoring the persistence of a trough. The attendant cloudiness and frequent cyclones produced above-normal rain and occasional snowfall throughout the summer over the northern Soviet Union. Wahl (1968) suggested that persistent snow may have played a similar role in the unusually cold summers of the 1830s in northern North America.
Feedbacks involving snow cover have been suggested by Namias (1962, 1981), who has argued that extensive snow cover over eastern North America should enhance the coastal baroclinicity and favor strong cyclones along the East Coast. These systems, in turn, reinforce the cold trough and contribute to its persistence over periods of several weeks or longer, as in the winters of 1960-1961 and 1976-1977. Dickson and Namias (1976) have attempted to link extensive snow/extreme cold over eastern North America with anomalies in the winter climatic regime of Greenland and northwestern Europe. According to the authors' hypothesis, the southward displacement and intensification of cyclones on the East Coast of the United States lead to a reduction of cyclone activity over the Iceland-Greenland region. This weakening of the Icelandic low reduces the advection of mild air into the northeastern North Atlantic and northern Europe, thereby increasing the frequency of polar outbreaks over the European land areas. The frequent recurrence of this abnormal winter regime during the 1960s may have contributed to the relatively extensive European snow cover during that decade. However, the role of snow cover in modifying the atmospheric circulation in such regimes remains speculative because other hypotheses (involving sea surface temperature anomalies, tropical forcing, etc.) can be formulated. Controlled model experiments are needed to explore the physical basis of seasonal anomalies in these specific cases, as has been done for anomalies such as the North American drought of 1988 and the ENSO event of 1982-1983 (e.g., Trenberth and Branstator, 1992; Palmer and Mansfield, 1986).
Iwasaki (1991) has recently reported a tendency for large-scale anomalies of snow cover to persist from December through February, but he found no indication of anomaly persistence from February to March in the longer data record. Iwasaki also reported an apparent lag relationship, in which winters with extensive snow cover over eastern Eurasia tend to be followed by winters with extensive snow cover over North America. Although this lag relationship is statistically significant (r = 0.52), at least three counterexamples occurred during the 1980s.
As the preceding subsection indicates, various roles have been attributed to snow cover in the evolution and persistence of monthly and seasonal atmospheric anomalies. Most of these roles involve feedbacks, which are notoriously difficult to unravel from observational data. Consequently,