A series of experiments were conducted with the Goddard general circulation model to determine the effect of variation in the location of Arctic sea ice boundaries on the model's mean monthly climatology. A control was defined as the mean of six January-February simulations with ice boundaries corresponding to climatologically minimum ice cover occurring simultaneously in the Davis Strait, East Greenland Sea, Barents Sea, Sea of Okhotsk and Bering Sea. An anomaly was similarly defined as the mean of two simulations with maximum ice conditions occurring simultaneously in the same regions. The extremes were estimated from 17 years of observed conditions in the Atlantic sector, and from five years of data in the Pacific sector.
When sea ice boundaries were at their maximum extent the following differences resulted in the January-February climatology as compared with minimum boundaries: Sea level pressure was higher by as much as 8 mb over the Barents Sea, by more than 4 mb in Davis Strait, and by slightly less than 4 mb in the Sea of Okhotsk. Pressure was lower by as much as 8 mb in the north Atlantic between Iceland and the British Isles, and in the Gulf of Alaska. Pressure rises of as much as 4 mb in the eastern subtropical regions of the North Atlantic and North Pacific accompanied pressure falls in the Gulf of Alaska and Icelandic region. Pressure also increased over the Mediterranean region. Geopotential heights at 700 mb were more than 80 gpm lower in the Gulf of Alaska, and more than 100 gpm lower in the Icelandic region. Changes of opposite sign occurred over the subtropics. Zonally averaged temperatures were cooler by 2°C below 800 mb between 50 and 70°N with little change elsewhere. The poleward flux of total energy was a maximum of 13% greater between latitudes 40 and 53°N.
The computed 700 mb geopotential differences were more than twice the inherent variability or "noise" of the model over a broad region of the North Pacific, and between Iceland and the British Isles. Pressure differences were more than twice the inherent variability in the Davis Strait, Gulf of Alaska and Barents Sea, and in the eastern subtropical Atlantic and Pacific. Statistical significance of zonally averaged differences was largest between 50 and 70°N where the confidence levels were as follows: for geopotential height differences, 99% (850 mb), 99% (700 mb) and 97% (500 mb); for temperature, 97% (850 mb) an 92% (700 mb); and sea level pressure, 94%. Confidence levels were high for changes in the Azores region.
On the basis of model results we conclude that ice margin anomalies are capable of altering local climates in certain regions of the high and mid-latitudes. Possible interactions between high latitudes and subtropical regions also are suggested.