JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 92, NO. C10, PAGES 10,835-10,843, SEPTEMBER 15, 1987 The Interpretation of the Icelandic Sea Ice Record P.M. KELLY, C. M. GOODESS, AND B. S. G. CHERRY Climatic Research Unit, School of Environmental Sciences, University of East Anglia, Norwich, England,United Kingdom The representativeness of the Icelandic sea ice record, one of the longestclimate indicatorsfor the North Atlantic sector,is assessed by statisticalanalysis. Correlation of the sea ice index with ice con- centrationdata for the adjacentseas suggests that the index is a reasonable measure of the ice cover of a large area of the northern North Atlantic sector. Associations with surface air temperature and sealevel pressure are strongbut regional rather than hemispheric in scale. The immediatecauses of the variable advance of ice onto the Icelandic coastsare clear in the analysisof the atmosphericcirculation data. Severe ice conditions are associatedwith enhanced anticyclonicity in the Icelandic region and stronger northerly and westerly winds over the GreenlandSea.It is concluded that the Icelandicindex is a useful indicator of the stateof the atmospheric flow north of Iceland and of the characterof the East Greenland Current, one of the major avenues of meridional heat exchange in the climate system. INTRODUCTION Iceland is situated at the margin of the seasonal ice zone in the path of the main outflow of ice from the Arctic basin in the East Greenland Current (Figure 1). The nature and quan- tity of ice in the East Greenland Current varies from year to year, as doesits extension as the East Icelandic Current north and east of Iceland [Koch, 1945; Wadhams, 1986]. It is the East Icelandic Current which frequently brings ice onto the shores of Iceland. The presenceor absenceof ice has always had significant implications for the Icelandic nation [Ogilvie, 1981]. Because of this vulnerability, many observations of unusual sea ice conditions and climatic events exist extending back to the colonization of the country some 1000 years ago. The record, basedon a variety of documentarysources, beginsin the ninth century A.D., and data are available for almost every year since 1600. The various forms of historical data were collated by Thoroddsen [1916/17], and the sea ice record was later updated and presented in index form by Koch [1945]. It is now maintained by the Icelandic Meteorological Office. Sigtryggsson [1972] and Ogilvie [1981, 1983] have critically examined the reliability of the historical data, and Ogilvie has presented series of revised indices for the period up to the 1780s. The Icelandic data set must be considered one of the most valuable indicators available to those concerned with the his- tory of the Earth's climate. Bergthorsson [1969] has used the record as an index of local air temperature fluctuations over the past 1000 years, and other investigators have used it as a measure of the state of the climate system further afield (see, for example, Bryson [1974], Lamb and Morth [1978], and Bryson and Swain [1981]). The location of Iceland beneath one of the major atmo- sphericcentersof action, the Iceland Low, in the zone affected by one of the hemisphere's major wind systems,the North Atlantic westerlies, meansthat the record may be a measureof the large-scale state of the atmospheric circulation. The Ice- landic record may also be a usefulindicator of the state of the oceanic componentof the climate system. The Icelandic region Copyright 1987by the American Geophysical Union. Paper number 7C0510. 0148-0227/87/007C-0510505.00 is one of major oceanographic significance, with the cold East Greenland Current flowing out of the Arctic adjacent to northward-flowing, warm Atlantic water. These currents transfersubstantial quantitiesof heat meridionally and play an important role in the Earth'senergy balance. The Green- land Sea is a major area of bottom water formation(see, for example, Swift [1986]) and the presence of extended ice cover generallyindicates stable stratification in the upper ocean layers and limitedbottom water formation in the region[Ma- Imberg, 1972]. It is in this general area that recenttrends in Arctic sea ice extent have been greatest [Walsh and Johnson, 1979a; Sear, 1984]. Finally, the Greenland Seais an area in which interaction between the ocean, ice cover and atmosphere [Walsh, 1983; Barry, 1985] may be an importantfactor. As the ice margin advances and retreats, the changing characteristics of the ocean-atmosphere boundary can produce major variations in energy transferbetweenthe oceans and atmosphere. As a consequence, the effects on air temperature of increasing at- mospheric levelsof carbon dioxide and other trace gases are believed likely to be at a maximumclose to the ice boundary [Dickinson, 1986]. For this reason, long cryospheric records representative of conditions over largeareas may be of value in detecting anthropogenic influences on the climate system [Barry, 1985]. While it can be argued on a priori groundsthat the Ice- landic record shouldprovide a valuableindex of the state of the climatesystem, there have beenfew empirical analyses to support this case. Although therehavebeena number of case studiesof severeice events,interpretation of the record has tended to be based on assumptions about representativeness and physical significance rather than on rigorous statistical analysis. To be of value, the representativeness of an indexin both space and time mustbe carefully assessed and its physi- cal significance hasto be determined. This can be achieved in a number of ways: through case studiesbased on detailed observational data; by theoretical argument and numerical modeling;or by statistical analysis. Each method has advan- tagesand limitations.In this study, we use statistical tech- niques to assess the representativeness of the Icelandicrecord and compare our resultswith the findingsof more detailed case studies of individual extreme events in order to confirm the causes of the variable approachof the ice toward Iceland. 10,835