4.4.2.3 Longer Records of Hemispheric Extent

The lack of comprehensive sea ice data prior to the satellite era hampers estimates of hemispheric-scale trends over longer time scales. Rayner et al. (2003) compiled a data set of sea ice extent for the 20th century from available sources and accounted for the inhomogeneity between them (Figure 4.10). There is a clear indication of sustained decline in arctic ice extent since about the early 1970s, particularly in summer. On a regional basis, portions of the North Atlantic have sufficient historical data, based largely on ship reports and coastal observations, to permit trend assessments over periods exceeding 100 years. Vinje (2001) compiled information from ship reports in the Nordic Seas to estimate April sea ice extent in this region for the period since about 1860. This time series is also shown in Figure 4.10 and indicates a generally continuous decline from the start of the record to the end. Ice extent data from Russian sources have recently been published (Polyakov et al., 2003), and cover essentially the entire 20th century for the Russian coastal seas (Kara, Laptev, East Siberian and Chukchi). These data, which exhibit large inter-decadal variability, show a declining trend since the 1960s until a reversal in the late 1990s. The Russian data indicate anomalously little ice during the 1940s and 1950s, whereas the Nordic Sea data indicate anomalously large extent at this time, showing the importance of regional variability.

Figure 4.10. Time series of NH sea ice extent for March and September from the Hadley Centre Sea Ice and Sea Surface Temperature (HadISST) data set (the blue and red curves, updated from Rayner et al., 2003), the April Nordic Sea ice extent (the black curve, redrafted from Vinje, 2001) and the August ice extent anomaly (computed relative to the mean of the entire period) in the Russian Arctic seas – Kara, Laptev, East Siberian and Chukchi (dotted green curve, redrafted from Polyakov et al., 2003). For the NH time series, the symbols indicate yearly values while the curves show the decadal variation (see Appendix 3.A).

Omstedt and Chen (2001) obtained a proxy record of the annual maximum extent of sea ice in the region of the Baltic Sea over the period 1720 to 1997. This record showed a substantial decline in sea ice that occurred around 1877, and greater variability in sea ice extent in the colder 1720 to 1877 period than in the warmer 1878 to 1997 period. Hill et al. (2002) have examined sea ice information for the Canadian maritime region and deduced that sea ice incursions occurred during the 19th century in the Grand Banks and surrounding areas that are now ice-free. Although there are problems with homogeneity of all these data (with quality declining further back in history), and with the disparity in spatial scales represented by each, they are all consistent in terms of the declining ice extent during the latter decades of the 20th century, with the decline beginning prior to the satellite era. Those data that extend far enough back in time imply, with high confidence, that sea ice was more extensive in the North Atlantic during the 19th century.

Continuous long-term data records for the Antarctic are lacking, as systematic information on the entire Southern Ocean ice cover became available only with the advent of routine microwave satellite reconnaissance in the early 1970s. Parkinson (1990) examined ice edge observations from four exploration voyages in the late 18th and early 19th centuries. Her analysis suggested that the summer antarctic sea ice was more extensive in the eastern Weddell Sea in 1772 and in the Amundsen Sea in 1839 than the present day range from satellite observations. However, many of the early observations are within the present range for the same time of year. An analysis of whaling records by de la Mare (1997) suggested a step decline in antarctic sea ice coverage of 25% (a 2.8° poleward shift in average ice edge latitude) between the mid-1950s and the early 1970s. A reanalysis by Ackley et al. (2003), which accounted for offsets between satellite-derived ice edge and whaling ship locations, challenged evidence of significant change in ice edge location. Curran et al. (2003) made use of a correlation between methanesulphonic acid concentration (a by-product of marine phytoplankton) in a near-coastal antarctic ice core and the regional sea ice extent in the sector from 80°E to 140°E to infer a quasi-decadal pattern of interannual variability in the ice extent in this region, along with a roughly 20% decline (approximately two degrees of latitude) since the 1950s.

In summary, the antarctic data provide evidence of a decline in sea ice extent in some regions, but there are insufficient data to draw firm conclusions about hemispheric changes prior to the satellite era.