We now examine how the decadal changes in the vicinity of Barrow and Eureka compare with those for the Arctic as a whole. Figure 3 shows the spatial pattern of decadal change for the 1990s (1989–98) minus the 1980s (1980–88) at two levels: 200 and 900 hPa. These fields are based on the TOVS Path-P temperature analyses with corrections, as discussed in the appendix. The months of January, March, April, and May are shown. January is used to illustrate the pattern of decadal change in the middle of winter; March–May are used to illustrate the evolution of these changes from the end of winter into spring.
FIG. 3. The decadal monthly mean temperature change between 1990s (1989–98) and 1980s (1980–88) at (top) 200 and (bottom) 900 hPa based on TOVS Path-P gridded satellite dataset. From left to right, the panels are for Jan, Mar, Apr, and May. Greenland is a data void.
For the month of January, the 1990s as compared with the 1980s were actually slightly warmer aloft, and cooler near the surface, for the Arctic as a whole (here considered the region poleward of 60°N) and Alaska in particular. It did warm at low levels over Scandinavia and the Barents Sea from the 1980s to the 1990s; this change is consistent with a shift to a systematically positive phase in the North Atlantic Oscillation (NAO). A positive sense to the NAO includes positive zonal wind anomalies across the North Atlantic, and hence the anomalous advection of relatively warm, maritime air masses across Scandinavia and the Greenland-Iceland-Norwegian (GIN) Sea (e.g., Hurrell 1995).
Much different results were found for the spring months. In March, the 1990s featured much colder (5 K) 200-hPa temperatures than the 1980s in a quasisymmetric pattern centered near the Pole. These cool conditions aloft in the 1990s persisted into April, but weakened, such that they were absent by May. These trends in springtime lower-stratosphere temperatures were followed by warm anomalies, for the most part, at 900 hPa. Here the most prominent decadal differences were in April, and in a less zonally symmetric pattern than aloft. Especially pronounced warming (3 K) occurred over Alaska, the Beaufort Sea, and northern Canada in the western Arctic, and the Kara Sea in the eastern Arctic. The decadal temperature changes at 900 hPa persisted into May. The decadal differences in temperature at 200 and 900 hPa (Fig. 3) over large areas of the Arctic are substantially greater than uncertainties in these temperatures, as gauged by differences between temperature fields calculated using different data sources and methods of analysis (Gaffen et al. 2000; Chelliah and Ropelewski 2000). Similar conclusions are valid if we had considered data from 50 hPa in place of 200 hPa.
Our maps for the temperature differences at 200 hPa between the 1990s and 1980s show that March had the clearest signature of the classic quasi-symmetric pattern associated with the AO. This result is consistent with the study of Pawson and Naujokat (1999), which found that the Arctic stratosphere was especially cold, and the polar vortex was especially strong, in the 1990s at the end of winter. The polar vortex has tended to persist longer into spring over the last decade (Waugh et al. 1999). March into April also seems to be the time of year in which lower-stratospheric conditions are most consistently related to those near the surface: temperature anomalies on monthly timescales in the troposphere tend to be out of phase with those in the lower stratosphere, with a crossover point near the 300-hPa level. The polar vortex weakens dramatically as spring progresses. By May the linkages between temperature anomalies in the lower stratosphere and troposphere are much less pronounced as the summertime circulation is usually in place in the stratosphere. The continuation of low-level warm anomalies into May in the 1990s might have been caused in part by reductions in snow and sea ice and hence surface albedo.
To compare the seasonality of the western Arctic with the AO index, we obtained the monthly AO values online (http://www.atmos.colostate.edu/ao/Data/ao_index.html). We then computed the monthly means for each decade as we did for Fig. 1; results are presented in Fig. 4. Similar to the temperature anomalies shown in Fig. 1, the AO index has a large positive anomaly in late winter during the 1990s compared to the previous four decades, with a peak in February and large values in January and March. The AO index models the general character of the polar vortex temperatures (Figs. 1a and 1c), although it peaks earlier.
FIG. 4. Decadal-averaged monthly AO index based on data provided by D. Thompson from his AO Web site (see text for address) for the 1960s–1990s.
It is interesting to examine how lower-stratospheric temperature anomalies in individual years compare with the decadal mean temperature signal presented above. Figure 5 shows the spatial distribution of temperature anomalies at 200 hPa for each March from 1990 to 1998. Anomalies are based on deviations from the 1980–98 mean formed from the corrected TOVS Path-P dataset. Three of the years, 1990, 1995, and 1997 featured zonally symmetric cold anomalies over virtually the entire Arctic. Three other years, 1993, 1994, and 1996 were also relatively cool near the Pole and for the Arctic as a whole, but in a less symmetric manner. The remaining three years, 1991, 1992, and 1998, were warmer, with decidedly asymmetric temperature anomaly patterns. Thus during the 1990s, the cold anomalies aloft, representing one of the principal signatures of the positive phase of the AO, were present strongly in 3 years, and modestly in 3 other years, of the 9-yr period. From a western Arctic perspective, substantial cold anomalies aloft occurred when the polar vortex was strongly enhanced, as in 1990, 1995, and 1997, and when the polar vortex was not as strong but displaced from the Pole to over the western Arctic, as in 1993.
FIG. 5. Mean temperature anomalies at 200 hPa in Mar for the 1990s, derived from gridded TOVS Path-P satellite temperature data.
Temperature anomaly fields at 200 hPa for March of 1999–2001 are shown in Fig. 6. These fields are based on the NCEP reanalysis using the mean for 1948–99 in calculating anomalies. The year 2000 featured continued cold conditions in the lower stratosphere as shown by Manney and Sabutis (2000). The years 1999 and 2001 had warm anomalies. In particular, the AO was in a significantly negative phase during the late winter and spring of 2001, which resulted in relatively large amplitude standing waves in the high-latitude flow in the Northern Hemisphere (not shown). It is unknown whether this represents a shift to a multiyear regime favoring a negative AO, or just a temporary hiatus from the largely positive AO state of the last decade or so.
FIG. 6. Mean temperature anomalies at 200 hPa in Mar for (left) 1999, (middle) 2000, and (right) 2001 based on the NCEP-NCAR reanalysis. The climatology is based on 1948–99, as in Figs. 1 and 2.
Return to previous section or go to next section
