U.S. Dept. of Commerce / NOAA / OAR / PMEL / Publications
Climate changes are being experienced in the Arctic. These changes pose challenges to the resilience of Arctic life including humans (Symon et al. 2005; Overland and Wang 2005). Considering that the Arctic domain is a relatively small fraction of the earth, it is likely that larger anomalies of temperature and other variables would occur in the Arctic-wide mean compared to the global mean. On the other hand, large multiyear anomalies are also possible based on internal positive feedbacks in the Arctic associated with sea ice, ocean, and land processes. Due to the lack of pan-Arctic observations in the past and the complexity of processes involved, coupled atmosphere–ocean general circulation models (AOGCMs) are tools for studying Arctic climate and its response to changing external forcing.
To assess whether recent changes in the arctic climate are outside the range of natural variability, it is helpful to compare observations from recent decades with those from earlier in the twentieth century. Surface air temperature (SAT) is related to regional energy budgets and is a robust climate parameter, in the sense that it can show large-scale anomaly patterns and is generally well observed (Lambert and Boer 2001). Previous studies show that warming during the last two decades exhibits the greatest trends in the high latitudes of the Northern Hemisphere (Hansen et al. 1999; Jones et al. 1999), and are attributed to anthropogenic forcing changes (Houghton et al. 2001; Broccoli et al. 2003). The two 20-yr periods of largest temperature anomalies in the Arctic for the twentieth century are 1925–44 (midcentury), and 1979 to present.
Although the warm anomalies in midcentury are recently receiving attention from the polar research community, the mechanisms behind them are still in debate (Polyakov et al. 2002; Overland et al. 2004). Bengtsson et al. (2004) suggest that the existence of the multiyear, midcentury warm anomalies are associated with considerable internal variations over several years initiated by the stochastic variations of the high-latitude atmospheric circulation and subsequently enhanced and maintained by sea ice feedbacks, particularly, over the Barents Sea. Overland et al. (2004) supports natural variability as the source of these warm anomalies, based on regional and temporal variability in the observed atmospheric circulation. Delworth and Knutson (2000) also propose that this warming was a manifestation of internal variability. By comparing index trends in observations and model simulations, Karoly et al. (2003) conclude that the observed warming from 1900 to 1949 over North America was likely due to natural climate variation, whereas the trend from 1950 to 1999 was consistent with simulations that include anthropogenic forcing from increasing atmospheric greenhouse gases and sulfate aerosols.
However, closure has not been reached on the ultimate cause of the midcentury warm event. The length and spatial coverage of the meteorological records in the Arctic, particularly upper air, do not allow a full analysis of the dynamics of the early/midtwentieth-century multiyear event, whereas results from AOGCMs provide useful information in studying the causes of these anomalies. In addition to model evaluation, we investigate the hypothesis that the midcentury multiyear warm event is based on intrinsic atmospheric variability amplified by internal Arctic feedback processes. Thus, we do not require the models to have a year-to-year correspondence to data, but the models should be able to replicate similar multiyear events.
A suite of scenario simulations were conducted by coupled AOGCM from worldwide sources for the Intergovernmental Panel on Climate Change (IPCC) Fourth Assessment Report (AR4). Section 2 summarizes these models and their ensemble runs together with the observed datasets used in the present study. Section 3 analyzes the simulation results for the Arctic in recent decades, and in the early part of the twentieth century with emphasis on the midcentury warm anomalies based on two scenarios: the twentieth-century climate in coupled models (20C3M) and the corresponding control runs (PIcntrl). Spatial distributions of temperature anomalies from a subgroup of models are discussed in section 4, followed by the conclusions.
Return to Abstract or go to next section