74
The internal variability of the atmospheric circulation, with a characteristic time scale of days to weeks, constitutes an important source of
sampling fluctuations in monthly means, seasonal means, annual means,
and even in means over averaging intervals up to a few years. It is evident from Fig. 2 that these dynamically induced sampling fluctuations
are bound to be even larger (by a factor of 3 or more) in estimates of
hemispheric-mean temperature variations based on data from individual
stations or localized regions, obscuring even interdecadal scale features in
the hemispheric or global 'signal'.
Superimposed upon this background noise are a number of real climate
signals that impact hemispheric- and global-mean surface air temperature.
The ENSO cycle dominates the variability of tropical-mean surface air
temperature and SST and, for lack of a competing interannual extratropical
signal, its signature is also clearly evident in time series of hemisphericmean and global-mean temperature. It shows up more clearly in the time
series of global-mean SST than in the time series of global-mean surface air
temperature based on station data because the former is less subject to the
sampling fluctuations described above. ENSO-like interdecadal variability
is evident in SST and cold season circulation patterns over the Pacific sector
(Nitta and Yamada 1989, Trenberth 1990, Trenberth and Hurrell 1994,
Graham 1994, Zhang et al. 1996). The North Atlantic Oscillation, which
impacts wintertime climate over Europe and most of Russia also exhibits
what appears to be statistically significant interdecadal variability (Hurrell
1995). Yet another climate signal worthy of mention in this regard is the
dramatic rise in SST and surface air temperature in a belt stretching from
Labrador across the North Atlantic and the Greenland, Barents and Kara
Seas that took place during the decade of the 1920s, followed by a cooling
during the 19508 and '60s, as documented documented by Kelly (1982),
Ellett and Blindheim (1992), Mysak et al. (1990) and Deser and Blackmon
(1993), among others. This very strong regional feature accounts for a
substantial fraction of the differences between the Northern and Southern
Hemisphere surface air temperature time series.
During the 1980s and early '90s, the North Atlantic Oscillation and the
ENSO-like variability over the North Pacific conspired to make wintertime
surface air temperatures over the Northern Hemisphere continents (poleward of 40°) warmer, relative to the surrounding oceans, than they were
during previous decades. These persistent circulation-related anomalies
have contributed substantially to the recent upward trend in hemispheric-
The internal variability of the atmospheric circulation, with a characteristic time scale of days to weeks, constitutes an important source of
sampling fluctuations in monthly means, seasonal means, annual means,
and even in means over averaging intervals up to a few years. It is evident from Fig. 2 that these dynamically induced sampling fluctuations
are bound to be even larger (by a factor of 3 or more) in estimates of
hemispheric-mean temperature variations based on data from individual
stations or localized regions, obscuring even interdecadal scale features in
the hemispheric or global 'signal'.
Superimposed upon this background noise are a number of real climate
signals that impact hemispheric- and global-mean surface air temperature.
The ENSO cycle dominates the variability of tropical-mean surface air
temperature and SST and, for lack of a competing interannual extratropical
signal, its signature is also clearly evident in time series of hemisphericmean and global-mean temperature. It shows up more clearly in the time
series of global-mean SST than in the time series of global-mean surface air
temperature based on station data because the former is less subject to the
sampling fluctuations described above. ENSO-like interdecadal variability
is evident in SST and cold season circulation patterns over the Pacific sector
(Nitta and Yamada 1989, Trenberth 1990, Trenberth and Hurrell 1994,
Graham 1994, Zhang et al. 1996). The North Atlantic Oscillation, which
impacts wintertime climate over Europe and most of Russia also exhibits
what appears to be statistically significant interdecadal variability (Hurrell
1995). Yet another climate signal worthy of mention in this regard is the
dramatic rise in SST and surface air temperature in a belt stretching from
Labrador across the North Atlantic and the Greenland, Barents and Kara
Seas that took place during the decade of the 1920s, followed by a cooling
during the 19508 and '60s, as documented documented by Kelly (1982),
Ellett and Blindheim (1992), Mysak et al. (1990) and Deser and Blackmon
(1993), among others. This very strong regional feature accounts for a
substantial fraction of the differences between the Northern and Southern
Hemisphere surface air temperature time series.
During the 1980s and early '90s, the North Atlantic Oscillation and the
ENSO-like variability over the North Pacific conspired to make wintertime
surface air temperatures over the Northern Hemisphere continents (poleward of 40°) warmer, relative to the surrounding oceans, than they were
during previous decades. These persistent circulation-related anomalies
have contributed substantially to the recent upward trend in hemispheric-