Observations of the extent of sea ice in the Greenland and
Labrador Seas for the 1953-1984 period led Mysak et al.
[1990] to conclude that a self-sustaining interdecadal climate
cycle exists in the Arctic. Mysak and Manak [1989]
observed a large positive 9-year sea ice anomaly in the Greenland
Sea, mirrored by a shorter 4-year positive anomaly in the
Labrador Sea. Tracking changes in the distribution of sea ice
with satellite imagery may be useful in the identification of
climate change [ Parkinson, 1992]. During the period
1968-1982 sea ice anomalies coincided with the passage of the ``Great
Salinity Anomaly'' (GSA) [ Mysak and Manak 1989]. The GSA
in the northern North Atlantic, first described by Dickson
et al. [1988], is an example of climate system variability
centered in a particular region, but affecting the climate of
distant regions. The GSA was a freshening of the North Atlantic
by 1%
(part per thousand) to a depth of 500 m
that impacted deep water formation, and thus ocean circulation
and global climate [ Dickson et al., 1988].
Sea surface temperature has also varied over decadal (and
other) timescales. Lazier [in press] noted that concurrent
with the GSA's change in salinity was a substantial change in sea
surface temperatures in the North Atlantic. Folland et al.
[1990] showed that the regional temperature decrease was not
reflected in the global ocean average, which was 
0.4
C
warmer in the latter half of the twentieth century. Their record
reveals approximately decadal ocean temperature oscillations
[ Folland et al., 1990]. Analysis of historical ocean
temperature data at a depth of 125 m in the North Atlantic by
Levitus et al. [1994] indicate the presence of a basin-wide
quasidecadal temperature oscillation between 1947 and 1990. A
decadal-scale pattern of temperature increase was also noted by
Parrilla et al. [1994], who compared ocean temperatures
from three surveys over the last 35 years. The authors note that
while the temperature increase is consistent with model
projections, the location of the increase (the interior ocean) is
not [ Parrilla et al., 1994].
The global thermohaline circulation (THC) of the ocean (a.k.a the ``conveyor belt''), is density-driven and can cause decadal variability in climate. A number of studies have shown internal oscillations in the strength of the THC at timescales of decades [ Weaver et al., 1991; Delworth et al., 1993; Stouffer et al., 1994]. Lynch-Stieglitz and Fairbanks [1994] used cadmium to calcium ratio and carbon isotope data to explore changes in oceanic circulation over time. The authors concluded that the two proxies, when used simultaneously, provide proof of a Pacific deep water source during glacial periods and circulation patterns that are different from those of the present. Delworth et al. [1993] showed variations in the strength of the THC at decadal timescales in coupled atmosphere-ocean general circulation model simulations. The variations were associated with salinity anomalies in the sinking region of the THC (North Atlantic) and resulted in decadal oscillations of sea surface temperatures (SST) which led to air temperature anomalies over the North Atlantic, northern Europe, and the Arctic. Patterns of SSTs simulated in Delworth et al.'s work [1993] are similar in some regards to observational results presented by Kushnir [1994]. Decadal variability in the THC has also been simulated by Weaver and Sarachik [1991], who attributed the oscillatory behavior to an advective phenomenon of the coupled atmosphere-ocean system. As is shown below (see ``Thresholds and nonlinearities'' section), the THC plays a major role in both potential changes of the state of the climate system, and in less drastic natural variability over decadal timescales.