174
to go inversely with the length of the averages (see section 3) for intervals
greater than 150 days, indicating independent intervals, and therefore a
purely white spectrum, beyond this time scale. The spatial organization is
clear by examining Fig. 5. The variance of the long term averages becomes
spatially organized and, for the longest intervals (beyond 500 days), clearly
has organized itself around Greenland and the Labrador Sea looking very
much like the North Atlantic Oscillation with hints of the Pacific North
American pattern clearly visible.
3.4 Variability in Coupled Models
There have been, at this writing, only two long coupled simulations of long
term natural variability known to us, that of Schneider and Kinter (1994)
and that of Manabe and Stouffer (1995).
Schneider and Kinter compare a 400 year atmosphere-only run with
seasonal variation in solar forcing to a 400 year coupled run (without flux
correction). While the coupling increases the variability, it doesn't increase
it that much and these authors conclude that the basic mechanisms for
long term variability is the Hasselmann mechanism, with high frequency
forcing by weather disturbances in a way that the " .. .integrating effects of
components with long memories, i.e., soil moisture, snow cover and heat
storage by the ocean appear to be important in determining the spectral
characteristics of the variability."
The Manabe and Stouffer calculation has a flux adjustment in both
heat and momentum in order to maintain the thermohaline circulation (see
Manabe and Stouffer, 1988) and is the same calculation that shows Atlantic
variability identified as internal ocean variability (Delworth, Manabe, and
Stouffer, 1993, see section 5). The global decadal averages are roughly
consistent with random samples chosen at random: the 25 year samples
had a variance about 5 1 / 2 times that of the five year samples. They also
showed that over much of the globe, a mixed layer model of the ocean
gives much of the excess variability over that attained by fixing SST at
its climatological march. The overall conclusion we draw is that much
of the modeled variability is consistent with the Hasselmann mechanism
for climate and that the additional role of the ocean is confined to local
regions. It should be pointed out that the ENSO phenomenon is not well
represented by these coarse resolution atmospheres and therefore that its
variability and contribution to climate must be investigated by special
purpose models.
to go inversely with the length of the averages (see section 3) for intervals
greater than 150 days, indicating independent intervals, and therefore a
purely white spectrum, beyond this time scale. The spatial organization is
clear by examining Fig. 5. The variance of the long term averages becomes
spatially organized and, for the longest intervals (beyond 500 days), clearly
has organized itself around Greenland and the Labrador Sea looking very
much like the North Atlantic Oscillation with hints of the Pacific North
American pattern clearly visible.
3.4 Variability in Coupled Models
There have been, at this writing, only two long coupled simulations of long
term natural variability known to us, that of Schneider and Kinter (1994)
and that of Manabe and Stouffer (1995).
Schneider and Kinter compare a 400 year atmosphere-only run with
seasonal variation in solar forcing to a 400 year coupled run (without flux
correction). While the coupling increases the variability, it doesn't increase
it that much and these authors conclude that the basic mechanisms for
long term variability is the Hasselmann mechanism, with high frequency
forcing by weather disturbances in a way that the " .. .integrating effects of
components with long memories, i.e., soil moisture, snow cover and heat
storage by the ocean appear to be important in determining the spectral
characteristics of the variability."
The Manabe and Stouffer calculation has a flux adjustment in both
heat and momentum in order to maintain the thermohaline circulation (see
Manabe and Stouffer, 1988) and is the same calculation that shows Atlantic
variability identified as internal ocean variability (Delworth, Manabe, and
Stouffer, 1993, see section 5). The global decadal averages are roughly
consistent with random samples chosen at random: the 25 year samples
had a variance about 5 1 / 2 times that of the five year samples. They also
showed that over much of the globe, a mixed layer model of the ocean
gives much of the excess variability over that attained by fixing SST at
its climatological march. The overall conclusion we draw is that much
of the modeled variability is consistent with the Hasselmann mechanism
for climate and that the additional role of the ocean is confined to local
regions. It should be pointed out that the ENSO phenomenon is not well
represented by these coarse resolution atmospheres and therefore that its
variability and contribution to climate must be investigated by special
purpose models.