334
transport parameterization. The limits, n = 0,00, correspond approximately to the
uncoupled cases of fixed surface heat flux and fixed surface temperature, respectively.
A power, m, of zero in the atmospheric moisture transport is equivalent to fixed
surface freshwater flux. Stronger dependence of moisture transport on temperature
gradient (increasing m) destabilises the THC. When moisture transport depends
weakly on temperature gradients (m ::; 1), the THC is more stable to perturbations
as n decreases (weaker restoring). For m 2 2, however, the THC is more unstable with smaller n, because the destabilising effect of anomalous moisture transport
outweighs the stabilising effect of anomalous thermal forcing of the THC. If zonal
mixing in the atmosphere is incomplete, the zonal mean temperature deviates from
the ocean temperature. Meridional atmospheric transports are less sensitive to variations in the ocean temperature gradient if the ocean area is small, and Newtonian
cooling is weaker. Simultaneously, a smaller ocean can compensate a given atmospheric energy budget imbalance only through a greater change in surface heat flux,
which translates into stronger Newtonian cooling. When flux adjustments are applied to obtain the correct model climate despite incorrect atmospheric transports,
the effect is that of choosing incorrect n or m, so transient behavior and model sensitivity are wrong although the mean state is correct. It is speculated that climate
drifts seen even in flux-adjusted coupled models are amplifications of residual drifts
in an incomplete, uncoupled spinup. The linearised model is non-normal, which
leads to amplifications of small perturbation through interference of non-orthogonal
eigenfunctions. Considerable excursions in temperature and salinity gradients can
occur, which are density compensated and hence are not effectively counteracted by
changes in the circulation unless the atmospheric transports respond to the changes
in ocean temperature.
1 Introd uction
This paper provides a conceptual framework to understand some feedbacks
involving meridional transports in the atmosphere and the thermohaline
circulation. Along the way, a strategy is presented that could be used to assess the strength of various feedbacks in a wide range of models. The analysis is motivated by the quest for a dynamical understanding of variations
in North Atlantic Deep Water (NADW) formation and Atlantic meridional overturning, which are both crucial for the northward heat transport
in the Atlantic (Hall and Bryden, 1982; Roemmich and Wunsch, 1985).
Paleoclimatic observations indicate that the strength of N ADW formation
may have varied significantly in the geologic past (see, for example, Boyle,
1990, or Sarnthein et at., 1994). Besides natural fluctuations in overturning
strength, anthropogenic ones may occur in the future: The coupled climate
model ofthe Geophysical Fluid Dynamics Laboratory (GFDL; Manabe and
Stouffer, 1994) shows a temporary reduction by one half of the Atlantic
thermohaline circulation, in response to a gradual doubling of atmospheric
Précédent

- 340/500

Suivant