Sensitivity of air-sea fluxes to SST perturbations

The sensitivity of long-term averaged air-sea fluxes calculated by a 3D atmospheric general circulation model to SST perturbations of an idealized spatial structure is investigated as a function of the SST perturbation amplitude, spatial scale, latitude, and season. This sensitivity is a dominant, yet largely unknown, parameter determining the stability and variability behavior of ocean-only model studies of decadal climate variability. The air-sea heat-flux anomaly induced by the SST perturbation varies linearly with respect to the SST perturbation amplitude in a wide range of perturbation amplitudes (+/-3 degrees C). The implied restoring time of the nonglobal SST perturbations by the air-sea fluxes is found to vary from 1.5 to 3 months (for a 50-m oceanic mixed layer and perturbation scale of less than 3.4 X 10(6) m) depending on the spatial scale, latitude, and season of the SST anomaly. The calculated restoring time for global perturbation is of the order of two years; and the latent heat flux is found to be the air-sea heat-flux component most sensitive to SST perturbations. Both longwave and shortwave radiative fluxes are much less sensitive than latent and sensible turbulent fluxes for nonglobal SST anomalies. The restoring time is found to be significantly longer for large-scale anomalies than for the small-scale ones, because of the dominant effect of the heat advection by wind over small perturbations. No significant difference is found between the restoring times for SST perturbations in midlatitudes and in the Tropics. SST perturbations are dissipated faster by the air-sea fluxes during the winter than during the summer. The air-sea freshwater flux anomaly is also found to strongly depend on the SST perturbation amplitude, and to vary almost linearly with the SST perturbation in the midlatitudes, but in a nonlinear way in the Tropics. The possible model dependence of the calculated restoring times is analyzed.