Changes in tropical sea surface temperature (SST) can produce changes in cloudiness that modify incoming solar shortwave (SW) radiation, which in turn affects SST. The effects of this negative feedback on Pacific interannual variability are examined in forced ocean model and hybrid coupled ocean-atmosphere model simulations. Two empirical schemes are used to model the large-scale, low-frequency response of surface SW to SST anomalies. The first scheme attempts to account for the nonlocal nature of the atmospheric response to SST based on patterns of covariability analyzed through singular value decomposition. Ion the observations the primary coupled mode of variability is composed of a SW anomaly in the central Pacific that covaries with anomalous SST in the eastern Pacific. This is applied in the model as a nonlocal SW feedback. The second scheme examines the effects of a purely local feeback with a spatially varying coefficient of magnitude chosen similiar to the first scheme. In almost all cases the second scheme behaved similarly to the first, presumably because the correlation scale of SST is large enough for El Ni¿o--Southern Oscillation (ENSO) dynamics that there is little sensitivity to the local approximation in the SW feeback. In simulations forced by time series of observed wind stress the SW feedback induced very minor SST damping.

Results from a simplified heat budget analysis showed that while the SW feedback increased the local heat flux damping on SST, it also induced a mean shallowing of the mixed layer. The resulting changes in both the local mean vertical temperature gradient and the zonal velocity response to the wind stress acted to oppose the local heat flux damping effects. When observed SW anomalies were modified as expected, and agreement with observed SST improved. In coupled simulations the SW feedbacks had greater impact than in the case of specified stress. The main effects were to decrease the magnitude of the warm and cold SST anomalies in central Pacific, while leaving the pattern and evolution of ENSO anomalies essentially unchanged elsewhere. The SW feedbacks thus produce a modest improvement of the model ENSO SST pattern compared with observations, although they tended to shorten the period of the model ENSO cycle. Overall the results suggest that large-scale SW feedbacks are of quantitative importance to simulating some aspects of the ENSO cycle but are not critical to the overall occurrence of the phenomenon. ¿American Geophysical Union 1994