Impact of tropical subseasonal SST variability on seasonal mean climate simulations

The purpose of this study is to examine the impact of subseasonal sea surfa ce temperature (SST) variability on the simulation of the seasonal mean ext ratropical circulation using a state-of-the-art high-resolution atmospheric general circulation model (AGCM). The format is a case study for January t hrough March 1989 (JFM89) and the primary emphasis is on regional scales ov er North America. The SST boundary conditions in the AGCM simulations were prescribed using observed weekly data. Experiments were made in which the w eek-to-week (subseasonal) SST variability was suppressed. In terms of the l argest spatial scales, the subseasonal SST variability has only a modest im pact; however, statistically significant modifications to the 500-mb height anomalies over North America were detected. Consistent with these changes in the height field, the seasonal mean North American rainfall anomalies we re particularly sensitive to the subseasonal SST variations, especially ove r the Pacific Northwest. Two possible mechanisms for this sensitivity were investigated with additio nal AGCM experiments and model diagnostics. The first mechanism, referred t o as a "stochastic'' effect, is defined by the hypothesis that the week-to- week SST variability only serves to enhance the amplitude of tropical preci pitation variability, which, in turn, modifies the midlatitude response. Wi th this stochastic effect, the details of the subseasonal SST evolution do not matter. In contrast, the second mechanism is a "deterministic'' effect in that the details of the evolution of the subseasonal SST matter. The exp eriments presented here indicate that the stochastic effect is small and th at the details of the subseasonal SST produce significant differences. This conclusion is supported by experiments with very large ensembles using a s omewhat lower-resolution AGCM and a nonlinear barotropic model. Finally, so me implications of these results for real-time forecasting are discussed.