The Stationary Wave Response to a Tropical SST Anomaly in an Idealized GCM

Abstract

The upper tropospheric stationary wave response to a tropical sea surface temperature (SST) anomaly is examined with an idealized general circulation model (GCM) as well as steady linear and nonlinear models. The control climate of the GCM is zonally symmetric; this symmetric climate is then perturbed by a dipolar SST anomaly centered at the equator. Two experiments, with anomaly amplitudes differing by a fact of two, have been conducted. The response is very linear in the amplitude of the SST anomaly. A steady, baroclinic model linearized about a zonally symmetric basic state simulates the GCM's stationary wave reasonably well when it is forced by anomalous heating as well as anomalous transients. When decomposing the GCMs flow into parts forced separately by heating and transients, tropical transients are found to play a dissipative role to first approximation, reducing the amplitude of the response to heating by a factor of two. The effects of extratropical transients are relatively weak. A steady nonlinear model is also used to evaluate the importance of transients and confirms the diagnosis based on the linear model. Part of the tropical transients seems to be forced by tropical convection and part by midlatitude disturbances propagating into the tropics. The anomalous extratropical transients include a part related to a shift in the model's storm track and a part related to barotropic instability of the stationary wave, but the effects of both of these changes are relatively weak due to the absence of strong extratropical climatic zonal asymmetries in the model. The dissipative role of transients in this model is contrasted with the positive feedback found by Held et al. (1989) in a GCM with realistic boundary conditions. The calculations in that paper are repeated, and the direct linear response to thermal forcing is found to be sensitive to the damping included in the model; but the positive feedback from the transients is robust to changes in the linear model. We speculate that a strong asymmetric storm track, with a well-defined barotropic decay region, is needed for the positive feedback to occur.