The Linearity of the Atmospheric Response to Tropical Pacific Anomalous Forcing

Abstract

In this paper, the authors show that the effect of a tropical Pacific anomalous forcing can he primarily linear or nonlinear depending on its sign and longitudinal position. Using a nine-level steady-state model both the linear and nonlinear steady-state responses to tropical anomalous diabatic warming or cooling in the midtroposphere were computed. These sources were centered at 130°W and at the date line. At 130°W the atmospheric response to tropical heating or cooling is primarily linear. The amplitudes are small and of opposite sign for heating and cooling. The response agrees well with the results of corresponding general circulation model (GCM) experiments. For a heating at the date line, the modification of the linear response by the nonlinear terms is substantial. The nonlinear response to the heating is much stronger than the linear response, whereas the nonlinear response to cooling is weaker. The main effect of the nonlinear terms is to modify the amplitudes; the structure of the response is only slightly adjusted. Both the linear and the nonlinear steady-state responses to tropical beating at the dale line result in an anti-Pacific-North American pattern. This is not in agreement with the results of corresponding GCM experiments. The tropically forced stationary perturbations can have a substantial effect on the stability properties of the planetary-scale time-mean state. This may lead to strong nonlinear transient feedback effects and consequently a strong modification of the direct steady-state response. We have shown that the effect of a persistent heating or cooling at 130°W hardly affects the stability properties of the time-mean state. However, the effect of a heating at the date line is to strongly enhance the low-frequency variability. We hypothesize that this causes large additional transient feedback effects that substantially modify the character of the direct linear or nonlinear steady-state response. This may account for the discrepancy between the steady-state model and general circulation model results for the heating case at the date line.