J. Climate, 20, 908-925.
College of Oceanic and Atmospheric Sciences, Oregon State University, Corvallis, OR
Adam H. Sobel
Department of Applied Physics and Applied Mathematics and Department of Earth and Environmental Sciences, Columbia University, New York, NY
Idealized experiments are conducted using a GCM coupled to a 20-meter slab
ocean model to examine the short term response to a localized positive equatorial SST anomaly, or "hot spot". A hot spot is imposed upon an aquaplanet with globally-uniform 28C SST, insolation, and trace gas concentrations designed to mimic tropical warm pool conditions. No boundary condition or external parameter other than the Coriolis parameter varies with latitude. A 15-member ensemble is initiated using random atmospheric initial conditions. A 2C equatorial warm anomaly is switched-on, along with ocean coupling (Day 0).
Enhanced deep convection rapidly develops near the hot spot, forcing an anomalous large-scale circulation that resembles the linear response of a dry atmosphere to a localized heating, as in the Gill model. Enhanced convection, the anomalous large-scale circulation, and enhanced wind speed peak in amplitude at about Day 15. Enhanced latent heat fluxes, driven primarily by an increase in vector mean wind, cause a loss of heat from the ocean in the vicinity of the hot spot before Day 20. Negative tropical-averaged wind speed anomalies occur between Day 20 and Day 50, driven by suppression of synoptic eddies. Suppressed latent heat fluxes due to suppressed eddy variance cause a warming of the remote tropics. The eddies are suppressed in association with anomalous low-level equatorial easterly flow. The wind-driven evaporative atmosphere-ocean exchange described above results in a 60-70 day oscillation in tropical mean (30N-30S) oceanic heat content, accompanied by a compensating out-of-phase oscillation in vertically-integrated atmospheric moist static energy. Beyond Day 70 of the simulation, positive SST anomalies are found across much of the tropical belt and slowly decay toward the 28C background state. This damping is contributed in large part by enhanced surface longwave emission.
Sensitivity experiments using 5 meter and 50 meter slab oceans demonstrate that the timescale of the transient oscillation in ocean heat content decreases with decreasing depth, as might be expected. A west Pacific hot spot experiment using realistic February SST, radiative forcing, and continents produces behavior qualitatively similar to that in the idealized experiments. In this case, however, a 15-member ensemble is not sufficient to produce a statistically significant oscillation in tropical ocean heat content.