Abstract
The observed atmospheric response to sea surface temperature (SST) anomalies associated with El Nino episodes is simulated with the use of a general circulation model (GCM) of the atmosphere. A series of experiments has been performed with the Rasmusson and Carpenter six-episode (El Nino) composite SST anomaly (SSTA) in the tropical Pacific superimposed upon the prescribed climatological SST (CSST). Five independent 400-day runs were made beginning in April of the El Nino year (Year 0) and ending in May of the year after the maximum SSTA in the tropical Pacific (Year +1). Each of the integrations started from different initial conditions selected from different years in a 20-year control run. The five-year El Nino integration was compared with the control run by analyzing the ensemble monthly, seasonal and annual mean statistics.The model's tropical response resembles the anomalous features found in previous studies. Some features are expected from the constraint imposed by the vorticity balance in the linear theory. For example, the upper level anticyclone pair and the lower level cyclonic circulation centers in the central Pacific are present for almost all months. These responses are significant and seem to be related to the location of both the maximum in CSST and the maximum in SSTA. An anomalous Walker Circulation is found in the vertical plane along the equatorial region. Rainfall departures from the long-term mean are associated with the anomalous Walker Circulation.An analysis of rainfall data over Northeast Brazil reveals sup- pressed rainfall in this area during the rainy season of Year(+1). Significant lower-than-normal rainfall is also found in the model's response during that period. This reduction in precipitation is associated with the downward branch of the anomalous Walker Circulation.The model's extratropical response is weaker than the tropical response and is not well organized. A Pacific/North American (PNA) pattern is present during the boreal winter, but its anomalous features are weaker than those simulated by Blackmon et al. (1983) (BGP), and the features are not all significant. The difference between our model response and BGP is attributed to different SSTAs employed as well as the inclusion of the annual cycle in the current experiment.A Pacific/South American pattern is indicated in the austral winter but the anomalous features associated with it are not all significant. There is large variability in the model's response on a month to month basis. This appears to be a result of the substantial variability in the anomalous forcing. (Abstract shortened with permission of author.)