1999 Volume 77 Issue 5 Pages 1083-1108
The annual cycle and the large-scale low-frequency variability of surface winds over the tropical Pacific simulated by an atmospheric general circulation model (AGCM) are examined. The AGCM was integrated for 10 years with the observed sea surface temperatures (SSTs) from 1979 to 1988 and the model outputs were compared with National Center for Environmental Prediction (NCEP) reanalysis.
The AGCM successfully simulates the climatological features associated with the annual mean and annual cycle of the surface winds. The amplitudes of the simulated annual harmonics are nearly comparable with observations over the south Pacific convergence zone (SPCZ). On the interannual time scale, the AGCM realistically captures two warm events (1982-83 and 1986-87), and one cold event (1988). The first two empirical orthogonal functions (EOFs) and their principal components (PCs) of simulated wind anomalies reproduce the important features, which are crucial for coupled air-sea interaction, like westerly anomalies associated with warm events, eastward propagation of westerly anomalies during 1982-83 and realistic convergence pattern in the first EOF mode over tropical eastern Pacific. Also, the simulated wind anomalies do not exhibit eastward migration during 1986-87 and the strength of anomalies is much larger during 1982-83 than 1986-87. Thus, the AGCM succeeds in reproducing the distinct characteristics of different warm events. It is found that the PCs of wind anomalies are strongly correlated to the PCs of precipitation anomalies. Also, the correlation between PCs of observed and modeled precipitation anomalies is very high (0.93 for both PC1 and PC2) at zero lag. Therefore, it appears that the AGCM's capability to represent the large-scale low-frequency variability of convective heating realistically contributes to its success in simulating the large-scale low-frequency part of surface winds.
The most apparent deficiencies in the AGCM simulation include the incorrect location of westerly maximum, and less westward extension of westerly anomalies in EOF1. Also, the mean south-westerlies over Peruvian coast are weaker and not extended up to 170°W, especially during January. To get more insight into the AGCM simulated winds, the response of a steady-state linear Matsuno-Gill type model to two type of forcing (AGCM simulated precipitation anomalies, and AGCM simulated precipitation anomalies+SST anomalies) is compared with AGCM simulated winds. The linear model, forced only by the AGCM precipitation anomalies, can reproduce the AGCM simulated surface winds. The results of the linear model, when it is forced by both AGCM precipitation anomalies and SST anomalies, suggest that the deficiencies of AGCM in the western Pacific are mainly related to the insufficient treatment of surface turbulent fluxes.