Department of Meteorology, School of Ocean and Earth Science and Technology, University of Hawaii at Manoa, Honolulu, Hawaii
Three 15-yr integrations were made with the ECHAM-4 atmospheric GCM (AGCM); in the first integration, the model lower boundary conditions were the observed monthly mean sea surface temperatures, and, in the second, the AGCM was coupled to the University of Hawaii 2.5-layer intermediate ocean model. In the third simulation, the SST climatology generated in the coupled run was used to create monthly mean SSTs, which were then used to drive the AGCM in an uncoupled configuration similar to the first run. The simulation of the intraseasonal oscillation (ISO) in these three runs was compared with data from the NCEP reanalysis and outgoing longwave radiation from NOAA polar-orbiting satellites, with particular emphasis on the boreal summer ISO.
The overall effect of coupling the AGCM to the ocean model is to improve the intraseasonal variability of the model. Upon coupling, the simulated boreal winter ISO becomes more spatially coherent and has a more realistic phase speed. In the May–June Asian monsoon season, the coupled run shows pronounced northward propagation of convection and circulation anomalies over the Indian Ocean, as in the observations, while northward propagation is absent in the uncoupled run. These improvements in the simulated ISO occur despite the fact that the coupled-run SST climatology has a substantial cold bias in both the Indian Ocean and the western Pacific warm pool. The improvement in the model ISO may be attributed to air–sea interaction whose mechanism is increased low-level convergence into the positive SST anomaly ahead of the convection anomaly.
The simulation of the August–October ISO is degraded upon coupling, however. The coupled-run basic state fails to produce the region of easterly vertical shear of the mean zonal wind, which lies on the equator during August–October. This region of easterly shear is critical for the emission of Rossby waves by equatorial convection associated with the ISO. In the absence of easterly shear, the observed northwestward propagation of convection is inhibited in both runs made using the coupled model basic state. The uncoupled AGCM run correctly locates the region of easterly shear and produces an August–October ISO that agrees well with observations.