El Niño-La Niña events simulated with Cane and Zebiak's model and observed with satellite or in situ data. Part II: Model forced with observations

Abstract

Various parameterizations derived from the observations are tested and proposed to improve the skill of the Cane and Zebiak's model in simulating the reality. The model is forced by FSU winds over the period 1980-1993. In particular, the way in which both the cold event and the easterlies in 1988 can be simulated better than with the parameterizations of the model is examined here. Introducing observed zonal current anomalies in the SST equation enhances the contrast between cold and warm anomalies in the eastern Pacific but does not induce a cold anomaly in 1988 in the central Pacific. This model deficiency is due to the parameterization of the temperature anomaly at 50 m, which precludes the temperature from being cold when the thermocline is shoaling. A new parameterization is determined based on the XBT subsurface temperature profiles. The SST simulated by the model is considerably improved with this new parameterization. When forced with observed SST, the atmospheric model simulates the observed easterlies in 1988. The model deficiency in simulating easterlies is therefore not due to the atmospheric component but to the oceanic one. However, the location of the simulated wind is always to the east of the observed ones because the atmospheric model simulates an internal heating term that is too tightly linked to the SST anomalies. The model performs better when it is forced by a combination of SST and cloud convection data. A new parameterization of anomalous convection heating is proposed based on convection data and introduced in the model to improve the wind simulation.