An evaluation of ENSO asymmetry in the community climate system models: A view from the subsurface

The asymmetry between El Niño and La Niña is a key aspect of ENSO that needs to be simulated well by models in order to fully capture the role of ENSO in the climate system. Here the asymmetry between the two phases of ENSO in five successive versions of the Community Climate System Model (CCSM1, CCSM2, CCSM3 at T42 resolution, CCSM3 at T85 resolution, and the latest CCSM3 + NR, with the Neale and Richter convection scheme) is evaluated. Different from the previous studies, not only is the surface signature of ENSO asymmetry examined, but so too is its subsurface signature. By comparing the differences among these models as well as the differences between the models and the observations, an understanding of the causes of the ENSO asymmetry is sought. An underestimate of the ENSO asymmetry is noted in all of the models, but the latest version with the Neale and Richter scheme (CCSM3 + NR) is getting closer to the observations than the earlier versions. The net surface heat flux is found to damp the asymmetry in the SST field in both the models and observations, but the damping effect in the models is weaker than that in the observations, thus excluding a role of the surface heat flux in contributing to the weaker asymmetry in the SST anomalies associated with ENSO. Examining the subsurface signatures of ENSO - the thermocline depth and the associated subsurface temperature for the western and eastern Pacific - reveals the same bias; that is, the asymmetry in the models is weaker than that in the observations. The analysis of the corresponding Atmospheric Model Intercomparison Project (AMIP) runs in conjunction with the coupled runs suggests that the weaker asymmetry in the subsurface signatures in the models is related to the lack of asymmetry in the zonal wind stress over the central Pacific, which in turn is due to a lack of sufficient asymmetry in deep convection (i.e., the nonlinear dependence of the deep convection on SST). In particular, the lack of a westward shift in the deep convection in the models in response to a cold phase SST anomaly is found as a common factor that is responsible for the weak asymmetry in the models. It is also suggested that a more eastward extension of the deep convection in response to a warm phase SST anomaly may also help to increase the asymmetry of ENSO. The better performance of CCSM3 + NR is apparently linked to an enhanced convection over the eastern Pacific during the warm phase of ENSO. Apparently, either a westward shift of deep convection in response to a cold phase SST anomaly or an increase of convection over the eastern Pacific in response to a warm phase SST anomaly leads to an increase in the asymmetry of zonal wind stress and therefore an increase in the asymmetry of subsurface signal, favoring an increase in ENSO asymmetry.