Roles of atmospheric physics and model resolution in the simulation of two types of El Niño

Abstract

Abstract In this study, we examined the possible impacts of atmospheric model resolution and physics on simulating two types of El Nino events. Four experiments were conducted using two different versions of the Community Atmosphere Model (i.e., CAM4 and CAM5) at two different resolutions. Compared with the CAM4 experiments, the CAM5 experiments yielded longer periods for both types of El Nino events, while the period yielded by the CAM5 experiments with a higher resolution was closest to the observations; the CAM5 experiments also had a larger anomaly amplitude of the thermocline depth during Eastern-Pacific (EP) El Nino events. All four coupled experiments could accurately simulate the phase locking of the Central-Pacific (CP) El Nino events but not the peak phase of the EP El Nino events; specifically, in the two CAM4 experiments, the peak phase appeared too early compared with the observations. Interestingly, the coupled experiment with a lower resolution or CAM5 could yield much stronger El Nino and larger El Nino-Southern Oscillation (ENSO) skewness than the experiment with a higher resolution or CAM4. The amplitude in the CAM4 experiment with a higher resolution was the lowest but was closest to that observed. Heat budget analysis of each of the four coupled experiments demonstrated that the thermocline feedback in the coupled model was much larger than that observed for the EP El Nino events. There existed a large deviation in the simulated zonal advection terms; specifically, the zonal advection term of the anomalous temperature gradient found by averaging the mean current over the Nino4 region was unrealistically negative during the developing phase of the CP El Nino events and directly affected the simulation of the CP El Nino events in the coupled experiments with a lower resolution.