Improving ENSO periodicity simulation by adjusting cumulus entrainment in BCC_CSMs

The simulation of El Niño-Southern Oscillation (ENSO) phenomenon is a challenging issue for coupled climate models. This study focuses on the ENSO periodicity simulated by Beijing Climate Center Climate System Models (BCC_CSM1.1 and BCC_CSM1.1m) which can reproduce reasonably well ENSO amplitude as observations. However, the major period of ENSO simulated by the BCC_CSMs is around 2.4 years, which is much shorter than that in observations. Compared with other 24 coupled models in Coupled Model Intercomparison Project Phase 5 (CMIP5), BCC_CSMs produce a very unrealistic ENSO peak period. Such a bias in simulating periodicity is suggested as a consequence of the severely underestimated air-sea coupling intensity in BCC_CSMs. To test this hypothesis, a quantitative method is developed to diagnose the linear ENSO frequency. As an effort to improve the ENSO simulation in BCC_CSMs, three experiments are performed with varying entrainment rates in the cumulus convection parameterization scheme of BCC_CSM1.1m. A more realistic ENSO period of about 3.3 years can be generated by the model with an inflated entrainment rate. When the cumulus entrainment is increased by 10%, the ENSO-related convective precipitation will enhance in the equatorial central to eastern Pacific. This anomalous convective heating induces an intensified surface westerly wind stress to the west of the anomalous convection center and as a result, the air-sea coupling intensity becomes larger, which contributes to a longer period of ENSO based on previous theories. In addition, the pronounced eastward extension of ENSO-related surface wind stress could also be the secondary factor to generate a lower frequency of ENSO in BCC_CSMs. Our study proposes a method to reduce the biases in ENSO periodicity simulation and puts more insights into the importance of adjusting atmospheric convection to reproduce ENSO properties in coupled model.