ENSO in GFDL's Newest Seasonal Forecast Model: Impacts of Ocean/Atmosphere Formulation and Resolution

We assess the simulation of El Niño/Southern Oscillation (ENSO) in FLOR, a coupled GCM developed by the NOAA Geophysical Fluid Dynamics Laboratory (GFDL), which now provides global multi-season forecasts to the public each month as part of the North American Multi-Model Ensemble (NMME). Compared to its lower-resolution predecessor (CM2.1), the FLOR model exhibits improved seasonal forecast skill and an ENSO simulation that is improved in some, but not all, aspects. Compared to CM2.1, the SST and rainfall anomalies in FLOR are shifted farther east during El Niño, improving ENSO's remote teleconnections over land. However, the FLOR ENSO is also even stronger than in CM2.1, with a shorter period and a more sharply-peaked spectrum. We trace the sources of these changes by examining an intermediary model (LOAR1), which is identical to FLOR except for a fourfold reduction in its atmospheric horizontal resolution to match that of CM2.1. We find that the simulated ENSO is as sensitive to the ocean/atmosphere formulation changes going from CM2.1 to LOAR1, as it is to the changes in atmospheric horizontal resolution change going from LOAR1 to FLOR. These model changes alter key ENSO feedbacks both above and below the ocean surface, in part through changes in the tropical Pacific climatology and its zonal and meridional asymmetries. Implications of these results for understanding, modeling, and forecasting ENSO are discussed.