Climate Prediction: ENSO, MJO and Teleconnections
Abstract

Abstract Author: Yehui Chang, Siegfried Schubert, Max Suarez, and Duane Waliser

Abstract Title: Dynamical MJO Hindcast experiments: Sensitivity to Initial Conditions and Air-Sea Coupling

Abstract: The MJO plays a key role in the variability of the tropical atmosphere on sub-seasonal time scales. It is also a key source of untapped predictability in the tropics and subtropics. A series of 35-day hindcast experiments were conducted using both the coupled and uncoupled versions of the GEOS-5 model to study the forecast skill on sub-seasonal time scales. The hindcasts were started daily from initial conditions generated by a coupled and uncoupled “relay” approach (using an early MERRA reanalysis) targeting the prominent MJO events observed during 1979, 1996-1997 and 2002. The study examines the ability of the GEOS-5 to predict the MJO and evaluates the dependency of the MJO prediction skill on the ocean and atmosphere initial conditions. It is found that the hindcasts (more than 2000 cases) from both the coupled model and AGCM exhibit realistic MJO signals. This occurs despite the fact that the model, in free running mode, is deficient in reproducing the observed MJO-related precipitation coupling and related variability. There is substantial skill in forecasting MJO events that at times extends out to 2-3 weeks, although there is a clear dependence of skill on the initial phase of the MJO. The coupled model generally has higher forecast skill than the AGCM forced with observed SST (by about 3-4 days). This appears to be due to more realistic latent heat fluxes in the coupled model. However, the AGCM performs better during April-May. This may be in part the result of the apparent SST predictability barrier in the coupled model that occurs in boreal spring.