Evaluation of MJO Predictive Skill in Multiphysics and Multimodel Global Ensembles

Area of Science:

• Atmospheric science and climate modeling.
• Ocean-atmosphere interaction and teleconnections.

Background:

• The Madden-Julian Oscillation (MJO) is a dominant mode of subseasonal variability in the tropics, significantly impacting global weather patterns.
• Accurate MJO prediction is essential for subseasonal forecasting, but remains a challenge for current climate models.
• Ocean coupling is hypothesized to improve the simulation and prediction of MJO dynamics.

Purpose of the Study:

• To evaluate the performance of two coupled atmosphere-ocean models, FIM-iHYCOM and CFSv2, in hindcasting the MJO.
• To assess the impact of different deep convection parameterizations (Grell-Freitas vs. simplified Arakawa-Schubert) within FIM-iHYCOM.
• To investigate the added value of multiphysics and multimodel ensembles for MJO prediction.

Main Methods:

• Monthlong hindcasts of the MJO were generated using FIM-iHYCOM (with two convection schemes) and CFSv2 for the period 1999-2010.
• Hindcasts were initialized weekly with four time-lagged ensemble members.
• Forecast skill was evaluated using a variant of the Real-time Multivariate MJO (RMM) index and root-mean-square errors (RMSEs) of zonal winds.

Main Results:

• FIM-iHYCOM with the Grell-Freitas scheme (FIM-CGF) produced the most skillful MJO forecasts, extending to 19 days.
• CFSv2 and FIM-CGF showed similar RMSEs, and their multimodel ensemble mean extended skillful prediction to 21 days.
• An atmosphere-only FIM-CGF model lost skill after 11 days, underscoring the importance of ocean coupling for MJO prediction.

Conclusions:

• Coupled atmosphere-ocean models, particularly FIM-CGF, demonstrate significant skill in MJO hindcasting.
• Multiphysics/multimodel ensembles provide added value only when constituent models exhibit comparable skill and error characteristics.
• The choice of MJO index can influence the assessment of model skill and error metrics.