MJO Predictability of Coupled Models and Seasonal Variation

The predictability of coupled models on MJO time scale is assessed by estimating the error growth rates of variables related to MJO activity. The retrospective forecasts by NCEP CFS and ECMWF model are analyzed to determine the error growth on intraseasonal time scale. The seasonal variation in the predictability is studied by examining the predictions made with initial conditions in different seasons. The error growths in predicting the precipitation, low-level and upper level winds in the Indo-Pacific region are estimated. These estimates are compared with earlier estimates of the predictions made by the earlier version of CFS. The MJO signal is extracted by data-adaptive methods and the predictability of the MJO signal and the total anomalies are compared.