A New Technique for Improved MJO Prediction

Predicting the Madden-Julian Oscillation (MJO) is key to global prediction on subseasonal- to-seasonal (S2S) timescales. The Real-time Multivariate MJO (RMM) index is commonly used to measure MJO prediction skill and used as a predictor for predictions of other parameters over the globe. This index has proven to be very useful in providing information of the planetary-scale circulation pattern associated with the MJO and eastward propagation of the MJO in a statistical sense. But it is known to be ineffective in accurately identifying longitudinal locations of convection centers for individual MJO events. This shortcoming of the RMM index has hindered its applications in predicting remote influences of the MJO, which sensitively depend on longitudinal locations of MJO convection centers.

Recently, we have developed a new method that identifies individual MJO events by tracking eastward motion of large-scale precipitation anomalies along the equator. This method allows several key parameters to be quantitatively and accurately defined for individual MJO events. The parameters include the longitude and time of MJO initiation and termination, speed and range of MJO propagation, life span and mean strength of the MJO, and intervals of neighboring MJO events. Prediction of these parameters is important in capturing the tropical forcing of subtropical and extratropical circulations at intraseasonal timescales, but is difficult to derive from the RMM index. The new MJO tracking method has been used successfully in quantifying the barrier effect on MJO propagation by the Indo-Pacific Maritime Continent and interpreting
the issue of MJO simulations by global models. With suitable minor adjustment, this method can be applied to real-time MJO forecast and provide an alternative technique for monitoring MJO events and measuring their prediction skill.

The objective of this proposed project is to revise and test the new method of MJO tracking for real-time application in enhancing our ability to predict the MJO and its related extremes. We plan to take the existing MJO tracking method and develop it into one that can be used in real-time forecast. The proposed work includes:

(1) Adjust the existing MJO tracking method to make it applicable to real-time forecast.

(2) Apply the real-time tracking method to CFSv2 historical reforecast to develop a statistical
base of MJO forecast skill as a function of the season and the MJO (initiation location, etc.).

(3) Compare MJO prediction measurement based on this new tracking method and other EOF-
based methods (RMM index, OLR index) and seek possible ways to complement each other.

(4) Test the new MJO tracking method or tracking-EOF hybrid method in real-time environment.
These steps will be applied to CFSv2 forecast only. If successful, we plan to expand this work to
S2S Prediction products or NMME in a follow-up project.

Relevance to the Competition:
This proposed research will target solicited area (1) of the Climate Test Bed: Testing and demonstration of an experimental prediction methodology (e.g. new calibration or post-
processing techniques, verification techniques) or system (e.g., experimental multi-model combinations, hybrid statistical/dynamical systems, merging of systems across timescales to advance subseasonal prediction) developed in the broader community for operational purposes.