MJO Forecasting during DYNAMO Period in Three General Circulation Models

The Madden-Julian Oscillation (MJO) is the dominant mode of tropical convection variability on the intraseasonal time scale. The MJO initiates in the western equatorial Indian Ocean and propagates eastward as a couplet between multi-scale convection and large-scale circulation. Through upscale/downscale modulations and tropical-extratropical tele-connection, the MJO influences the weather activity and climate variability around the globe. The recurrent nature of the MJO with a period of 30-60 days offers an opportunity to bridge the gap between weather forecasting and seasonal prediction.

The predictability of the MJO is basically determined by three factors: i) the model fidelity in the representation of the MJO; ii) the settings of sea surface temperature; and iii) the accuracy of initial conditions. Due to significantly underestimated MJO intensity in old NCEP reanalysis I and II, the MJO predictability initialized with these two datasets is only about a week. When the MJO signal in these two reanalyses was recovered to be comparable with the observations, the MJO predictability in UH model, measured with Wheeler-Hendon index, reaches two weeks.

Initialized with much improved NCEP analysis, we carry out experimental MJO forecasting with UH coupled model weekly during DYNAMO field campaign period (Oct 2011-Mar 2012). We validate model forecasts of MJO initiation and propagation with latest CFSR and inter-compare with the results from the coupled CFSv2 and atmosphere-only GFS operational forecasts. Since atmospheric initial conditions for all three models are virtually the same, so the differences in the performance will be largely due to model differences (air-sea coupling and other physics). The successes and weaknesses of these models in forecasting MJO initiation and propagation will be reported.