Visualizing and verifying probabilistic forecasts of the Madden-Julian Oscillation

We describe a new approach for presenting probabilistic forecasts of the Madden-Julian Oscillation (MJO) based on the community standard Real-time Multivariate MJO (RMM) index, using forecasts from version 2 of the Predictive Ocean Atmosphere Model for Australia. This new display overcomes the difficulty of interpreting a dispersive ensemble plume and directly quantifies the probability for the MJO to occur in each of its eight RMM-defined phases as well as the weak phase. Beyond monitoring and interpreting predictions of the MJO, this new approach also provides a basis for forecast verification using probability-based skill scores. Here we present a clear and concise quantitative summary of this innovative method for accessing probability of the state of the MJO in an ensemble forecast. This new method compliments the traditional MJO ensemble forecast display and verification and will benefit global forecasting centers, international MJO working groups, and the World Meteorological Organization Subseasonal to Seasonal Project.