[A-AS03] Large-scale moisture and organized cloud systems

[AAS03-09] Subseasonal prediction of Madden-Julian Oscillation using machine learning

Takuya Jinno¹, Hiroaki Miura¹, *Kengo Nakai², Yoshitaka Saiki³, Saori Sakai¹, Tamaki Suematsu¹, Daisuke Takasuka⁴, Tsuyoshi Yoneda³ (1.The University of Tokyo, 2.Tokyo University of Marine Science and Technology, 3.Hitotsubashi University, 4.Japan Agency for Marine-Earth Science and Technology)

Keywords:Madden-Julian Oscillation, machine learning, Subseasonal prediction

By employing machine learning techniques, we construct a model for predicting the time series of the Realtime multivariate MJO index, which is a proxy that detects the MJO as an entity of active convection coupled to large-scale circulation at the intraseasonal time scale. The number of days for which the prediction is valid is measured for each of the 16 different phases of initial data in the MJO phase space. The model was evaluated to have a capability of predicting the RMM time-series for more than one month by applying a widely accepted criteria for the predictability of the MJO.