Predictability from Weather to Intraseasonal Timescales Using Stochastic Ensembles

Stochastic ensembles of the ECMWF model are used for the MJO predictability studies and a high-resolution coupled atmosphere-ocean (WRF-HYCOM) model for TCs. To address the multiscale interactions and their impact on predictability of tropical cyclones (TCs) and the Madden-Julian Oscillation (MJO), we focus on scale-dependent error growth and its impact on predictability of TCs and the MJO. The first stochastic perturbation method is the stochastic kinetic-energy backscatter scheme (SKEBS), which allows perturbing individual fields with random patterns with prescribed temporal and spatial scales. By choosing the relevant spatial and temporal decorrelation times, we can mimic uncertainty in the location of convection or uncertainty in the large-scale background field. A second perturbation method is the stochastic perturbation of physical tendencies (SPPT). The main focus of these perturbations experiments is to better understand and quantify error growth and its impact on predictability. In particular, we conduct model experiments using the SKEBS perturbation generator to perturb the model on various spatial scales from synoptic, MJO, to global. These experiments will allow us to quantify the impact of scale-dependent error growth on predictability and prediction skill of MJO propagation through the Maritime Continent.