The Taiwan mesoscale ensemble prediction system (EPS) based on the Weather Research and Forecast model (WRF), also called WEPS, is designed to provide reliable ensemble forecasts for the East Asian region centered on Taiwan. The most skillful ensemble is obtained if model-uncertainty is represented in addition to initial condition uncertainty. A number of numerical prediction experiments are conducted to obtain the optimal configuration consisting of multiple physics-schemes, and two stochastic parameterization schemes: The Stochastic-Kinetic Energy Backscatter (SKEB) Scheme and the Stochastically Perturbed of Physics-Tendency (SPPT) scheme. The performance of the best configuration of WEPS is objectively verified against ECMWF reanalysis and a high-resolution simulation. Further analysis finds little impact of the stochastic schemes on quantitative precipitation forecasts and a single Typhoon case study. We conclude that for best performance over the East Asian domain, stochastic parameterizations alone are not sufficient to represent model-uncertainty and need to be augmented with a multi- physics suite.
