Preliminary Tests of Short-range Ensemble Forecast of Precipitation over Korea

Although numerical weather predictions have been improved dramatically, quantitative precipitation forecast (QPF) remains as one of the most difficult fields to predict accurately. Forecast errors arise from the uncertainties in both the initial conditions and model formulations. To compensate those uncertainties there have been many efforts toward ensemble forecast. The use of ensemble methodologies has resulted in the improved skill of medium-range weather prediction (e.g., Toth and Kalnay 1993; Molteni et al. 1996). Motivated by this success, application of ensemble technique to short-range forecast also has begun (e.g., Hamill and Colucci 1997; Stensrud et al. 1999) and it shows the ensemble average give a more accurate forecast than a single realization of the forecast model. Since the most single-model ensemble prediction systems consider uncertainties primarily in the initial conditions, those have the possibilities of biases present in the model. To account for other sources of uncertainty, Stensrud et al. (2000) considered model physics perturbations, and Ebert (2001) tested multi-model ensemble prediction using 7 different model forecasts.