Application of the multi-model ensemble forecast in the QPF

Abstract

Based on the ensemble forecasting data of the China Meteorological Administration, European Centre for Medium Range Weather Forecasts, Japan Meteorological Agency, National Centers for Environmental Prediction, and the United Kingdom Met Office in the TIGGE Datasets, the multi-model ensemble forecasting techniques of the precipitation have been investigated. For the quantitative precipitation forecast(QPF) in the Northern Hemisphere, the bias-removed ensemble mean (BREM) forecast is more skillful and more stable than each individual model. In addition, the multi-model ensemble forecasts of the precipitation for the extreme events with freezing rain and snow over southern China in early 2008 have been conducted by using the bias-removed ensemble mean with running training period (R-BREM). For the light rain forecasts in central and southern China, the R-BREM technique has the largest TS scores for the 24h–192h forecasts except for the 72h forecast among all individual models and multi-model ensemble techniques. For the moderate rain, the forecast skill of the R-BREM technique is superior to those of individual models and multi-model ensemble mean.