Refined Forecasting of Surface Temperature Based on Multi-Model Ensemble Technology
Based on the 3-60 h surface air temperature forecasting data of ECMWF,JMA,T639,WRF and observation data of Zhengzhou intensive automatic weather station from 1st June to 30th September 2012,the multi-model super ensemble forecasts of the 3 h surface air temperature of 11 stations in Zhengzhou from 29th August to 27th September 2012 was conducted by using four kinds of ensemble methods,including EMN,BREM,WBREM and SUP.The absolute error is used to evaluate the forecasting effect.Results showed that there is a significant improvement for most stations using BREM,WBREM and SUP,while the forecast effect of EMN for 11 stations is not obvious.Among the four methods,the forecast effect of BREM and WBREM is relatively better and more stable,since the absolute forecasting error of each station is around or below 2 ℃.Although forecasting error of SUP on individual station is smaller,the forecast effect is not stable,because there is significant error which is more than 2 ℃ for individual stations.For the forecasting effect of air temperature on Zhengzhou observation station,the error of 20:00 BT is less than 08:00 BT,the absolute error of SUP is significantly smaller than other methods which is forecasted from 20:00 BT.Overall,BREM,WBREM and SUP can provide better reference for the refined forecasting of Zhengzhou.
NETL
NSTL
万方数据
维普
