基于多模式集成技术的地面气温精细化预报  

Refined Forecasting of Surface Temperature Based on Multi-Model Ensemble Technology

作　　者：冯慧敏[1,2] 智协飞[1] 崔慧慧[2] 李荣[2] Feng Huimin Zhi Xiefei Cui Huihui Li Rong(Nanjing University of Information Science and Technology, Nanjing 210044, China Zhengzhou Meteorological Office,Zhengzhou 450007, China)

机构地区：[1]南京信息工程大学,南京210044 [2]郑州市气象局,郑州450007

出　　处：《气象与环境科学》2016年第4期73-79,共7页Meteorological and Environmental Sciences

基　　金：中国气象局华中区域气象中心科技发展基金项目(QY-Z-201302)资助

摘　　要：基于ECMWF、JMA、T639、WRF四个数值模式2012年6月1日—9月30日地面气温3—60 h预报资料和郑州加密自动站资料,利用多模式集合平均(EMN)、消除偏差集合平均(BREM)、加权消除偏差集合(WBREM)及多模式超级集合(SUP)4种方法,对2012年8月29日—9月27日郑州城区11个站点地面逐3 h气温进行多模式集成预报试验,采用绝对误差对预报结果进行检验评估,结果表明:在30天的预报期内,BREM、WBREM及SUP对于大多数站气温预报效果有明显改善,而EMN方案对11个站预报效果改善则不太明显;4种方案中,BREM和WBREM预报效果相对较好且稳定,各个站上3—60 h预报的绝对误差均在2℃附近或以下;SUP方案虽然对个别站预报误差较低,但是其预报效果并不稳定,一些站点的个别预报时效误差大于2℃。对于郑州观测站的气温预报而言,4种集成方案20时起报的气温误差明显小于08时起报的误差,并且20时起报的SUP集成方案绝对误差明显小于其他方案的绝对误差。总体而言,BREM、WBREM及SUP三种集成方案能够给郑州精细化预报业务提供较好的参考。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 30 th September 2012,the multi-model super ensemble forecasts of the 3 h surface air temperature of 11 stations in Zhengzhou from 29 th August to 27 th 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 than2 ℃ 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.

关 键 词：多模式集成 气温 精细化预报 

分 类 号：P457.3[天文地球—大气科学及气象学]