基于相似集合预报技术的臭氧预报释用研究  

Research on the interpretation and correction of numerical ozone forecast based on Analog Ensemble

作　　者：李梓铭 赵秀娟 孙兆彬 徐敬 张小玲[3] 邱雨露 尹晓梅 熊亚军 乔林 LI Zi-ming;ZHAO Xiu-juan;SUN Zhao-bin;XU Jing;ZHANG Xiao-ling;QIU Yu-lu;YIN Xiao-mei;XIONG Ya-jun;QIAO Lin(Institute of Urban Meteorology,China Meteorological Administration,Beijing 100089,China;Environmental Meteorology Forecast Center of Beijing-Tianjin-Hebei,Beijing 100089,China;School of Atmospheric Sciences,Chengdu University of Information Technology,Chengdu 610225,China)

机构地区：[1]北京城市气象研究院,北京100089 [2]京津冀环境气象预报预警中心,北京100089 [3]成都信息工程大学大气科学学院,四川成都610225

出　　处：《中国环境科学》2020年第2期475-484,共10页China Environmental Science

基　　金：北京市重大科技项目(Z181100005418014);科技部公益性行业(气象)专项(GYHY201206015);四川省科技计划项目(2018SZDZX0023,2018JY0011)。

摘　　要：利用相似集合预报技术(AnEn),采用2a的睿图-化学子系统(RMAPS-CHEM)历史预报结果和观测资料,对2018年6月1日~9月30日模式在京津冀地区13个城市共70个国控站点逐小时预报的O3浓度进行了释用订正研究,结果表明:随着集合成员数的增加,AnEn方法的预报效果呈现出先上升后下降的趋势,并且越过临界集合成员数后,预报效果逐渐降低,因此确定14为最优集合成员数.不同预报因子的权重敏感性不同,其中以RMAPS-CHEM本身预报的O3权重最高,其它依次为2m温度、2m相对湿度、10m风速和边界层高度.经过AnEn方法的释用订正,有效降低了O3浓度的预报误差.AnEn方法对O3浓度的时空变化趋势以及浓度值大小都有很好的订正效果,从所有站点的检验结果来看,AnEn方法订正后的O3浓度与观测值之间的相关系数较模式结果提升68.6%,均方根误差降低25%.AnEn方法对O3预报释用订正的效果存在明显的区域特征和日变化特征,白天时段对预报的提升主要集中在京津冀东部地区和大城市地区,夜间主要是在城市地区更加显著;AnEn方法夜间的订正效果优于白天,部分站点夜间提升效果(相关系数)超过250%.AnEn方法订正后的O3概率密度函数整体更接近实况,特别是在O3的低值区(35μg/m3以下)和高值区(200μg/m3以上)订正效果更佳.针对典型O3污染过程中北京、天津、石家庄3个城市的对比检验表明,AnEn方法在0~48h的预报时效内表现优于48~96h.不同城市体现出一定的区域差异,天津最优,北京和石家庄次之.AnEn方法对RMAPS-CHEM预报的O3浓度订正效果整体改善明显,可以在区域光化学污染数值预报工作中进行更加广泛的应用.A numerical correction of the O3 concentration forecasted by the Rapid refresh Multi-scale Analysis&Prediction System-CHEM(RMAPS-CHEM v1.0)was performed by using the Analog Ensemble(AnEn)method with 2a historical observations at 70 sites and forecast datasets in Beijing-Tianjin-Hebei region.The correction ability of the AnEn was evaluated against observed data from June to September 2018.The results showed that the performance of AnEn first increased and then decreased with the increase of ensemble member.The optimal performance was achieved when 14 members were used.The sensitivities of main forecast factors were different:the forecasted O3 from the RMAPS-CHEM showed the greatest weight,followed by temperature at 2m,relative humidity at 2m,wind speed at 10m,and planetary boundary layer height.The predicted O3 concentration from the RMAPS-CHEM was improved significantly using the AnEn method.The improvement was obvious for both the magnitude and spatiotemporal variation of predicted O3 concentrations.The correlation coefficients between observed and AnEn-predicted O3 concentrations were 68.6%larger than those between observed and RMAPS-CHEM-predicted concentrations.The root mean square errors after correction were 25%smaller.In addition,the performance of the AnEn method showed obvious spatial and diurnal variations in Beijing-Tianjin-Hebei region.In the daytime,the improvement was more evident in the eastern part of Beijing-Tianjin-Hebei region and in large cities,while at night,significant improvement mainly occured in urban areas.The performance of AnEn was generally better at night than in the daytime.Besides,the probability density distribution of O3 modified by the AnEn method is closer to the observation than that of the RMAPS-CHEM,especially in areas with low(<35μg/m3)and high(>200μg/m3)O3 concentrations.Evaluation at three cities during a typical O3 pollution event indicated that the performance of the AnEn method was better in the forecast period of 0~48h than in 48~96h.The AnEn method performed best

关 键 词：相似集合 RMAPS-CHEM 臭氧预报 释用 

分 类 号：X515[环境科学与工程—环境工程]