最优化集成方法在城市臭氧数值预报中的应用研究  

Application of Optimal Consensus Forecast in Urban Ozone Prediction

作　　者：吴剑斌[1,2] 肖林鸿 晏平仲 李健军[3] 皮冬勤 陈焕盛[1,2] 赵江伟 王威[3] 

机构地区：[1]中国科学院大气物理研究所,北京100029 [2]中科三清科技有限公司,北京100029 [3]中国环境监测总站国家环境保护环境监测质量控制重点实验室,北京100012

出　　处：《中国环境监测》2017年第4期213-220,共8页Environmental Monitoring in China

基　　金：国家重点研发计划项目(2016YFC0208803);中国科学院战略性先导科技专项(XDB05030204);环保公益性行业科研专项(201509014);国家自然科学基金资助项目(41505105)

摘　　要：基于多模式(NAQPMS、CMAQ、CAMx、WRF-Chem)空气质量数值预报业务系统的滚动预报结果,结合站点观测资料,评估了最优化集成方法在城市臭氧数值预报中的可行性和预报效果。一年的评估结果表明:当训练期为15 d时,最优化集成方法能够得到相对较好的结果。总体而言,最优化集成方法对城市臭氧浓度变化趋势和浓度水平的预报效果明显优于单个模式,且在大部分城市优于多模式的最优预报,其预报值和观测的相关系数提高0.11以上,均方根误差降低约10μg/m^3;该方法对城市臭氧污染等级的预报能力也明显优于单个模式,特别是轻、中度污染。此外,在模拟偏差较大的城市,最优化集成方法对预报效果的改进更为显著;在模拟偏差较小的城市,该方法仍可进一步提升预报效果。An air quality operational forecasting system,consisting of NAQPMS,CMAQ,CAMx and WRF-Chem numerical models,provides real-time air pollutants predictions. In this study,the method of optimal consensus forecast（ OCF） is applied to assemble ozone prediction from these four air quality models. The performance of verification shows that OCF can get better results with the training period of 15 days. In general,the OCF is distinctly superior to individual model in predicting the variation and concentration level of ozone for most cities,with the correlation coefficient increased by over 0. 11 and the root mean square error decreased by nearly 10. In addition,the performance of the ozone prediction by OCF depends on the skill of each member of the ensemble. The improvement of ozone prediction by OCF could be more significant when the model deviation is larger in some cites,and become smaller under an ensemble of better model skills.

关 键 词：臭氧 数值预报 集合预报 最优化集成方法 

分 类 号：X84[环境科学与工程—环境工程]