基于集合卡尔曼滤波的污染源反演方法研究

项目名称： 基于集合卡尔曼滤波的污染源反演方法研究

项目编号： No.41205091

项目类型： 青年科学基金项目

立项/批准年度： 2013

项目学科： 大气科学学科

项目作者： 唐晓

作者单位： 中国科学院大气物理研究所

项目金额： 25万元

中文摘要： 排放源数据的不确定性是目前大气污染数值模拟研究普遍面临的难题。污染源反演方法的发展为减小排放源不确定性提供了新的途径，但现有污染源反演方法基本都没有考虑模式误差，目前有研究发现忽略模式误差会导致反演出现排放源负值和20%以上的反演偏差。此外，现有反演研究主要依赖卫星资料，对地面观测资料的利用还远远不够。本研究将基于集合卡尔曼滤波算法（EnKF）发展一种合理考虑模式误差、有效同化地面观测资料的污染源反演新方法。以京津冀地区臭氧前体物NOx和VOCs排放源为反演优化对象，本研究首先通过不确定性分析方法对NOx和VOCs模拟的模式误差进行详细分析，并依此构建一阶线性马尔科夫误差演变模型将模式误差动态耦合到EnKF反演算法中，在此基础上发展能有效应用地面观测资料的长时间逐时反演方案。然后，开展京津冀地区2010年夏季NOx和VOCs排放源的理想和实际反演试验，对发展的污染源反演方法进行验

中文关键词： 污染源反演；不确定性分析；集合卡尔曼滤波；蒙特卡罗；资料同化

英文摘要： The uncertainty in emission inventory has been a great challenge for the numerical modeling research on atmospheric pollution. Emission inversion method provides the chances of solving this problem, but almost none of the related research so far has released its full potentials through appropriately addressing the issue of model error. It was suggested that negligence of model error could lead to inversion bias by over 20% or negative values for the emission. Besides, most of the previous emission inversion studies rely on satellite data without making enough use of surface observation data. In this study, we will develop a new emission inversion method based on ensemble Kalman filter algorithm (EnKF). Model errors will first be incorporated into EnKF properly, based on which a new long-time hourly inversion strategy will be developed with application of surface observation data in emission inversion. This new strategy will be applied to reduce the uncertainty in the two ozone precursors NOx and VOCs emission over Beijing-Tianjin-Hebei areas. We will conduct a detailed analysis of model error in NOx and VOCs simulation and then construct a first-order linear Markov error model to incorporate the model error into the EnKF algorithm. Then a long-time hourly assimilation strategy will be developed to apply NOx and

英文关键词： emission inversion；uncertainty analysis；ensemble Kalman filter；Monte Carlo；data assimilation