基于遥感同化的PM2.5源清单优化方法研究

项目名称： 基于遥感同化的PM2.5源清单优化方法研究

项目编号： No.41301380

项目类型： 青年科学基金项目

立项/批准年度： 2014

项目学科： 天文学、地球科学

项目作者： 姚凌

作者单位： 中国科学院地理科学与资源研究所

项目金额： 25万元

中文摘要： PM2.5的人体健康效应已受到舆论的广泛关注，PM2.5源清单的准确性对于PM2.5模式预报结果的优劣起关键作用，如何获取高空间分辨率、精确的PM2.5源清单是当前研究的难点问题。目前基于直接调查、源反演模型的源清单订正方法都一定程度上受到地面监测资料空间分辨率较低的限制，本项目从利用遥感资料提高订正数据源空间分辨率的角度入手，研究：（1）基于LM-BP算法优化的人工神经网络模型反演PM2.5浓度，以之代替地面观测资料进行同化；（2）引入显式的PM2.5源清单订正因子，研究基于集合卡尔曼滤波（EnKF）的遥感反演与模式模拟浓度同化的源清单优化方法；（3）以京津冀地区为实验区域，对PM2.5源清单优化方法进行实证研究，提供可靠的不确定性范围。通过本研究方法优化的PM2.5源清单，可为PM2.5污染区域联防联控政策制订以及提高空气质量模式PM2.5预报能力提供可靠的支持。

中文关键词： 数据同化；模型-观测同化；机器学习；优化方法；

英文摘要： The effects of PM2.5 on human health have been widespread concern, the accuracy of PM2.5 source inventory plays a key role in the PM2.5 air quality model. How to obtain the accurate PM2.5 source inventory with high spatial resolution is a difficult problem of current research. Currently research on source inventory optimization based on direct investigations or source inversion model are somewhat limited by the lower spatial resolution of ground monitoring data. From the point of view of improving the spatial resolution of the revised data resoure with remote sensing data, this research focuses on three sections as follows.(1) PM2.5 concentration inversion based on LM -BP algorithm optimized artificial neural network model, used to assimilate instead of ground-based observations data; (2) The assimilation inversion of CAMx mode analog concentration and remote sensing data based on Ensemble Kalman Filter (EnKF) with the introduction of explicit revised factor of the PM2.5 source inventory;(3) Empirical research on the PM2.5 source inventory optimization method in the Beijing - Tianjin - Hebei region to provide reliable uncertainty range.The optimized PM2.5 source inventory obtained through this research can provide reliable support for not only the policy-making of regional PM2.5 pollution defense and joint cont

英文关键词： data assimilation；Model-observation assimilation；machine learning；optimization method；