基于气流轨迹聚类的大气污染输送影响

基于中尺度气象预报模式(MM5)、混合单粒子拉格朗日积分(HYSPLIT)轨迹模式模拟和K均值聚类算法，利用气流轨迹聚类判断不同尺度大气输送型对城市空气的质量影响. 采用MM5模式对2006年珠江三角洲地区四季代表月(1，4，7，10月)的气象场进行了模拟，将模拟结果输入到HYSPLIT模式中，以计算广州市上空气团每日逐时的12 h后向轨迹；利用K均值聚类算法按轨迹移动方向和速度将各月气流轨迹线聚为有代表性的5类，计算各类输送型出现时段广州市ρ(PM₁₀₎和ρ(SO₂)的平均值. 结果表明，珠江三角洲地区低层大气输送季节性变化明显，按输送特征及其对城市污染物浓度的影响差异，可将输送型分为局地输送、城市间输送和远距离输送3类. 结合污染源排放清单得出的污染源空间分布，分析结果表明，广州市大气污染较重时段主要受特殊气象条件和珠江三角洲地区周边城市排放源的影响，本地源排放与周边城市污染物输送的叠加使大气污染加重. 

Impacts of Air Pollutant Transport Based on Air Trajectory Clustering

An air trajectory clustering methodology based on the MM5-HYSPLIT model and K-means clustering algorithm was developed to analyze the impact of different scale air pollutant transport on urban air quality.Meteorological fields of January,April,July and October in the Pearl River Delta(PRD) in 2006 were obtained by running the MM5 modeling system.We then calculated daily and hourly 12-hour backward trajectories in Guangzhou City utilizing the hourly MM5 outputs introduced to the HYSPLIT model during the peri...