Abstract: The transport process of air pollutants in Hefei in 2014 was simulated, using the HYSPLIT(hybrid single-particle Lagrangian integrated trajectory)-4 backward trajectory model and NCEP(National Centers for Environmental Prediction) reanalysis data, and the mass concentration data of conventional atmospheric pollutants such as SO2, NO2, PM10 and PM2.5 were also used. The results showed that the air trajectories in Hefei had an obvious seasonal variation, and that the potential sources of pollutants in different seasons in Hefei were different.In January, the airflow in the northwest direction accounted for 43%, and the local airflow between cities accounted for 57%. The concentrations of SO2, NO2, PM10, and PM2.5 were very large. In April, the airflow from the Yangtze River Delta region accounted for 58%. The concentrations of various pollutants were very low. In July, the local airflow between cities accounted for 48%, and the airflow from the Yangtze River Delta region accounted for 24%. The PM10 concentration was higher, and the concentrations of other pollutants were lower. In October, the local airflow between cities accounted for 50%, the northern airflow accounted for 27%, and the airflow from the Yangtze River Delta region accounted for 23%. The PM10 concentration was higher, and the concentrations of other pollutants were lower. The main sources of air pollution in Hefei were mainly from the northern air masses, the air masses in the Yangtze River Delta region, and the local urban air masses.

Key words: air pollution, HYSPLIT-4 model, backward trajectory, cluster analysis