Over the past few years, regional air pollution has frequently occurred in Mid-Eastern China, especially in Beijing. As the primary pollutants in urban air, atmospheric particulate matter (PM) not only leads to the decrease of atmospheric visibility, but also increases the mortality and morbidity of respiratory system diseases. In order to estimate the concentration, distribution and other properties of PM_2.5, a physical model is used. In building model, AOD at 550nm, the Effective Radius, ground-based PM_2.5 data, the relative humidity (RH) and boundary layer height (PBL) data are used, pollutant concentration and PM characteristics are taken into account comprehensively to build the retrieval model for near surface concentration of PM_2.5. The result shows that the correlation coefficient (R²) of retrieval and ground-based PM_2.5 is greater than 0.60, showing this method is possible for retrieving PM_2.5 in Beijing area.