By using the hourly PM_2.5 surface observation data from 12 air quality monitoring stations in Beijing of 2014 and the Aerosol Optical Depth (AOD) products at 3 km, the spatial and temporal distribution characteristics of surface PM_2.5 and AOD of two satellites were analyzed, moreover, based on the spatial matching, the regression model of PM_2.5 versus AOD was established. Results show that the PM_2.5 concentration is higher in downtown region than that in suburban region while the lowest value appears in Dingling station. The daily variation of PM_2.5 concentration shows bimodal and unimodal pattern in urban areas and suburban areas, respectively. The spatial distribution characteristics of AOD are similar to those of PM_2.5 concentration, there is significant AOD gradient along the mountainous border. The AOD of Terra in the mornings are larger than those of Aqua in the afternoons in the urban, but in the suburbs, both of the AODs are essentially the same. The determination coefficient(R²) between AOD of Aqua and PM_2.5 concentration is 0.11 higher than that between AOD of Aqua and PM_2.5 concentration, and the average coefficients of determination (R²) between AOD of Aqua and Terra and PM_2.5 concentration in urban stations are higher than those in suburban stations. Comprehensively, the correlation coefficient of AOD of Aqua and PM_2.5 is the highest in urban areas, which means that Aqua satellite is more adaptive to monitor and inverse ground-level PM_2.5 in urban areas.