CO2 and CH4 are the main greenhouse gases emitted by cities. Mobile observation methods with high precision and high spatio-temporal resolution are beneficial to understand the details of their distribution in cities, and the dynamic changes of sources. This work analyzes the existing mobile monitoring methods for atmospheric greenhouse gases. On this basis, a mobile observation system is built based on self-developed equipment off-axis integrated cavity spectrometer, a three-dimensional anemometer and vehicle differential GPS, and a matching greenhouse gas monitoring system is developed. The aerial observation data analysis software system has observed the CO2 and CH4 of typical roads in Hefei City, analyzed the hot spots of typical CH4 concentration, and observed CH4 in the landfills. The results show that the distribution of CO2 concentration in the first ring road of Hefei has a good correlation with the impact of urban non-point source emissions, while the distribution of CH4 has a poor correlation with it, and it is greatly affected by point sources. CO2 and CH4 concentration distribution in the second ring road is closely related to the distribution of surrounding forests, water sources and business circles. Generally, the average (median) concentrations (median) of CO2 and CH4 in the first and second rings in the morning and evening peak periods are higher than those in the idle time, and the concentrations in the first ring are higher than those in the second ring. Three-dimensional ultrasonic anemometers and GPS are used to calculate real-time natural wind speed and wind direction. The result showsthat the hot spots of CH4 concentration on the road is mainly from natural gas filling stations, biochemical pools, natural gas vehicles, etc. Among them, the correlation coefficient between CH4 and CO2 emitted by natural gas vehicles is about 70%. The emissions are large during idling, starting and slow driving. The high CH4 concentration captured in Feidong and Feixi, domestic waste landfills, is related to the incomplete sealing layer of the landfill and the unorganized release of the surrounding waste incineration power plant workshop. The Gaussian diffusion model estimated that the CH4 emission rate when the workshop door of the Feixi landfill was opened was an order of magnitude higher than when it closed. The fugitive emission of CH4 from Beicheng and Lujiang landfills was smaller than the first two. This study proves that the urban mobile observation system can provide a reference for establishing a comprehensive urban carbon emission monitoring system on the one hand and provide basic data for the study of urban greenhouse gas concentration characteristics on the other hand.