Studying urban air-transport phenomena is highly complex, because of the heterogenous flow patterns that can arise. The main reason for these is the variable topology of urban areas, however, there is a large number of influencing variables such as meteorological conditions (e.g., wind situation, temperature) and anthropogenic factors such as traffic emissions. During a one-year CO2 measurement campaign in the city of Basel, Switzerland, steep CO2 gradients were measured around a large building. The concentration differences showed a strong dependency on the local flow regimes. Analysis of the field data alone did not provide a complete explanation for the mechanisms underlying the observed phenomena. The key numerical parameters were defined and the influence of turbulent kinetic energy dependency on the time interval for the Reynolds decomposition was studied. A Reynolds-Average Navier-Stokes Computational Fluid Dynamics (CFD) approach was applied in the study area and the CO2 concentrations were simulated for six significant meteorological situations and compared to the measured data. Two flow regimes dependent on the wind situation, which either enhanced or suppressed the concentration of CO2 in the street canyon, were identified. The enhancement of CO2 in the street canyon led to a large difference in CO2 concentration between the backyard- and street-sides of a building forming the one wall of the canyon. The specific characteristics of the flow patterns led to the identification of the processes determining the observed differences in CO2 concentrations. The combined analysis of measurement and modeling showed the importance of reliable field measurements and CFD simulations with a high spatial resolution to assess transport mechanisms in urban areas.