In this paper, an automated algorithm, which is used to identify the spectral gap during the heavy haze pollution process, reconstruct acquired data, and obtain pure turbulence data, is developed. Comparisons of the reconstructed turbulent flux and eddy covariance (EC) flux show that there are overestimations regarding the exchange between the surface and the atmosphere during heavy haze pollution episodes. After reconstruction via the automated algorithm, pure turbulence data can be obtained. We introduce a definition to characterize the local intermittent strength of turbulence (LIST). The trend in the 15 LIST during pollution episodes shows that when pollution is more intense, the LIST is smaller, and intermittency is stronger; when pollution is weaker, the LIST is larger, and intermittency is weaker. During the heavy pollution process, air quality on an hourly scale is determined via turbulent motion. At the same time, the LIST at the city site is greater than that at the suburban site, which means that intermittency over the complex city area is weaker than that over the flat terrain area.Urbanization seems to reduce intermittency during heavy haze pollution episodes, which means that urbanization reduces the 20 degree of weakening in turbulent exchange during pollution episodes. This result is confirmed by comparing the average diurnal variations in turbulent fluxes at urban and suburban sites during polluted and clean periods. The sensible heat flux, latent heat flux, momentum flux, and turbulent kinetic energy (TKE) in urban and suburban areas are all affected when pollution occurs. Material and energy exchanges between the surface and the atmosphere are inhibited. Moreover, the impact of the pollution process on suburban areas is much greater than that on urban areas. Urbanization seems to help reduce the 25 consequences of pollution.