Abstract: A serious haze weather event from February 25 to March 12, 2013 was selected in this paper.Based on the environmental data from monitoring stations of Chengdu, meteorological data from weather stations, 0.75°×0.75° reanalysis data of ERA and wind profile radar data in Chengdu University of Information Technology, meteorological characteristics and effect of different meteorological conditions on pollutants diffusion in this process were analyzed.A forecast equation of PM_2.5 concentration was established.The results show that it is stable in upper level.Surface is controlled by the uniform pressure field or weak low pressure and no precipitation.These conditions are favorable to formation of pollution.However, there is low trough system in upper altitude.If low pressure enhances on surface, haze dissipation probability increases.Low wind speed and downdraft in vertical direction avail to product haze weather.Temperature inversion in boundary layer is another important feature.Strong ground temperature inversion exists from 0 m to 200 m, and its structure has inhibitory effect on convection current and in favor of accumulating vapor to form the mist, thus there are more chances to generate haze weather.When there is a haze weather event, the height of mixing layer is generally low.A PM_2.5 concentration model is established using a stepwise regression method, and its back inspection is done.Another haze weather event in spring in Chengdu is forecasted using this model.Pollution grade can be accurately forecasted, and its prediction results are relatively satisfactory.

Key words: Haze, Temperature inversion, Wind profile, Stepwise regression