Abstract: By analyzing the temporal and spatial characteristics of errors between the daily forecast results of WRF-Chem model and the actual observations in Tianjin from 2019 to 2021,the following conclusions are drawn：（1） The model forecast results have the best correlation with O₃ （0.90） and the worst correlation with PM₁₀ （0.59）. （2） The overestimation of pollutant concentration in the model prediction results is more significant,and the absolute error of PM₁₀,PM_2.5 and O₃ prediction is significantly higher than the other three pollutants. （3） The order of relative error magnitude of model prediction is O₃<CO<NO₂<SO₂<PM₁₀<PM_2.5,which is the highest in summer,approximate in autumn and winter,and the lowest in spring. The relative error in summer,autumn and winter is the largest in PM_2.5,while in spring it is the largest in PM₁₀. （4） From the spatial distribution characteristics of the relative error of the concentration prediction of the two main pollutants,PM_2.5 is relatively large in Jizhou,central urban area and Dagang;PM₁₀ is larger in Xiqing,Wuqing,Tanggu and Baodi in the central region. （5） The worst prediction effect of the model is PM₁₀,followed by PM.

Key words: WRF-Chem model, air pollution concentration, forecast error, Tianjin