Prediction and Source Contribution Analysis of PM2.5 Using a Combined FLEXPART Model and Bayesian Method over the Beijing-Tianjin-Hebei Region in China

Fine particulate matter (PM2.5) has a serious impact on human health. Forecasting PM2.5 levels and analyzing the pollution sources of are great significance. In this study, Lagrangian particle dispersion (LPD) model was developed by combining FLEXPART Bayesian inventory optimization method. The LPD capacity for real-time forecasting determination PM2.5, which refers to contribution ratio spatial distribution each type (industry, power, residential, transportation). we applied Beijing-Tianjin-Hebei (BTH) region optimize priori emission estimates during 15–20 March 2018. results show that (1) have certain degree overestimation compared with posteriori flux most areas BTH; (2) after optimization, correlation coefficient (R) between forecasted observed concentration increased an average approximately 10%, root mean square error (RMSE) decreased 30%, IOA (index agreement) index 16% at four observation sites (Aotizhongxin_Beijing, Beichenkejiyuanqu_Tianjin, Dahuoquan_Xintai, Renmingongyuan_Zhangjiakou); (3) main mainly originated from industrial residential emissions, while power factory transportation accounted only small proportion. forecasts in analysis can be used as corresponding reference information environmental governance protection public