Air Quality Forecasting with Inversely Updated Emissions for China
Huangjian Wu, Lei Kong, Xiao Tang, Lili Zhu, Jiang Zhu, Zifa Wang
Abstract
Air quality forecasting relies heavily on accurate emission inventories. However, “bottom-up” inventories are often compiled for past years, leading to significant uncertainty in forecasting. In this study, we propose an operational air quality forecasting approach that incorporates online updates of emissions. Alongside the conventional 7 day forecasting, our approach conducts 24 h ensemble simulations for the ensemble Kalman filter to estimate emissions inversely. Specifically, it updates the emissions of CO, SO 2, NO x, volatile organic compounds, and other precursors of PM 2.5 and PM 10 by assimilating observations of CO, SO 2, NO 2, O 3, PM 2.5, and PM 10, respectively. Compared with the same ensemble forecasting for both emission estimations and forecasting, this approach reduces the computational cost by 84%, making it feasible for operational forecasting. We apply this approach to operational air quality forecasting in China. The results from January to February 2022 show that the root-mean-square errors are reduced by 7.1% from 40.7 to 37.8 μg m –3 for PM 2.5 . Similar reductions of 8.2–30.5% are found for CO, SO 2, NO 2, O 3, and PM 10 . Moreover, the updated emissions reveal the impact of emission control measures during the Beijing 2022 Winter Olympics, indicating reductions in NO x emissions of 53.5%, 42.7%, and 48.6% in Beijing, Zhangjiakou, and Hebei, respectively.