Observation of Targeted Air Pollutants in Beijing–Tianjin–Hebei Region

This study explores the potential for improving forecasts of PM2.5 concentration through the observation of a pollution event that occurred from 14 to 18 November 2020, in Beijing–Tianjin–Hebei (BTH) region. The WRF (Weather Research and Forecasting) model and NAQPMS (Nested Air Quality Prediction Model System) are employed to simulate PM2.5 concentrations, and sensitivity experiments are conducted to identify key pollution species (i.e., sensitive species) and areas (i.e., sensitive areas) that significantly impact the PM2.5 forecasts. Additionally, Observing System Simulation Experiments (OSSEs) are carried out to assess the improvements in PM2.5 predictions resulting from reducing the initial uncertainties associated with the sensitive species identified in sensitive areas. The results indicate that errors in the initial organic carbon (OC) concentration make the largest contribution to forecast errors among all species, including SO₂, NO₂, CO, PM10, PM2.5, and black carbon (BC). Consequently, OC is identified as the target species to be observed during the pollution event. Furthermore, to conduct OSSEs, eight sub-regions are selected within the modeling domain according to predefined criteria. A comparison among these sub-regions shows that the Shandong Peninsula has the greatest influence on the PM2.5 levels of the entire region through its OC concentration, which makes the Shandong Peninsula a sensitive area for targeted observation. In subsequent experiments, additional OC observations are assimilated into the initial conditions for all eight sub-regions, and the resulting improvements in PM2.5 forecasts are evaluated. The results show that enhancing the accuracy of the initial OC field for the Shandong Peninsula significantly improves the PM2.5 forecasts for the entire BTH region. In conclusion, prioritizing OC observations in the Shandong Peninsula is essential for reducing PM2.5 forecast errors in the BTH region, and targeted observations can effectively enhance the accuracy of these forecasts.