Reply on RC3

Abstract

Emission inventories are essential for modeling studies and pollution control, but traditional emission inventories have large uncertainties and are often not real-time because they are highly human resource demanding to develop. In this study, a four-dimensional variational assimilation (4DVAR) system was developed to optimize sulfur dioxide (SO2) emissions by assimilating hourly SO2 concentrations. An observation system simulation experiment was conducted to evaluate the performance of the system. This evaluation indicates that the 4DVAR system can effectively reduce the uncertainty in SO2 emissions at a regional level. The 4DVAR system was then applied to optimize SO2 emissions during the early period of COVID-19 (from January 17 to February 6, 2020), and the reduction in SO2 emissions was assessed in comparison with the 2016 inventory. The hourly surface SO2 observations were assimilated. The results show that the emissions in 2020 decreased by 18.0 % compared with those in 2019, indicating a significant decrease between 2019 and 2020 due to the COVID-19 related lockdown. Three forecast experiments were conducted using emissions in 2016, 2019, and 2020 to demonstrate the effects of optimized emissions. The root mean square error in 2020 decreased by 47.9 % and the correlation coefficient increased by 300.0 % compared with 2016 emissions. This suggests that the 4DVAR system can effectively optimize emissions to describe the actual change in SO2 emissions during special events and improve the forecast skill.