Development of 3D Variational Assimilation System Based on GRAPES-CUACE Adjoint Model (GRAPES-CUACE-3D-Var V1.0) and Its Application in Emission Inversion

Abstract. The adjoint method is known for its efficient calculation of sensitive information. After decades of development, assimilation technology based on adjoint method has gradually become an important tool for emission inversion. On the basis of GRAPES-CUACE aerosol adjoint model, and combined with the optimization algorithm and pollutants observations, the GRAPES-CUACE 3D variational (GRAPES-CUACE-3D-Var) assimilation system was further developed, and was used in the inversion of BC emissions in Beijing-Tianjin-Hebei region. The results show that the newly constructed GRAPES-CUACE-3D-Var assimilation system is reasonable and reliable, and can be applied to the emission inversion in Beijing-Tianjin-Hebei region. Compared to the simulations using the a priori BC emissions, the model simulations driven by the a posterior BC emissions in two inversion schemes are in better agreement with measurements. The correlation coefficient between the simulations and the observations is increased from 0.2 before the inversion to 0.7 and 0.64, respectively, and the NMSE is reduced from 0.38 to 0.22 and 0.24, respectively, and the NMB is decreased from 51.53 % to 43.37 % and 40.90 %, respectively, in the two inversion schemes. The spatial distributions of the a posterior BC emissions in the two inversion schemes are consistent with the distributions of the a priori BC emissions. The high-value areas are mainly located in the south of Beijing, Tianjin, central and southern Hebei, and northern Shandong. On the whole, the inversion scheme with a large observation ratio has better optimization effect. The observation information of the target time has a great influence on the a posterior BC emissions in a short period before the target time, and the influence decreases with the reverse time sequence.