Air quality models, such as the Community Multiscale Air Quality (CMAQ), Weather Research and Forecasting-Chemistry (WRF-chem), the Comprehensive Air Quality Model with Extensions (CAMx), and the Nested Air Quality Prediction Modeling System (NAQPMS) models, are effective tools for atmospheric pollution analysis. They are widely used in the prediction and analysis of the temporal and spatial distribution of atmospheric pollutants. However, there are uncertainties in terms of emission sources, initial conditions (ICs), boundary conditions (BCs), and chemical processes that affect the accuracy of these models (Bouttier & Courtier, 1999;Hanna et al., 1998;Moore & Londergan, 2001). In particular, the ICs input the atmospheric state at a specified time into the model, playing an important role in air quality prediction. Data assimilation (DA), which can incorporate observations into the model, has been shown to be an effective method to improve the ICs.Since the assimilation of the initial field is significant for building an accurate analysis field, which is crucial for improving the short-term forecast and analyzing the vertical diffusion and extinction of aerosol, many studies and multiple DA methods have been conducted and have obtained significant assimilation effects (