Identifying and optimizing the key parameterization processes and parameters associated with land-atmosphere interactions in WRF-Chem model to better predict O3 pollution

The Weather Research and Forecasting  with Chemistry model (WRF-Chem) is one of the state-of-art models for studying air quality. Its meteorological module and chemical module are fully coupled, making it an ideal model for exploring the interaction between meteorological and chemical processes. However, it still needs great improvement in simulating near-surface ozone in a heavy pollution event. Except for the emissions, the model parameterization processes and parameters are one of the most critical factors, meanwhile, existing great uncertainties. It is of great significance to find out the most important parameterization processes and parameters that affect the simulation results for accurate simulation. The model simulation performance is usually estimated by comparing the simulation variable with observations of some indicators, such as relative humidity, wind speed, temperature, short-wave radiation flux and boundary layer height, which have important effects on ozone concentration. By calculating the sensitivity of ozone concentration and these factors to the parameters through some parameter sensitivity experiments, the key parameters and physico-chemical processes for ozone simulation can be found out. At present, it is known that the land-atmosphere coupling process has a great influence on ozone simulation, but it is not clear which mechanism and parameter are the key factors. For this purpose, a series of parameter sensitivity experiments were designed. This study considered the land surface process, planetary boundary process, cloud microphysics process, near-surface layer process and cumulus cloud process. Six microphysics schemes, three groups of near-surface schemes, six boundary layer schemes and three cloud microphysics schemes with the best performance in WRF-Chem were selected, and a total of 120 simulations were performed. The Morris one-at-a-time (MOAT) method was used to screen out the physical and chemical processes and parameters that have important effects on ozone pollution and adaptive surrogate modeling-based optimization (ASMO) method was used to optimize these key parameters, which can explore the role of different physical processes in regulating land-atmosphere interaction, quantify the uncertainty of model physical processes, and provide evidence to improve the model physical parameterization, so as to improve the near-surface ozone simulation.