Improved method of visibility parameterization focusing on high humidity and aerosol concentrations during fog–haze events: Application in the GRAPES_CAUCE model in Jing-Jin-Ji, China

Abstract Atmospheric visibility affects people's everyday lives and its simulation accuracy is therefore important, especially in low visibility conditions. Two fog–haze pollution episodes (episode 1, 16–21 December 2016 and episode 2, 30 December 2016–7 January 2017) in the Beijing–Tianjin–Hebei (Jing-Jin-Ji) region of China are simulated using the GRAPES_CUACE atmospheric chemistry model. The simulation of low visibility is significantly overestimated by the model. The observational data show that a decrease in visibility is closely related to the increase of the mass concentration of aerosol particles and the increase of the relative humidity. However, the contribution of aerosol particles to extremely low visibility is less than that of the relative humidity. When visibility is  80% or near saturation. High humidity promotes the hygroscopic growth of aerosols, and it also favors the formation of fog droplets. Besides the strongly enhanced aerosol extinction coefficient due to hygroscopic growth at high relative humidity conditions that may be underestimated by the IMPROVE scheme, the further degradation of visibility may also depend on extinction from fog droplets which is absent in the IMPROVE scheme. An improved visibility parameterization focusing on low visibility is proposed based on the IMPROVE equation by including the direct extinction coefficient b f of fog droplets in the original parameterization. Focusing on the two fog–haze episodes, the simulation results with the improved parameterization show an overall improvement in the simulation of low visibility over the whole Jing-Jin-Ji region. The errors in the simulation of visibility by the improved parameterization averaged over the Jing-Jin-Ji region are reduced by 65%–88% relative to the original parameterization. In addition, the improved parameterization greatly improves the visibility simulation accuracies (VSAs) of 5 km, especially 3 km.