资源环境科技发展态势分析平台

The explosive growth of PM2.5 mass usually results in extreme PM2.5 levels and severe haze pollution in eastern China, and is generally underestimated by current atmospheric chemistry models. Based on one such model, GRAPES_CUACE, three sensitivity experiments - a "background" experiment (EXP1), an "online aerosol feedback" experiment (EXP2), and an "80% decrease in the turbulent diffusion coefficient of chemical tracers" experiment, based on EXP2 (EXP3) - were designed to study the contributions of the aerosol-radiation feedback (AF) and the decrease in the turbulent diffusion coefficient to the explosive growth of PM2.5 during a "red alert" heavy haze event in China's Jing-Jin-Ji (Beijing-Tianjin-Hebei) region. The results showed that the turbulent diffusion coefficient calculated by EXP1 was about 60-70m(-2)s(-1) on a clear day and 30-35m(-2) s(-1) on a haze day. This difference in the diffusion coefficient was not enough to distinguish between the unstable atmosphere on the clear day and the extremely stable atmosphere during the PM2.5 explosive growth stage. Furthermore, the inversion calculated by EXP1 was obviously weaker than the actual inversion from sounding observations on the haze day. This led to a 40 %-51% underestimation of PM2.5 by EXP1; the AF decreased the diffusion coefficient by about 43 %-57% during the PM2.5 explosive growth stage, which obviously strengthened the local inversion. In addition, the local inversion indicated by EXP2 was much closer to the sounding observations than that indicated by EXP1. This resulted in a 20%-25% reduction of PM2.5 negative errors in the model, with errors as low as 16% to 11% in EXP2. However, the inversion produced by EXP2 was still weaker than the actual observations, and the AF alone could not completely explain the PM2.5 underestimation. Based on EXP2, the 80% decrease in the turbulent diffusion coefficient of chemical tracers in EXP3 resulted in near-zero turbulent diffusion, referred to as a "turbulent intermittence" atmospheric state, which subsequently resulted in a further 14%-20% reduction of the PM2.5 underestimation; moreover, the negative PM2.5 errors were reduced to 11% to 2%. The combined effects of the AF and the decrease in the turbulent diffusion coefficient explained over 79% of the underestimation of the explosive growth of PM2.5 in this study. The results show that online calculation of the AF is essential for the prediction of PM2.5 explosive growth and peaks during severe haze in China's Jing-Jin-Ji region. Furthermore, an improvement in the planetary boundary layer scheme with respect to extremely stable atmospheric stratification is essential for a reasonable description of local " turbulent intermittence" and a more accurate prediction of PM2.5 explosive growth during severe haze in this region of China.