边界层方案对南京地区PM2.5浓度模拟的影响

基于WRF-Chem空气质量模式采用YSU、MYJ、MYNN2、ACM2，4种边界层方案和基于ACM2的敏感性实验方案ACM2R2（模式最低6层的湍流扩散系数阈值改为2m²/s）和ACM2R5（模式最低10层的湍流扩散系数阈值改为5m²/s），对南京地区冬季出现的一次天气污染过程进行模拟，分析了不同边界层方案对南京地区PM_2.5浓度模拟的影响.结果表明，6种方案对地面气象要素和风温湿廓线的模拟显示了较为合理的日变化特征及高度变化趋势，不同边界层方案之间气象要素差异较小.各方案均较好地模拟出了边界层高度逐日变化和日变化特征，以及PM_2.5浓度随时间的变化趋势，但是YSU、MYJ、MYNN2和ACM2，4种边界层方案在夜间对PM_2.5浓度的模拟均存在较大程度的高估，而ACM2R2和ACM2R5在ACM2的基础上显著地降低了PM_2.5浓度，甚至转为低估.从整个时段的偏差上看，ACM2R2最接近观测值.这是由于ACM2R2夜间湍流扩散系数较高，更利于源于地表的污染物向上扩散，使得ACM2R2方案在夜间相比原ACM2方案模拟所得的地表PM_2.5浓度更低，从而改善了原方案高估的现象.ACM2R5夜间湍流扩散系数更高，后期扩散过强从而产生较大低估.这些结果表明湍流扩散系数的不同是PM_2.5浓度模拟差异的重要原因，准确地参数化边界层方案夜间湍流扩散系数对于提高PM_2.5浓度模拟的准确度十分必要.

Influence of different boundary layer schemes on PM2.5 concentration simulation in Nanjing

Based on the WRF-Chem (Weather Research and Forecasting Chemistry Model) air quality model, this paper used four boundary layer schemes including YSU (Yonsei University), MYJ (Mellor Yamada Janjic), MYNN2 (Mellor Yamada Nakanishi Niino Level 2), and ACM2 (Asymmetric Convective Model2), and two ACM2-based modified schemes ACM2R2 (the turbulence diffusion coefficient threshold of the lowest 6 layers was set to 2m²/s) and ACM2R5 (the turbulence diffusion coefficient threshold of the lowest 10 layers was set to 5m²/s), the effects of different boundary layer schemes on PM_2.5 concentration simulation in Nanjing were analyzed. The results showed that the surface meteorological elements and wind/temperature/humidity profiles simulated by the six schemes presented reasonable diurnal variation characteristics and height variation, and the difference of meteorological elements among different boundary layer schemes was small. However, the four boundary layer schemes YSU, MYJ, MYNN2, and ACM2 overestimated PM_2.5 concentration at night, while ACM2R2 and ACM2R5 significantly reduced the PM_2.5 concentration, even sometimes to a level of underestimation. In terms of the deviation of the whole period, ACM2R2 was closest to the observed value. This was due to the higher turbulent diffusion coefficient of ACM2R2 at night, which was more conducive to the upward diffusion of pollutants from the surface, and made the surface PM_2.5 concentration of ACM2R2 scheme lower than that of the original ACM2 scheme at night, thus amended the overestimation of the original ACM2 scheme. These results show that the difference in turbulent diffusion coefficient is an important reason for the difference in PM_2.5 concentration simulation, and it is necessary to accurately parameterize the turbulent diffusion coefficient in the boundary layer scheme at night to improve the accuracy of the PM_2.5 concentration simulation.