珠三角关键大气挥发性有机物的模拟精度评估

大气挥发性有机物（VOCs）是导致臭氧污染的关键前体物，是城市空气质量建模不可或缺的重要组成部分，但由于其非常复杂的构成和来源以及监测数据缺乏，目前对其模拟精度的了解仍非常有限。本文利用嵌套网格空气质量模式预报系统（NAQPMS）对珠江三角洲（简称珠三角）地区2017年9月21日至11月20日的VOCs开展了模拟试验，并利用光化学监测网8个地面站点的VOCs浓度监测数据，对模式模拟的关键VOCs组分进行了精度评估。结果发现，模式对强活性的甲苯、乙烯和二甲苯具有较高的模拟精度，模拟浓度偏差百分比为0.4%～26.6%，模拟能较好再现其日均浓度变化趋势和日变化的双峰特征。但是模式对化学反应活性强且与植物排放密切相关的异戊二烯具有很大的模拟偏差，偏差比近100%，无法再现其白天浓度高、夜间浓度低的观测日变化特征。通过分析发现，现有模拟系统主要考虑人为污染物排放而未考虑生物源排放，可能是导致这一模拟偏差的关键原因。同时，评估结果也表明模式在VOCs空间分布模拟上仍面临很大的不确定性。本文结果揭示了珠三角VOCs模拟面临的关键不确定性，表明融合VOCs观测数据来揭示并减小VOCs模拟的不确定性具有非常迫切的需求。

Simulation Accuracy Evaluation of Key Atmospheric Volatile Organic Compounds in the Pearl River Delta

VOCs (volatile organic compounds), key precursors to the ozone pollutant, are indispensable parts of urban air quality modeling. Owing to their complex composition and the lack of monitoring data, understanding of their simulation accuracy is still poor. Here, simulations of VOCs in the Pearl River Delta region were performed using the nested grid air quality model prediction system (NAQPMS) from 21 September to 20 November 2017. First, monitoring of VOC concentrations from eight ground stations of the photochemical assessment monitoring stations was conducted to evaluate the accuracy of key VOC components. The results showed that the model has high simulation accuracy for toluene, ethylene, and xylene with concentration deviation ratios of 0.4%–26.6%, which can well reproduce the trend of daily average concentration and double-peak characteristics of diurnal variation. However, the model has a large simulation deviation for isoprene with strong chemical reaction activity and closed relation to plant emissions. The deviation ratio was nearly 100%, which cannot reproduce the diurnal variation characteristics of high concentration during the day and low concentration at night. The total amount of VOCs emitted by plants in the Pearl River Delta region was relatively high. Hence, ignoring the biological VOC emissions in the current simulation system can be the key reason for this simulation deviation. In addition, the results of simulation evaluation showed that the model still has great uncertainty in the VOC spatial distribution. With these, combining VOC experimental data with simulations to reveal and reduce the uncertainty of VOC simulations is necessary.