基于GAMs模型分析成都市气象因子交互作用对O3浓度变化的影响

为探究成都市大气环境中气象因子交互作用对臭氧（8h浓度平均最大值，统一用O₃表示）浓度变化的影响特征，利用成都市2014~2019年逐日大气污染物资料以及同期的气象资料，采用广义相加模型（generalized additive models，GAMs）分析气象因子对O₃浓度变化的影响效应.结果表明，单影响因素的GAMs模型中，O₃浓度与最高气温、日照时数、相对湿度、风速、降水量、最大混合层厚度（maximum mixed depth，MMD）和通风系数（ventilation coefficient，VC）间均呈非线性关系，无论全年还是夏季，最高气温、日照时数、MMD和相对湿度对O₃浓度影响均较大，值得注意的是，夏季相对湿度和降水量对O₃浓度变化的影响较全年更加显著.在构建O₃浓度变化的多气象因子GAMs模型中，除平均风速以外的其他气象因子共同作用对O₃浓度变化有显著影响，就全年而言，构建的GAMs模型判定系数（R²）为0.849，方差解释率为85.1%，最高气温是全年O₃浓度变化的主导影响因素；夏季GAMs模型的R²为0.811，方差解释率为81.3%，而夏季最高气温和MMD同为重要影响因素.GAMs交互效应模型中，就全年而言，最高气温与日照时数、相对湿度、降水量间交互作用，以及日照时数和MMD间交互作用对O₃浓度变化影响显著，结合三维可视化图形直观分析气象因子交互作用对O₃浓度变化的影响特征，发现强高温+强日照+MMD （2000 m左右）+无降水条件协同作用下有利于O₃的生成；就夏季而言，仅最高气温分别与日照时数和VC交互作用对O₃浓度的影响显著，夏季强高温+强日照+水平方向小风速有利于近地层O₃浓度的生成.运用GAMs模型能够对O₃污染的主导气象因子进行识别，并定量化分析气象因子单效应及其交互作用对O₃浓度变化的影响特征，对O₃浓度污染防控研究具有重要指示意义.

Interactive Effects of the Influencing Factors on the Changes of O3 Concentrations Based on GAMs Model in Chengdu

To explore the influence characteristics of the interaction effects between meteorological factors on ozone(O₃) concentration in Chengdu, daily air pollutants and meteorological data from 2014 to 2019 were collected. Generalized additive models(GAMs) were adopted to explore the effects of different factors on O₃ concentration. The results showed that the relationship between O₃ and maximum temperature, sunshine hours, relative humidity, wind speed, precipitation, maximum mixed depth(MMD), and ventilation coefficient(VC) was non-linear. Specifically, the maximum temperature, sunshine hours, MMD, and relative humidity had a significant influence on O₃ concentration throughout the year. It is worth noting that the influence of relative humidity and precipitation on O₃ concentration during summer was more significant than that for the whole year. In the multi-meteorological factors GAMs of O₃ concentration, the meteorological factors mentioned above, except average wind, had significant impacts on O₃ concentration change. For the whole year, the judgment coefficient(R²) was 0.849 and the variance explanation rate was 85.1%. The maximum temperature was the most important influencing factor on O₃ concentration throughout the year. During summer, corresponding R² was 0.811 and the explanation rate of variance was 81.3%; however, maximum temperature and MMD were the dominant meteorological factors. In the interaction GAMs, for the whole year, the interaction between maximum temperature and sunshine hours, relative humidity, and precipitation, and the interaction between sunshine hours and MMD had a significant impact on O₃ concentrations. The interaction between maximum temperature and sunshine hours played a leading role in changes of O₃ concentration. The high temperature+strong radiation+MMD(about 2000 m) +no precipitation were conducive to the formation of O₃ concentration, but in summer, only the maximum temperature, sunshine hours, and VC had the most significant effect on the O₃ concentration, and strong high temperatures+strong radiation+the little horizontal wind in summer were conducive to the formation of O₃ concentration near the surface. In summary, GAMs model can not only be used to identify the dominant influencing factors of O₃ pollution, but also quantitatively analyze the influence of single effects and interaction of influencing factors on O₃ concentration, which has great significance for the prevention and control of O₃ pollution.