【摘要】 以杭州市2013—2016年秋冬季PM_2.5及影响因素为研究对象,构建统计动力学反演模型,研究PM_2.5浓度系统的影响因素与演化特征。结果显示,（1）模型反演预测值与实际监测值的相关系数0.8316,贡献率大于1%的各驱动项之和为90.85%,反演效果较好;模型中线性项贡献率之和为21.65%,非线性项贡献率之和为62.88%,PM_2.5浓度变化受非线性影响更加显著。（2）包含NO₂、SO₂、CO的各驱动项贡献率之和分别为10.77%、8.42%、6.58%;包含降雨（PRE）、风速（WIND）的各驱动项贡献率之和分别为20.5%、17.02%,NO₂、SO₂、PRE、WIND是杭州市PM_2.5浓度变化的最主要影响因素。（3）贡献率大于2%的各驱动项中,平均气压（PRS）、相对湿度（RHU）、日照时长（SSD）对PM_2.5呈负反馈线性稳定影响作用,PM_2.5×SO₂、NO₂×PRS、PRE×SSD、PM_2.5×WIND等交互项对PM_2.5浓度变化呈正反馈非线性影响,PM_2.5×O₃、SO₂×NO₂、PRE×WIND等交互项对PM_2.5浓度变化呈负反馈非线性稳定作用。（4）杭州市PM_2.5浓度变化系统是一元二次非线性驱动演变系统,通过数值模拟分析可知其PM_2.5浓度变化系统呈双定态式,PM_2.5浓度变化系统在定态值0.2413与0.3769之间波动变化,系统平衡态不稳定。结果表明,统计动力反演模型可定量解析出PM_2.5浓度变化的线性驱动项与非线性驱动项的系数值及贡献率,有助于揭示PM_2.5浓度变化系统的演化特征,研究结果可应用于大气污染控制与环境管理等方面。更多还原

【Abstract】 To reveal the evolution characteristics of PM_2.5 concentration in autumn and winter in 2013—2016, and its relationships with the influencing factors, the statistical dynamic inversion model between PM_2.5 and its driving factors was constructed by taking Hanzhou city as a case study. The results showed that:（1） The correlation coefficient between the predicted value and the actually monitoring value was 0.8316 in the dynamic inversion equations, and the sum contributions of all the drivers with contribution rate above 1% were 90.85%. The performance of the inversion model was good. The contributions of the linear effects of the driving factors were 21.65%, while, the contributions of nonlinear effects were 62.88%. Thus, the changes of PM_2.5 concentrations in Hangzhou City were more strongly influenced by the nonlinear effects of the driving factors. The contribution rates of NO₂, SO₂ and CO were 10.77%, 8.42% and 6.58%, respectively. The contribution rates of precipitation（PRE） and wind（WIND） were 20.5% and 17.02%, respectively. Therefore, NO₂, SO₂, PRE and WIND were the most important driving factors of the changes of PM_2.5 concentration in Hangzhou city.（3） The driving factor with a contribution rate more than 2% was assumed to have an important influence on the change of PM_2.5 concentration. Among these factors, the average pressure（PRS）, the relative humidity（RHU） and the sunshine duration（SSD） had negative feedback effect on the linear stability of PM_2.5 concentration. The interaction terms of PM_2.5-SO₂, NO₂-PRS, PRE-SSD and PM_2.5-WIND had positive feedback effect on the nonlinear change of PM_2.5 concentration. The interactive terms of PM_2.5-O₃, SO₂-NO₂ and PRE-WIND had negatively nonlinear feedback effects on the stability of PM_2.5 concentration.（4） PM_2.5 concentration change was a one-place quadratic nonlinear dynamical system. Through numerical simulation analysis, we found that the system of PM_2.5 concentration change had double steady states. The values in steady state of PM_2.5 concentration varied between 0.2413 and 0.3769. The equilibrium state of the PM_2.5 concentration change system was unstable. Our results demonstrate that the nonlinear dynamic inversion model can be used to quantitatively analyze the linear and nonlinear effects of the driving forces of the PM_2.5 concentration changes and their contributions, and further to reveal the evolution characteristics of PM_2.5 concentration change. Our findings will help to provide some scientific references for air pollution control and environmental management.更多还原