本文选取了雅安市2015~2018年的空气污染物SO₂、NO₂、CO逐日、逐小时浓度监测数据和同期气温、气压、相对湿度、雨量与10 min平均风速5个气象要素逐日数据，采用多尺度趋势分析法得出3种空气污染物浓度的时间变化特征，然后通过灰色关联度分析法、相关系数法从不同角度来分析污染物浓度与各气象要素的相关性。结果表明，SO₂、NO₂浓度的年变化呈现先上升后下降的波动变化趋势，CO浓度持续下降；3种污染物浓度的季节变化都表现为冬春高、夏秋相对低的特征；SO₂、NO₂浓度的月均值都呈先减后增的周期性变化趋势，CO浓度的波动变化较为平缓；各气象要素对SO₂、NO₂、CO浓度的影响程度不同，不同年份、不同季节3种空气污染物浓度与各气象要素之间的关联度大小也不相同；SO₂浓度与气温、相对湿度、雨量正相关，与气压、10 min平均风速负相关，而NO₂、CO浓度与气温、雨量以及10 min平均风速都呈负相关，与气压、相对湿度呈正相关。
This paper selects the daily and hourly concentration monitoring data of air pollutants SO₂, NO₂ and CO in Ya’an City from 2015 to 2018, and the daily data of 5 meteorological elements of temperature, pressure, relative humidity, rainfall and 10 min average wind speed in the same period, using multi-scale. Trend analysis method obtained the time variation characteristics of three kinds of air pollutant concentrations, and then analyzed the correlation between pollutant concentration and various meteorological elements from different angles by grey correlation analysis method and correlation coefficient method. The results showed that the annual variation of SO₂ and NO₂ concentration showed a trend of rising and then decreasing, and the CO concentration continued to decrease; as for seasonal changes , three pollutants present a higher concentration in winter and spring, lower concentration in summer and autumn relatively; the concentrations of SO₂ and NO₂ show a periodic changes with decreasing first and then increasing, and the fluctuations of CO concentration are relatively flat; the influence of meteorological elements on SO₂, NO₂, CO concentration is different, and the correlation degree between 3 air pollutants with meteorological elements is also different in different years and seasons; the concentration of SO₂ is positively correlated with air temperature, relative humidity and rainfall, and negatively correlated with air pressure and 10 min average wind speed, while the concentration of NO₂, CO is both negatively correlated with the temperature, rainfall and 10 min average wind speed, positively correlated with air pressure and relative humidity; both BP neural network model and stepwise regression method have better prediction effects of CO than the concentration of SO₂ and NO₂.