Study on variation characteristics and estimation model of PM_2.5 concentration in Beijing in winter

Based on the hourly air quality monitoring data in Beijing in January 2017, the hourly and regional changes of the concentration of PM_2.5 in Beijing were discussed, and the estimation models of PM_2.5, PM₁₀ and major gaseous pollutants of 12 sites were constructed by multiple stepwise regression. The results showed that the hourly concentration change trend of PM_2.5 concentration in Beijing was high at night (18:00-07:00) and low in the daytime (08:00-17:00), the monthly averaged PM_2.5 concentrations of 3 sites in Huairou Town, Changping Town and Dingling were relatively low (78-94 μg∕m ³), while that of other sites were between 106-128 μg∕m ³. The correlation between PM₁₀, 4 kinds of gaseous pollutants and PM_2.5 was in the order of PM₁₀>CO>NO₂>O₃>SO₂. The adjustment coefficient ${R_{\mathrm{adj}}}^2$ of the PM_2.5 concentration estimation model of the 12 sites was all more than 0.96, and the standard error SE was between 13.6-24.5. It is more effective to estimate the PM_2.5 concentration of the site by using the estimation model of the site itself, and the effect of estimating the PM_2.5 concentration was poorer by using the data of other sites.