基于地理国情数据的PM_(2.5)因子多元回归分析

Multiple regression analysis on PM_(2.5) impact factors based on geographic conditions monitoring data

针对对PM2.5和与其相关的测绘成果进行的相关分析和回归模型分析相对不足的问题,该文以河北省石家庄市为研究对象,在相关因素分析的基础上,运用多元线性回归分析与多元非线性回归分析两种方法对石家庄市PM2.5浓度与扬尘地表、工业企业分布、地表覆盖以及道路等地理国情数据进行回归建模,并进行对比分析,根据判定系数R2得到最优建模方法及PM2.5重要影响因素及其影响关系。结果表明,该实验中多元非线性回归分析能获得较好的拟合效果,由模型可以看出扬尘地表、未利用地、人造覆盖面积与PM2.5呈正相关,是影响PM2.5的重要因素。该研究结果对于认识空气中PM2.5的来源与分布特征具有重要的参考价值。

As air pollution in China has been increasingly exacerbated, the relevant factors analysis and modeling on the PM2.5 sources and distribution is becoming a significant research direction. In this paper, Shijiazhuang City in Hebei Province was chosen as the research area, based on the analysis of relevant factors, regression modeling was processed on Shijiazhuang PM2.5 concentration and certain geographical condition data, such as ground dust distribution, distribution of industrial enterprises in key industry, land cover and road data, using multiple linear regression analysis method and multivariate nonlinear regression analysis method. After comparative analysis, the optimal modeling method and the relationship between PM2.5 and its significant impact factors were figured out with the measure of the coefficient of determination R2. The results showed that the multiple nonlinear regression analysis method could get a better fitting result, and the ground dust, unused land, and area of artificial were important factors positively correlated to PM2.5 reflected from the model, which would provide a reference for the understanding of the origin and distribution of PM2. 5 in the air.