Liu Bing-Jie, Peng Xiao-Min, Li Ji-Hong
School of Forestry, Northeast Forestry University, Harbin 150040, China.
School of Geography and Ocean Science, Nanjing University, Nanjing 210023, China.
Huan Jing Ke Xue. 2018 Dec 8;39(12):5296-5307. doi: 10.13227/j.hjkx.201805051.
The land use regression (LUR) model is one of the most important systematic methods to simulate the temporal and spatial differentiation of the atmospheric pollutant concentration. To explore the adaptability of the LUR model to the simulation of air pollutants at the national scale in China and the temporal and spatial variation characteristics of fine air particulate matter (PM) in China in 2015 and its correlation with different geographical elements, we built a LUR model. The LUR model is based on a geographically weighted algorithm using PM data acquired from the national control monitoring site in 2015 as the dependent variable and applying factors such as the type of land use, altitude, population, road traffic, and meteorological elements as independent variables. Based on model regression mapping, we obtained the distributions of monthly and annual PM concentrations nationwide in 2015 and analyzed the temporal and spatial variation characteristics of PM concentrations using the Hu line as a reference line. The results indicate that introducing the geographically weighted algorithm can significantly reduce the residual Moran's Ⅰ of the LUR model, weaken the spatial autocorrelation of residuals, and improve the coefficient of determination , which is better to reveal the complex relationship between the spatial distribution and impact factors of PM. Cropland, forest, grass and urban industrial and residential land, and meteorological elements and major roads noticeably impact the PM concentration. Different spatial distributions of different geographical elements have distinct effects on PM. The PM shows distinct temporal and spatial differences on both sides of the Hu line. The PM concentration is relatively high in developed cities with a large population and high industrialization levels. The concentration of PM is higher in winter and gradually decreases in autumn, spring, and summer.
土地利用回归（LUR）模型是模拟大气污染物浓度时空分异的重要系统方法之一。为探究LUR模型在中国全国尺度上对空气污染物模拟的适应性以及2015年中国细颗粒物（PM）的时空变化特征及其与不同地理要素的相关性，我们构建了一个LUR模型。该LUR模型基于地理加权算法，以2015年从国家控制监测站点获取的PM数据作为因变量，并将土地利用类型、海拔、人口、道路交通和气象要素等因素作为自变量。基于模型回归制图，我们得到了2015年全国PM月浓度和年浓度的分布情况，并以胡焕庸线为参照线分析了PM浓度的时空变化特征。结果表明，引入地理加权算法能够显著降低LUR模型的残差Moran's Ⅰ，减弱残差的空间自相关性，提高决定系数，能更好地揭示PM空间分布与影响因素之间的复杂关系。农田、森林、草地、城市工业和居住用地以及气象要素和主要道路对PM浓度有显著影响。不同地理要素的不同空间分布对PM的影响各异。PM在胡焕庸线两侧呈现出明显的时空差异。在人口众多、工业化水平高的发达城市，PM浓度相对较高。PM浓度在冬季较高，在秋季、春季和夏季逐渐降低。
