【作者】 张波；

【导师】 王卫；

【作者基本信息】 河北师范大学 ， 自然地理学， 2020， 硕士

【摘要】 近年来随着我国社会经济的快速发展,人类活动和自然界会排放大量污染物质,这就使得大气中大气颗粒物越来越多,而其中PM_2.5不仅对空气质量和能见度等有重要的影响,而且还对当地居民的人身健康造成严重威胁。京津冀地区由于受到其独特的地形和气候特征的影响,使得大气污染物难以向外迁移扩散,导致京津冀地区PM_2.5污染状况尤为严重。为深入认识京津冀地区地面污染源因子、气象条件与大气颗粒物浓度间的复杂影响关系,揭示研究区大气PM_2.5时空分布规律,本文以京津冀为研究区,首先以遥感反演的气溶胶光学厚度（AOD）因子、气象因子、土地利用因子与近地面层PM_2.5浓度为基础数据构建了混合效应模型,在此基础上,通过引入两个变量（大气AOD数据的平方,AOD数据与边界层高度的乘积）建立了非线性混合效应模型进一步探讨其关系;针对由于AOD数据缺失导致的预测结果缺失的情况,又引入当天所有可用PM_2.5监测站点上测量的平均PM_2.5浓度,建立了第二段模型来对其进行填补。最终获得最高覆盖度、高时间分辨率（每天）、空间分辨率为10×10km的PM_2.5浓度时空变化模拟结果。研究结果显示:京津冀地区2013-2016年的第一阶段非线性混合模型预测精度经交叉验证后R²分别为0.79、086、0.84和0.85,RMSE和RPE分别为25.52μg/m³、32.19%、19.67μg/m³、26.26%、17.72μg/m³、27.93%和16.85μg/m³、28.13%,与线性混合效应模型相比,较明显提高了区域PM_2.5浓度的模拟效果。另外,统计得2013-2016各年第二阶段模型对网格单元填补个数分别为67612、66553、66001和66069,PM_2.5浓度预测均值分别提高到了78.20μg/m³、66.66μg/m³、54.44μg/m³和49.27μg/m³。上述研究结果表明非线性混合效应模型并进行补值后可明显提升模型预测精度,而通过引入恰当因子提升模型空间预测范围,有效弥补地面监测站点时间和空间上的数据缺失,为我国环境治理和规划提供科学参考。更多还原

【Abstract】 In recent years,with the rapid development of China’s social economy,human activities and nature will emit a large amount of pollutants,which has caused more and more atmospheric particulate matter in the atmosphere,and PM_2.5.5 not only has an important impact on air quality and visibility,It also poses a serious threat to the personal health of local residents.Due to the unique topography and climatic characteristics of the Beijing-Tianjin-Hebei region,it is difficult for air pollutants to migrate out and diffuse,resulting in a particularly serious PM_2.5.5 pollution in the Beijing-Tianjin-Hebei region.In order to gain a deeper understanding of the complex influence relationship between ground pollution source factors,meteorological conditions and atmospheric particulate matter concentration in the Beijing-Tianjin-Hebei region,and to reveal the temporal and spatial distribution of atmospheric PM_2.5.5 in the study area,this paper takes Beijing-Tianjin-Hebei as the study area,first using remote sensing inversion Sol-optical thickness（AOD）factor,meteorological factor,land use factor and PM_2.5concentration of near-surface layer were used as basic data to construct a mixed effect model.On this basis,two variables（square of atmospheric AOD data,AOD data）were introduced The product of the height of the boundary layer)established a nonlinear mixed effect model to further explore its relationship;for the lack of prediction results due to the non-randomness of the AOD data,the average PM_2.5measured on all available PM_2.5.5 monitoring stations for the day was introduced concentration,a second-stage model was established to fill it.Finally,the simulation results of the spatiotemporal variation of PM_2.5.5 concentration with the highest coverage,high temporal resolution（every day）,and spatial resolution of 10×10km were obtained.The research results show that the prediction accuracy of the first-stage nonlinear mixed model in the Beijing-Tianjin-Hebei region from 2013 to2016 is cross-validated,R2 is 0.79,086,0.84 and 0.85,RMSE and RPE are25.52μg/m³,32.19%,19.67μg/m³,26.26%,17.72μg/m³,27.93%and 16.85μg/m³,28.13%,greatly improving the simulation effect of regional PM_2.5.5 concentration.In addition,the statistics show that the number of grid cells filled by the second-stage model in 2013-2016 is 67612,66553,66001 and 66069 respectively,and the average predicted PM_2.5.5 concentration has been increased to 78.20μg/m³,66.66μg/m³ and54.44μg/m³ and 49.27μg/m³.The above research results show that the nonlinear mixed-effect model and the supplementary value can greatly improve the prediction accuracy of the model,and by introducing appropriate factors to improve the spatial prediction range of the model,it can effectively make up for the lack of data in the time and space of the ground monitoring station,and provide environmental protection for China.更多还原