Examination of the Mathematical Models Used in Atmospheric Source Apportionment Studies

查明造成特定地点空气中颗粒物组成的各种来源，是制定有效的空气质量管制政策的重要前提。在过去的十年中，越来越多地使用元素和化学分析方法来表征空气中的颗粒物和收集的降水。随着这些大型数据集的出现和计算机规模和速度的不断提高而变得实用的新的多变量统计方法，有了更多的机会从收集的颗粒或降水的特征数据中提取有关大气过程的信息。由于一些元素在地球化学上有很强的相关性，元素数据中存在冗余，这限制了在识别来源方面有影响的参数的数量。另一个数据来源是计算机控制的扫描电子显微镜对单个颗粒的物理表征(大小、形状、质地)。本项目将研究几种基于特征向量的统计方法，以调查空气中颗粒物的区域尺度输送。通过主成分分析、经验正交函数和三模因子分析，可以更好地了解气溶胶组成的时空变化规律，并确定与区域气象的关系。此外，还将开发从扫描电子显微镜获得的单个颗粒的图像中提取颗粒纹理的定量测量的方法。将探讨分维、特征向量和傅立叶分析方法。通过与气体物质的反应对颗粒表面的化学修饰对视觉表面纹理和图像分析方法的结果的影响将被检验。

Identifying the various sources responsible for the composition of airborne particles at a given location is an important prerequisite to developing effective regulatory policies for air quality. During the past decade, there has been increasing use of elemental and chemical analytical methods to characterized airborne particulate matter and collected precipitation. With the availability of these large data sets and new multivariate of statistical methods that have been made practical by the continuing increases in computer size and speed, there is an enhanced opportunity to extract information about atmospheric processes from data characterizing collected particles or precipitation. Since some elements are strongly geochemically related, there is redundancy in the elemented data that limits the number of parameters that are influential in identifying sources. Another source of data is physical characterization (size, shape, texture) of individual particles by computer-controlled scanning electron microscopy. This project will examine several eigenvector-based statistical methods to investigate regional scale transport of airborne particulate matter. By using principal components analysis, empirical orthogonal functions, and three-mode factor analysis, is it expected that the patterns of spatial and temporal variations in the aerosol compositions can be better understood and relationships with the regional meteorology defined. In addition, methods will be developed for extracting quantitative measures of particle texture from images of individual particles obtained with a Scanning Electron Microscope. Fractal dimension, eigenvector and Fourier analysis methods will be explored. The effects of chemical modification of the particle surface by reaction with gaseous species on the visual surface texture and results of the image analysis methods will be examined.