Method to estimate PM10 concentration based on eigenvalues of CCD images in Lanzhou urban area

In this study, photogrammetry was used to estimate the PM10 concentration value in Lanzhou urban area. First, the CCD images were collected in a fixed position from a heating period in Lanzhou, that is, from November in 2009 to March 2010. At the same time, PM10 concentration was measured. Secondly, image eigenvaluess (definition) are calculated through eight methods. Thirdly, the relationship between image eigenvalues and PM10 concentration are built through regression analysis method. From the study, the results show that the negative exponential relationship between image definition value derived from Sobel algorithm and PM10 concentration is the optimal model (R²=0.83, n=113). Additionally, image eigenvalues can be utilized to estimate PM10 concentration. The inversion results with comparative high accuracy could determine the air pollution levels when particulate matter is the main pollutant.