Remote sensing detection of the spatial pattern of urban air pollution in Los Angeles

Traditional monitoring method of PM_2.5 concentrations with field campaigns cannot accurately identify the spatial pattern of air pollution in urban areas. Remote sensing techniques have been applied to monitor the distribution of atmospheric particulate pollution. However, remotely sensed aerosol data products with low spatial-resolution cannot reveal the spatial variations of urban air pollution. In this study, urban aerosol optical depth (AOD) data with 500 m resolution was generated using the Moderate Resolution Imaging Spectroradionmeter (MODIS) image data for the Greater Los Angeles area. The AOD was then used to build a land-use based regression (LUR) model (Model B) for mapping the urban PM_2.5 concentration, by combining with population density and leaf area index. The accuracy of the modeling method was evaluated by comparing with the results of LUR model (Model A) without AOD and of Ordinary Kriging (OK) interpolation. The results show that: (1) the AOD values varied over the city, and were higher in the downtown area; (2) correlation coefficient of LUR model increased from 0.28 to 0.35 by incorporating AOD data; and (3) the proposed LUR model (B) can well reveal the distribution of air pollution with a smaller relative error than the Ordinary Kriging interpolation method. It is suggested that the AOD aided LUR model offers a potential to reveal the spatial pattern of PM_2.5 pollution with “high spatial resolution” in urban areas, and can thus provide support for mitigating the growingly concerned air pollution in city worldwide.