Land use regression models using satellite aerosol optical depth observations and 3d building data from the central cities of Liaoning province, China

Land use regression (LUR) modeling is a promising method for assessing the spatial variation of air pollutant concentrations. We developed an LUR model for air pollutants (SO₂, NO₂, and PM₁₀) in the central cities of Liaoning Province using monitoring data collected during 2013. We evaluated whether the addition of annual satellite aerosol optical depth (AOD) observations and five canyon indicators (building height, building coverage ratio, floor area ratio, building shape coefficient, and high-rise building ratio) improved the LUR models. Out-of-sample “10-fold” cross validation was used to quantify the accuracy of the model predictions. Our results showed that the gross domestic product (GDP) and the distance to the nearest industrial emissions were the common variables for the models. Annual AOD demonstrated weak correlations with air pollutant concentrations because of its instantaneity, low resolution, and limited precision; however, it was useful for improving the coefficient of determination (R2) of the LUR models. The full models incorporating the annual AOD data and canyon indicators showed further improvement. The improvements of R2 were 0.22, 0.19, and 0.39 for SO₂, NO₂, and PM₁₀, respectively, demonstrating that the consideration of canyon indicators could still be valuable and could be used in LUR models.