Development of nitrogen dioxide and volatile organic compounds land use regression models to estimate air pollution exposure near an Italian airport

Abstract The aim of this study was to evaluate the small scale spatial variability of nitrogen dioxide (NO 2 ) and selected VOCs (benzene, toluene, acrolein and formaldehyde) concentrations using Land Use Regression models (LURs) in a complex multi sources domain (64 km 2 ), containing a mid-size airport: the Ciampino Airport, located in Ciampino, Rome, Italy. 46 diffusion tube samplers were deployed within a domain centred in the airport over two 2-weekly periods (June 2011–January 2012). GIS-derived predictor variables, with varying buffer size, were evaluated to model spatial variation of NO 2 , benzene, toluene, formaldehyde and acrolein annual average concentrations. The airport apportionment to air quality was investigated using a Lagrangian dispersion model (SPRAY). A stepwise selection procedure was used to develop the linear regression models. The models were validated using leave one out cross validation (LOOCV) method. In this study, the use of LURs was found to be effective to explain spatial variability of NO 2 (adjusted-R 2  = 0.72), benzene (adjusted-R 2  = 0.53), toluene (adjusted-R 2  = 0.50) and acrolein (adjusted-R 2  = 0.51), while limited power was achieved with the formaldehyde modeling (adjusted-R 2  = 0.24). For all pollutants LURs output showed that the small scale spatial variability was mainly explained by local traffic. The airport contribution to the observed spatial variability was adequately quantified only for acrolein (0.43 (±0.69) μg/m 3 in an area of about 6 km 2 , SW located to the airport runway), while for NO 2 and formaldehyde, only a little portion of the spatial variability in a limited portion of the study domain was attributable to airport related emissions.