A Land Use Regression model for predicting concentrations of benzene and 1,3-butadiene accounting for small scale biofuel burning

摘要

BackgroundTraffic is often the most important emission source for ambient air pollution in urban areas, although other sources locally can be at least as important. Small scale biofuel burning emits a wide range of air pollutants, of which many are also substances in vehicle exhaust. AimThis study aimed to develop a Land Use Regression model (LUR) to predict concentrations of benzene and 1,3-butadiene, accounting for emissions from both traffic and burning of biofuels. MethodsYearly average concentrations of benzene and 1,3-butadiene were estimated from diffusive monitoring at 39 sites during three weekly measuring periods in 2011. The majority of sampling sites were placed in areas with high density of biofuel stoves and boilers, based on information from the chimney sweeper's registry. A large number of geographical variables were collected quantifying the area of land use, population density, traffic and heating equipment surrounding each measuring site. This information was later combined in a regression model, to explain the geographical variation in the measured levels. ResultsThe estimated yearly average concentration of benzene was 0.78 μg/m3, ranging from 0.40 to 1.33 μg/m3. The corresponding concentration of 1,3-butadiene was 0.10 ngm3, ranging from 0.06 to 0.20 μg/m3. For benzene, the final LUR model explained 75% (adjusted r2) of the variation (cross validation (CV) 69%) and included five variables, of which two referred to traffic and one to the number of pellet boilers within 100 meters. The final 1,3-butadiene model explained 56% of the variation (CV 37%) and contained five variables of which three referred to biofuel burning. In both models an increasing number of pellet boilers within 100 meters were associated with higher levels of benzene and 1,3-butadiene. Conclusion This study show that benzene and 1,3-butadiene can be predicted with reasonable precision using geographical information, and that small scale biomass burning is an important predictor variable.