Monitoring Air Pollution Using GIS: Case Study for the City of Belgrade

Abstract

The methodology for modeling the distribution of certain air pollutant for the city of Belgrade in winter 2015 is presented in the paper. Land Use Regression (LUR) was used for modeling and visualization of spatial distribution of air pollution concentration in the city. NO2 concentrations were sampled at 46 locations, and predictive variables were calculated based on the road category, traffic intensity, demographic data, altitude, household furnaces and land use. These variables were calculated using buffers of different sizes. Linear regressions between NO2 and predictive spatial variables were calculated. Afterwards, the most significant predictors were used for multivariate regression model. Quality of the final model was checked using measurement available at certain locations. The RMSE of 9.8 μg/m³ and the coefficient of determination (R2) of 0.6 were obtained. These results indicate that traffic has the largest impact on air pollution concentration especially near the major roads. Prediction should help in deciding which air protection measures are to be taken in order to preserve and improve the city environment. The lack of data that are collected by using quite a few sensor stations is still rather limiting for the successful monitoring of air pollution in the city of Belgrade.