Saving old cities: Land use regression model for traffic emissions in the Historical Peninsula of Istanbul

This study aims to develop a pollution distribution model for estimating traffic related intra-urban concentrations of nitrogen dioxide (NO ₂ ) levels. Weekly concentrations of NO ₂ were measured at 45 different locations in the Historical Peninsula of Istanbul during spring, summer and winter seasons in 2010. The range of NO ₂ was 14.2-155 µg/m ³ . A land use regressing (LUR) model was developed to explore the impact of independent variables on the measured levels. Independent model variables were selected based on land use characteristics, traffic and road network information, and meteorological data. Results suggest that 150 metre range is the most effective buffer zone for NO ₂ distribution characteristics in the study area. Average wind speed and temperature data have significant influences (up to 25%) on the prediction performances. Better estimations were produced for spring and winter seasons, particularly for in land stations compared with costal ones. As a result, the overall prediction performance of the constructed model is satisfactory (R ² = 0.64).