Modelling the impact of road traffic on ground level ozone concentration using a quantile regression approach

Abstract

Road traffic is both a major source of ozone precursors (e.g. nitrogen oxides and hydrocarbons) and a potential local sink for ozone in the form of fresh nitric oxide (NO) that depletes ozone. This study investigates the effect of road traffic characteristics on ground level ozone concentration (ppb) applying a quantile regression model (QRM). QRM has certain advantages over other regression methods, including its applicability to non-normal ozone distribution and its ability to handle non-linearities in the relationship of ozone with its covariates. The paper is developed in two parts. In the first part ozone concentrations at urban and rural sites have been compared using data from 80 ozone monitoring sites throughout the UK. The model results in an average urban decrement of about 7 ppb (26%) but indicates variations at various quantiles of the ozone distribution, for instance the difference is 5.25 ppb (25%) and 10.78 ppb (30%) at quantile 0.1 and 0.99, respectively. In the second part the effect of road-traffic characteristics (traffic flow, speed and fleet composition) on urban decrement has been modelled, as a case study in Leeds, UK. The relationship between urban decrement and road traffic characteristics changes at different regimes of ozone distribution indicating a highly non-linear association. Flow of cars, buses and articulated heavy vehicles seem to have the strongest effect on urban decrement; however buses are the only category showing significant effect at all quantiles. The effect of average speed and motorcycles flow was not significant. The results of QRM show that up to 86% ozone variations between rural and urban sites can be explained with the help of traffic characteristics. The effect of various traffic scenarios on urban decrements has been investigated.