Investigation of the impact of short-timescale NOx variability on annual mean oxidant partitioning at UK sites

Empirical assessments of annual mean [NO2] have often made use of expressions describing its non-linear [NOx]-dependence, based on fitting to a large number of observations. These studies have also shown that annual mean [NO2] data possess a large amount of scatter, such that the target [NOx] that corresponds to the EU annual mean NO2 limit value (40 μg m−3) can vary considerably from one site to another. A contribution to this scatter is shown to result from site-to-site variations in the [NOx]-averaging of the non-linear impact of the chemical coupling of NOx and O3 over the annual range of ambient conditions. The variability in the partitioning of oxidant (OX) into its component forms of NO2 and O3 is used as a diagnostic, with a particular focus on the impact of short-timescale [NOx] variability. Annual mean monitoring data from 95 UK sites reporting co-located measurements of O3 and NOx are used in the analysis, consisting of 906 site-years of annual mean concentrations measured over the period 1991-2011. The ratio of the upper and lower quartiles of the annual distribution of hourly mean [NOx] (the quartile ratio, QR) is used as a practical indicator of [NOx] variability. The data in a series of QR intervals are shown to deviate systematically from an idealised [NO2]/[OX] vs. [NOx] curve. The data are well-described by [NOx]-averaging the idealised curve, with QR being well correlated with the range of [NOx] over which the curve is averaged, providing strong evidence of a systematic impact of short-timescale [NOx] variability. A combined analysis of the data for sites nominally categorised as roadside, background and rural shows that (although scattered) the values of QR in each category apparently display systematic trends with annual mean [NOx], showing progressive increases in QR with decreasing [NOx] at roadside and polluted background sites. These dependences are used to interpolate between [NO2]/[OX] vs. [NOx] expressions based on constant QR to produce representative expressions for background and roadside sites. These may be appropriate for application in empirical policy models, with the high level of scatter for each site category being rationalised by the associated range in QR.