An evaluation of methods for normalising short campaign outdoor personal exposure data to improve long-term representativeness using background site measurements

Using a portable air sampler is a highly effective means to measure personal exposure and to map spatial distributions in high resolution. Nevertheless, the available data is descriptive only of the specific period of measurement, largely due to the influence of meteorological factors. Data normalisation has been conducted to mitigate the impact of weather conditions on air pollutant concentration. However, this technique has been mostly used on long-term data obtained from stationary air monitoring stations using machine learning. This study presents the first application of simple background-site-based normalisation methods to short-term walking personal exposure measurements. Several normalisation approaches were evaluated using simultaneous data from a nearby reference air quality monitoring station in an urban neighbourhood in Birmingham. The methods were applied to Particle Number concentration(PNC), Particulate Matter mass (PM1, PM2.5, and PM10), and Black Carbon (BC) obtained from walking measurements with the aim of reducing temporal sampling influences and increasing representativeness of short-term mobile measurement relative to long-term background conditions. The results showed that some methods appreciably altered the raw data, thereby making them comparable to the long-term background conditions. For certain pollutants (PNC and BC), the normalized concentration was still greater than the annual average at the reference site, which was expected due to the presence of additional local sources along the walking route. These findings demonstrate that background-site-based normalisation can be utilized to evaluate both personal exposure and the spatial distribution of pollutants so as to reflect conditions at a certain point in time.