UK hyperlocal air pollution model – Use of inverse modelling to increase the spatial detail of air pollution predictions

Abstract

Background: Air pollution models are increasingly able to provide forecasts of NOx, NO₂, O₃, PM₂.₅ and PM₁₀ at country scales (every 1-2km) and close to roads (every 10-20m). However, a significant limitation of these models is the lack of road traffic flow, speed and emissions. Aim: To develop a hyperlocal model which predicts air pollution every 20m across all of UK’s roads. Method: Here we address the lack of detailed traffic counts, speed and emissions by utilising ~12,000 diffusion tube and local sensor measurements in the UK, with a road traffic emissions optimisation method (EOM). The EOM uses model sensitivity coefficients in Taylor series expansions of the models’ response to varying emissions, allowing Monte Carlo emissions calibration to be performed at every monitoring location. This has allowed the creation of previously unavailable road emissions estimates, as well as the calibration of emissions on those roads that were already in the model. For other sources we used a combination of European, UK National Atmospheric Emissions Inventories. Air pollution was predicted using the WRF met. model and the CMAQ-urban coupled model. Results: The EOM has enabled an emissions calibration process of unprecedented scale – improving traffic emissions estimates and enabling air pollution predictions at 20m resolution for every road in the UK. The NOx, NO₂ and O₃ model performance compares favourably with out of sample measurements and the EOM addresses the uncertainty NOx, NO₂ emissions from road transport, which has hitherto been under predicted. Conclusions: We have demonstrated what is possible with coupled CTM and local scale air pollution models, when combined with sufficiently detailed emissions inventories. This new approach is ideally suited for use with low cost sensor networks globally, providing data, including for PM₂.₅ and PM₁₀, for local populations, policy development and health research, across entire cities and even countries.