Led by: Eleanor Smith, Met Office
In recent years the importance of understanding the impact of poor air quality on our health has moved higher up the political agenda. Efforts to fully understand the relationship between pollution episodes and the onset or deterioration of our health have been stepped up. It is acknowledged that pollution measurements from roadside and other locations do not provide the whole story as most of us do not spend a significant time in polluted outdoor locations. Instead, developing a clearer understanding of our actual exposure to polluted air (both indoor and outdoor) is key to making the link between pollution levels and health responses.
Air quality exposure assessments are typically performed using temporally or spatially limited modelling or measurement data. This places numerous limitations on health impact assessments. Whilst personal air quality exposure occurs at the individual scale, it is not practical to model this explicitly. However, it is also true that government policy, interventions and many other actions are by necessity based on wider population data and are also applied to broad cross sections of that population.
The Met Office provides the national air quality forecast. The modelling infrastructure created to deliver the best possible forecast is also ideally suited to generate what the meteorological community call a reanalysis. This is a model simulation in which observational data from the past is used to constrain the model, with the combined model-observation fusion providing a greatly improved re-creation of the past reality across the entire modelling domain. Such an approach requires a model and a system by which measurement data can be used within the model or an associated processing system. Such approaches, also used for forecasting and in meteorological modelling, are commonly called data assimilation.
The intention of this work is to generate 15 years of air quality concentrations for the whole of the UK. This will provide useful data for those looking to generate health impact studies, which usually require a generous timescale in order to analyse the health impact signals. It would also benefit those researchers involved in more localised (street level) air quality exposure calculations who will use the dataset as boundary conditions for their more localised analyses.