Sub-grid Variability and its Impact on Exposure in Regional Scale Air Quality and Integrated Assessment Models: Application of the uEMEP Downscaling Model

Abstract

Regional scale air quality and integrated assessment models are necessarily limited in their spatial resolution, particularly when applied to larger regions such as Europe or for global applications. The EMEP MSC-W chemical transport model and the GAINS integrated assessment model, which makes use of source receptor information precalculated by the EMEP MSC-W model, use resolutions of 0.1°×0.1° and 0.3°×0.2°. These resolutions cannot account for variability at finer scales. Variability within grids, both concentration and population distributions, can be significant. To improve exposure calculations that take into account the sub-grid variability, the uEMEP model has been applied in this study to provide suitable parameterisations of the sub-grid variability within an EMEP MSC-W grid. The uEMEP model is applied on a European domain at 250 × 250 m2. This provides roughly 7000 sub-grids within each EMEP MSC-W 0.3°×0.2° grid to analyse the sub-grid variability in exposure. The analysis derives an exposure correction factor (ECF) for each source sector in each EMEP MSC-W grid. This factor provides information on the impact of sub-grid resolution on the population weighted concentrations within each grid. The population weighted standard deviation is also calculated, providing information on sub-grid variability to derive frequency distributions and parameterised forms of the sub-grid variability suitable for health risk assessments. The impact of sub-grid variability is assessed for the pollutants PM2.5, NOX and NO2. For NO2, additional chemistry calculations are carried out on the NOX frequency distributions. We show that the parameterised form of the sub-grid variability provides close to equivalent results as the original uEMEP calculations. Exposure to PM2.5 increases by around 13% when the sub-grid variability is taken into account for a 2030 emission scenario. For NO2, this increase is around 31% for the same scenario. A similar increase is found in the health risk indicator ’population attributable fraction’ when no lower concentration threshold is used in the concentration–response function. When applying the World Health Organizations recommended threshold value we find that including the sub-grid variability increases the calculated population attributable fraction by 32% and 46% for PM2.5 and NO2 respectively. We conclude that sub-grid variability should be taken into account in future exposure calculations for regional scale modelling and impact assessments.