The EU Air Quality Directive on air pollution levels, mirrored in Swedish legislation, has far-reaching consequences for Swedish city administrations. Of special importance is the PM10 legislation, as the Swedish EPA estimates that approximately 80% of Swedish cities will have to assess PM10 concentrations.
To handle these problems, two new Internet-based modelling tools have been developed for use by all Swedish municipalities to assess their air pollution levels and how they compare with the EU Air Quality Directive targets. The first is SIMAIRroad (Gidhagen et al., 2009), which is related to traffic emissions, and the second is SIMAIRrwc, rwc standing for residential wood combustion (Omstedt et al., 2009).
Sources of air pollution
Sweden's air quality is strongly influenced by its geographic location and climate in Northern Europe. In the winter, the temperature is often below zero degrees and rain and snow are common, causing slippery roads and a need for space heating.
For anti-skid treatment, studded tyres and sand are frequently used, creating road wear particles; accordingly, approximately 80–90% of all PM10 emissions (Ketzel et al., 2007) originates from non-exhaust emissions.
Another local source of particles is residential wood combustion, a common primary or supplementary space heating source. Increased use of small-scale biofuel is one alternative being considered given the phasing out of fossil fuels. However, high-emitting old wood stoves can have negative impacts on air quality.
Long-range transport from emissions outside Sweden is another important source of air pollution. Local authorities can therefore only improve part of the air quality, the part related to local sources. An important step is therefore to separate local, urban and regional contributions to total air pollution concentrations.
SIMAIRroad and SIMAIRrwc both use the same principles: a coupled model system of different models for the local, urban and regional geographical scales, using the best available emission data, but presented in a very simplified way. This is done using a combination of pre-calculated concentrations from larger-scale models and outputs from fast-computing local models.
Regional background concentrations
On the regional scale, modelling is performed using the MATCH multi-scale atmospheric transport and chemistry model (Robertson et al., 1999; Andersson et al., 2007). The MATCH model is driven by the HIRLAM weather forecast model and uses a 44 x 44 km2 grid over Europe. Emissions are taken from the EMEP 50 x 50 km2 inventors (http://www.emep.int). Sea salt particles are included using a method developed by Foltescu et al. (2005).
The MATCH model does not yet include secondary organic aerosols. Therefore, PM10 measurements are also included using two-dimensional variational data assimilated with PM10 data from a few regional stations in Scandinavia and background fields from the MATCH model.
Urban background concentrations
Urban background concentrations are simulated on a 1 x 1 km2 grid using emission data from the Swedish Database for Emissions to the Environment (SMED; http://www.smed.se).
Two different model approaches are used for the dispersion. For ground-level sources, such as traffic exhaust and small-scale wood combustion, an adjoint modelling approach is used. This model is based on determining an influence area upwind from a receptor point, within which all emissions are aggregated to the final concentration. Each of the cells in the urban 1 x 1 km2 grid constitutes a receptor point. The dispersion of stack emissions is treated in a separate Gaussian point source model (Omstedt, 1988).
The meteorological data used are from the routine operating Mesan system (Häggmark et al., 2000), based on the optimal interpolation technique. The background field is a six-hour forecast from the HIRLAM model with 22-km horizontal resolution. All available measurements from synoptic and automatic stations, radars and satellites are analyzed on an 11 x 11 km2 grid with a 3-h resolution.
Road and traffic information
The Swedish National Road Database (NVDB) is a national road and vehicle database, containing up-to-date information about, for example, road-coordinates, functional road class, speed limit, number of lanes and road width. A parallel database includes measured traffic volumes on state-owned roads, while traffic volumes on municipal roads have been simulated with a traffic demand model. Emission factors for the exhaust part of PM10 emissions are calculated by ARTEMIS (http://www.trl.co.uk/artemis). A semi-empirical model is used for the non-exhaust part (Omstedt et al., 2005).
Data from Swedish chimney sweeps
Detailed local data for residential wood combustion are included at the user’s request for different cities. This is done in cooperation with the Swedish Association of Master Chimney Sweeps.
Local-scale models used by SIMAIRroad
SIMAIRroad uses two different dispersion models. If the road of interest is surrounded by buildings along one or both sides, the OSPM street canyon model (Berkowicz, 2000) is used. If the road is not surrounded by buildings or obstacles (i.e., open road conditions), then a simplified Gaussian line source model for ‘infinite line sources’ is used (Gidhagen et al., 2004).
Local-scale models used by SIMAIRrwc
SIMAIRrwc uses two different local-scale dispersion models, one for point sources and one for traffic sources; both models are Gaussian. The point source model is based on the OML model (Berkowicz et al., 1986; Omstedt, 1988). The traffic model is a line source model for ‘finite line sources’.
SIMAIRroad has been validated against measurements of PM10, NO2 and benzene for more than 20 streets in Sweden (Andersson and Omstedt, 2009). The model reproduces both the data averages and variations (expressed as the coefficient of variation). SIMAIRroad is able to account for the main features of day-to-day mean PM10 variability, especially the peak PM10 concentrations in late winter and early spring commonly experienced in Nordic countries where studded tyres and sand are used as anti-skid treatments.
In accordance with the EU Directive on ambient air quality, the acceptable uncertainty of modelling estimation is defined as the maximum deviation between the measured and calculated concentration levels, over the period considered and at the limit value in consideration, without taking into account the timing of the events. The average annual modelling uncertainty for PM10 is for instance defined as ±50%.
As shown in the figure to the right, SIMAIRroad is able to calculate mean value statistics, 90 percentile and 98 percentile daily mean values of PM10 that are well within the ±50% the EU requires model estimates of yearly mean PM10 values. In the comparison, most of the data, i.e., 70% of the data given, is within ±25%, which is the quality objective for fixed measurements in accordance with the EU Directive.
The correlation between modelled and measured concentrations of PM10 is also strong for areas with residential wood combustion, as shown in the figure below. SIMAIRrwc describes the main features of day-to-day mean PM10 variability and reproduces both the data average and variations reasonable well.
National research studies
SIMAIR has recently been used in some national studies. In Omstedt et al. (2011) the model was used for the evaluation of possible Air Quality improvements due to the reduction of studded tyres use in Sweden. In Andersson et al. (2008) and Omstedt et al. (2010 and 2012) SIMAIR was also applied for evaluation of the future prospects to achieve the standards according to the EU Air Quality Directive as well as the Swedish national environmental goals.
Development and improvements
Recently, a new application of SIMAIR has been developed; SIMAIR Scenario tool. This application will be an important support for authorities and decision-makers for the visualization of emission scenarios regarding Air Quality, health effects and external costs.
A new emission model for non-exhaust road traffic induced particles is currently tested and evaluated within SIMAIR. The model has been developed within the Nordic project NORTRIP.
References and further reading
See attached document, SIMAIR publication list, in the column on the right hand side.