Aarhus University Seal / Aarhus Universitets segl

No. 57: Source Contribution to harmful air pollution in Copenhagen

Jensen, S.S., Brandt, J., Ketzel, M., Plejdrup, M. 2013: Kildebidrag til sundhedsskadelig  luftforurening i København. Aarhus Universitet, DCE –Nationalt Center for Miljø og Energi, 85 s. - Videnskabelig rapport fra DCE - Nationalt Center for Miljø og Energi nr. 57. http://www.dmu.dk/Pub/SR57.pdf


Background and objectives

As a basis for regulation of air pollution, it is important to have information about how much each source contributes to the concentration of air pollutants, population exposure and the related health impacts.

Source apportionment provides information about the potential for improvement of air quality due to emission reductions of a given source.

Air pollution in Copenhagen is influenced by local sources in the city and sources outside the city in Denmark and Europe. Thus, there is a regional contribution of air pollution that is transported over long distances which can’t be controlled locally. Source apportionment provides information on the size of the regional contribution of the different air pollutants, and how much the city contributes, and how much traffic in the streets contributes.

The aim of the project is to prepare a source apportionment for nitrogen oxides (NOx), nitrogen dioxide (NO2), PM10 (particles less than 10 microns) and PM2.5 (particles less than 2.5 microns) and number of particles for urban background concentrations and street concentrations in Copenhagen to estimate how much the different sources contribute. The source apportionment will show how much the different sources contribute to the concentration, respectively, for urban background and street concentrations (in micrograms per cubic meter and in percentage). The focus of the project is on air pollutants that are harmful to public health.

The study

The study focuses on NOx, NO2, PM10 and PM2.5 and number of particles. NOx and NO2 are good indicators of combustion where traffic is a dominant source. There are health-related air quality limit values for NO2 where the air quality limit as an annual average is 40 µg/m3 which is exceeded at the air quality measuring station at H.C. Andersens Boulevard in Copenhagen. PM10 and PM2.5 have a variety of sources and long range transport of air pollution plays a major role. There are health-related air quality limit values for PM10 of 40 µg/m3 as annual mean value and for PM2.5 of 25 µg/m3 as an annual mean which are not exceeded at measurement stations in Copenhagen. Although the limits are not exceeded there are great health impacts associated with PM10 and PM2.5. There is no air quality limit values for number of particles, and health impacts as well as source inventory for the number of particles are uncertain due to lack of knowledge.

Urban background concentrations are determined by the sources in the city - in this case the Municipalities of Copenhagen and Frederiksberg, and the regional contribution. Air pollution in a street is determined by traffic sources in the street as well as the contribution of urban background concentrations. These different contributions are calculated with air quality models based on emission data, meteorological data and topological data. Regional air pollution is calculated with DEHM (Danish Eulerian Hemispheric Model). Urban background pollution is calculated with UBM (Urban Background Model) and street pollution is calculated with OSPM (Operational Street Pollution Model). For Denmark, the SPREAD emission model (Spatial High Resolution Emissions to Air Distribution Model) is used which has a spatial resolution of 1km x 1km. Source apportionment calculations are carried out for 2010, the latest year for emissions data. The UBM model takes into account the emissions that are within 20 km from the receptor points. Urban background concentrations are calculated for all grid cells of 1kmx1km covering the Municipalities of Copenhagen and Frederiksberg, and source contributions are calculated. An average has been calculated to represent the source contributions to urban background pollution.

The contribution to air pollution levels of NOx, NO2, PM10 and PM2.5 is calculated by turning off each of the sources in the model and the contribution of each emission category is calculated as the difference between the model results when all sources are included and model results where each source is turned off. This approach takes into account non-linear atmospheric chemistry in the calculations which are important for the calculation of NO2 as it involves reactions between NO, NO2 and O3.

DEHM does not currently include secondary organic particles (SOA). Therefore, the contribution of SOA is added based on analysis of measurements. Furthermore, a contribution is added for ‘unknown mass’ to make the total mass fit with measurements for PM10 and PM2.5. Unknown mass is, among others, likely to include water bounded in the particles.
In the model system the source apportionment for the urban background concentrations is divided into the following contributions - a regional contribution and a local contribution.

The regional contribution represents the air pollution that blows over the outer municipal boundary of the Municipality of Copenhagen. The regional contribution is calculated with the DEHM model (and for particles also including SOA and unknown mass) with contributions from neighbouring municipalities. Neighbouring municipalities are defined as the area within a distance of up to 20 km away from Copenhagen's outer municipal boundary. Furthermore, the contribution from ship traffic in Øresund (sound between Zealand and Sweden) is included where the local contribution from Øresund is defined at a distance up to 20 km away from the municipal boundary. The contribution from distances greater than 20 km is calculated with the DEHM model and the local contribution with the UBM model. The local emission sources are broken down by SNAP codes 1-9 in the Municipalities of Copenhagen and Frederiksberg Municipality on a 1kmx1km grid.

In the current model setup, DEHM results can’t be compared directly with e.g. concentration levels from the regional background measurement station Lille Valby/Risø outside Copenhagen, as DEHM delivers upstream concentrations to UBM which subsequently calculates the contribution from local sources within a distance of up to 20 kilometers of a given calculation point. DEHM modeled concentrations are about half the measured values of the measurement station Lille Valby/Risø.

Source apportionment for the above represents the year 2010.

Source apportionment for streets has been completed for 99 busy streets in Copenhagen and Frederiksberg, where the source apportionment is subdivided into vehicle categories for 2010. For a single street – H.C. Andersens Boulevard - a more detailed inventory of sources of NOx has been given where vehicle categories are further subdivided in Euro emission classes and types of fuel (gasoline/diesel). Here, the source apportionment is for 2015 based on a previous study.

In addition the relative distribution of the health-related external costs of the different sources is provided with the EVA-system (Economic Valuation of Air pollution). These calculations represent source apportionment for the regional background concentrations. The EVA-system is an integrated model system that applies the regional background model DEHM, dose-response relationships and pricing of health effects to calculate the external costs of air pollution and to estimate how the external costs are allocated to the different emission sectors. The calculations represent 2008 based on a previous study.

Main Conclusions and Project Results

 Main Contributions

The regional contribution to NOx and NO2 are relatively small and NOx and NO2 are dominated by local sources i.e. sources in the city and traffic sources in the streets. The opposite is true for PM10 and PM2.5 where the regional contribution is dominant, and local sources mean less. The regional contribution originates from sources in Denmark and Europe outside the outer municipal boundary of Copenhagen. Measured concentrations of the number of particles follows the pattern of NOx and NO2, that is, the regional contribution is relatively small and air pollution with ultrafine particles is dominated by local sources as sources in the city and traffic sources in the streets.

Source Contributions to Urban Background Concentrations

NOx pollution dominated by local sources

As expected, the concentration of NOx exceeds NO2 as NOx includes NO and NO2. The regional modeled DEHM contribution to NOx pollution in the urban background is relatively small (about 5 µg/m3 or 25% of urban background).

The contribution from neighbouring municipalities is about 4-5 µg/m3 or 25-27% of urban background and the same size as road traffic within the City of Copenhagen and Frederiksberg, and the contribution from neighbouring communities are also dominated by road traffic.

Previous assessments of ship traffic in Danish waters and its contribution to NO2 urban background pollution in Copenhagen with the DEHM model estimated about 2 µg/m3 in 2007 (Olesen et al. 2009; Jensen et al. 2010). The ship traffic in Øresund contributes with 0.6-0.8 µg/m3 (about 4% of urban background) while the rest is from regional air pollution. Emission control requirements under the International Maritime Organization (IMO) for NOx and expected increase in ship traffic is likely to offset one another leading to a contribution from ship traffic that is roughly the same in 2007 and 2020 in absolute terms.

The contribution that blows over the outer municipal boundary of Copenhagen includes the modeled DEHM contribution, the contribution from neighbouring municipalities and the contribution from ship traffic in Øresund. Together these contributions constitute around 55% for NOx and 60% for NO2 of urban background pollution.

NOx pollution is dominated by local sources in and just around Copenhagen that contribute to 75% of urban background concentrations, where the main source is road traffic with around 5 µg/m3 or 26-28% of urban background. Local sources include here also neighbouring municipalities (mainly traffic), ship traffic in Øresund and all sources in Copenhagen within the outer municipal boundary.

Other local NOx sources in Copenhagen of any significance is combined heat and power plants and district heating plants, including waste incineration plants (1-2 µg/m3 or 8-10% of urban background), industry - including non-road mobile machinery (0.4 - 0.5 µg/m3 or 2-3%) and non-industrial combustion, for example, combustion in households and commercial and service, dominated by wood smoke (0.2-0.3 µg/m3 or about 2%).

Ozone pollution dominated by European sources

Ozone is formed from emissions of nitrogen and hydrocarbon compounds and also depending on sunlight and temperature. Danish sources contribute to ozone formation on a large scale together with European sources while the Danish sources actually contribute to reduction of ozone levels in Denmark. For example, DEHM calculates the regional background level for ozone to about 61 µg/m3. This level is modified by NOx emissions from local sources which converts ozone to NO2 in reaction with NO. The result is that the ozone concentration in the urban background air ends up being about 52 µg/m3.

Particle pollution dominated by regional contribution

The regional concentration contribution is dominant for the pollutants PM10 and PM2.5. The regional contribution is determined by emission sources in Denmark and Europe outside Copenhagen outer municipal boundary. As expected, the regional concentration contribution of PM10 is slightly higher than for PM2.5 as PM10 also contains the mass of particles with a diameter between 2.5 and 10 microns. The regional contribution consists of the contributions from DEHM, SOA and unknown mass. The regional contributions for PM10 and PM2.5 are 17 µg/m3 and 11 µg/m3 of the urban background, respectively, compared to the measured urban background of 19 µg/m3 and 13 µg/m3, respectively. The regional contribution for PM10 and PM2.5 constitutes 89% and 86% of urban background, respectively.

Neighbouring municipalities also contribute with a smaller contribution of about 0.5 µg/m3 (about 3%) while ship traffic in Øresund within 20 km from the municipal boundary contributes very little. The contribution that blows over the outer municipal boundary of the Municipality of Copenhagen includes the modeled DEHM contribution plus SOA and unknown mass, neighbouring municipalities and ship traffic in Øresund. Together these contributions are about 92% of PM10 and 89% of PM2.5 of urban background concentrations in Copenhagen. The local sources of Copenhagen and Frederiksberg contribute about 1.5 µg/m3 (about 8% of urban background) and is predominantly non-industrial combustion (mainly wood stoves) and road traffic where non-industrial combustion contributes about twice as much as road traffic. Non-industrial combustion contributes about 0.9 µg/m3 and road traffic about 0.3-0.4 µg/m3.

Emissions from non-industrial combustion are dominated by emissions from combustion sources. The main sources of emissions of particles from households are wood stoves, wood boilers and wood pellet boilers. The uncertainty is high on the estimation of emissions from especially stoves and boilers. The uncertainty is related to both the level of emissions and the geographic distribution. The Danish Energy Agency estimates wood consumption per municipality. This means that emissions from residential wood combustion are evenly distributed throughout the municipal area. The distribution is assumed to lead to an overestimation of emissions from residential households in the City of Copenhagen and Frederiksberg. Other sources such as combined heat and power plants and district heating plants, including waste incineration plants (SNAP 1), and industry - including non-road mobile machinery (SNAP 0808) contribute very little. Note that ship traffic outside Øresund is included in the regional DEHM modeled contribution.

The contribution from different emissions sources to urban background concentrations is shown in Table 3.1.

Source Contribution to Health-related External Costs in Denmark

In Table 3.2, the contribution to health-related external costs in Denmark is given as a percentage for the Danish emission sectors calculated with the EVA- system. The calculations are for 2000 and 2008.

The dominant contributors from Danish emissions sectors in 2008 are in the following order 1) Agriculture, 2) non-industrial combustion, and 3) road transport.

The external cost is dominated by the number of years of life lost which in turn is dominated by the particle mass. Agriculture is one of the main contributors due to the contribution to atmospheric formation of secondary inorganic particles (ammonium sulfate and ammonium nitrate) due to agricultural emissions of ammonia.

There is currently not enough knowledge to separate the health effects from the inorganic secondary particles and from primary particles such as soot. In these calculations it is assumed that health effects are the same for all types of particles. However, a sensitivity analysis was performed in which primary particles are assumed to be more harmful to health than secondary particles and this analysis shows that external costs for stoves and road traffic will increase compared to emissions from agriculture under this assumption.

Non-industrial combustion has increased significantly from 2000 to 2008 due to increase in number of wood stoves and wood boilers and their use.

Source Apportionment for Street Contribution

Street concentrations include the contribution from the urban background as well as the contribution from street emissions for the specific street. The urban background contribution to street concentrations is larger for PM10 and PM2.5 than for NOx and NO2.

The average vehicle distribution for the 99 streets examined in Copenhagen is: 80% passenger cars, 16% vans and 4% for trucks and buses. As vehicle distribution is different from street to street there will also be differences in the source apportionment from street to street.

As expected, the street contribution is higher for NOx than for NO2 since NOx is the sum of NO and NO2. The modeled street contribution for NOx is up to about 65 µg/m3 and for NO2 up to about 30 µg/m3. In the assessment of the contribution of NOx it should be kept in mind that modeled calculations underestimate the street contribution to NOx as the street contribution for H.C. Andersens Boulevard is measured to about 110 µg/m3. The size of the street contribution mainly depends on annual average daily traffic but also on vehicle distribution, travel speed and street geometry.

Passenger cars contribute the most to the street contribution of NOx and NO2. Vans, trucks and buses contribute about equally but it varies from street to street depending on vehicle distribution especially for heavy-duty vehicles. The heavy-duty vehicles (trucks and buses) typically contribute more than vans, and trucks typically contribute more than buses. Despite the fact that trucks and buses account for only about 4% of traffic they contribute relatively much as emission factors for trucks and buses are about 10 times higher than for passenger cars and vans. If it is assumed that the number of particles is proportional to NOx emissions then the distribution of number of particles is the same as described above.

As expected, the street contribution is higher for PM10 than for PM2.5 as PM10 includes a higher proportion of non-exhaust which is road, tire and brake wear and re-suspension. The street contribution for PM10 is up to about 9 µg/m3 and for PM2.5 up to about 4 µg/m3. The street contribution for the number of particles is up to about 12,000 particles per cm3. The size of the street contribution mainly depends on annual average daily traffic but also on vehicle distribution, travel speed and street geometry.

Passenger cars contribute the most to the street contribution for PM10 and PM2.5 followed by vans. Passenger cars and vans contribute on average, respectively, 62% and 22% for PM10, and 57% and 28% for PM2.5. The heavy-duty traffic generally contributes less than vans, and buses generally contribute less than trucks. Trucks contribute on average 11% for PM10 and 10% for PM2.5 and buses contribute on average 7% for both PM10 and PM2.5. The reason that passenger cars and vans contribute relatively more to PM10 than to NOx compared to heavy-duty traffic is due to the non-exhaust contribution. Despite the fact that non-exhaust contribution is less for passenger cars and vans per vehicle compared to heavy-duty vehicles, the contribution from passenger cars and vans is relatively higher for PM10 and PM2.5 due to the large number of passenger cars and vans. Furthermore, the existing low emission zone reduced PM exhaust from the heavy-duty traffic.

The average source apportionment for the street contribution for the different air pollutants for the 99 busy streets is shown in Table 3.3.

Source apportionment for NOx for H.C. Andersens Boulevard by Euro emission classes and by fuel type (gasoline and diesel) in 2015 showed that about 84% of NOx emissions come from diesel vehicles, and about 29% come from older vehicles up to and including Euro 3.

Discussion of uncertainties

The source apportionment is largely based on air quality modeling. There are uncertainties associated with the models' description of the physical and chemical processes, and uncertainties related to the input data used especially emission data. A general way to assess the uncertainties in the models is to compare model results with measurements. Comparison between model and measurements in urban background show good agreement for NOx and NO2, and for PM10 and PM2.5 but only due correction for SOA and unknown mass. The model underestimates street concentrations especially for NOx but shows good correlation for NO2.

Currently, DEHM does not include secondary organic particles (SOA). Therefore, the contribution of SOA is added based on analysis of measurements. SOA is estimated from measured organic matter (OM). A large part of this is supposed to be SOA, and in the present report it is assumed that all OM is SOA. Furthermore, unknown mass is added as a residual contribution to fit model results to measurements of the total mass of PM10 and PM2.5. Additionally, sea salt is not presently included in the model and therefore part of the unknown mass is sea salt. A substantial part of unknown mass is supposed to be water bound to the particles. SOA and unknown mass are around 150% of the modeled DEHM contribution. If these contributions were included in the emissions sectors it might give a different allocation between emissions sectors. On the other hand, the source apportionment for the regional background is based on anthropogenic emissions, and SOA and unknown mass are primarily from natural emissions. Therefore, the relative distribution between emissions sectors may not change even if SOA and unknown mass were encountered for in the model.

There are also uncertainties on the calculation of the health-related external costs modeled with the EVA-system which is also based on DEHM. It is assumed that the external costs is the same for Copenhagen as for Denmark as a whole which is an approximation. An important assumption is that the primary and secondary particles are equally harmful to health. A sensitivity analysis was carried out assuming primary particles to be 1.3 times more harmful on average and secondary particles 0.7 times less harmful on average. This sensitivity analysis shows that non-industrial combustion (primarily wood stoves and boilers) which is directly emitted particles have greater weight while the share of agriculture becomes less as the contribution here is mainly related to the secondary particles, otherwise there are only minor changes in the contribution from the different emissions sectors. The EVA-system calculates external costs on the basis of regional concentrations. However, increased spatial resolution might increase the contributions from local urban sources such as road traffic and stoves.

Urban background concentrations are modeled with UBM and include contributions from DEHM (and also SOA and unknown mass) as well as contributions from neighbouring municipalities, ship traffic in Øresund and all emissions sources within the Municipalities of Copenhagen and Frederiksberg.

In particular, uncertainties are related to the amount of non-industrial combustion which is dominated by wood stoves and boilers. Non-industrial combustion accounts for 5-7% of urban background pollution for particle mass.

There are also uncertainties for other mobile sources (construction machinery, trains etc.) However, the contribution from non-road mobile machinery is very modest with 0.2-0.3% for particle mass. There is an ongoing investigation by the Danish Environmental Protection Agency to further quantify the contribution from this source. A sub-report on trains shows that the contribution from trains is very limited (Miljøstyrelsen, 2013).

There is considerable uncertainty associated with the contribution of particle numbers in urban background since number of particles is not included in the model at present, and no attempt was made to establish a source apportionment for number of particles. The main sources of number of particles in the urban background are assumed to be traffic and non-industrial combustion (wood stoves and boilers). It is not possible at present to qualify this further due to lack of knowledge.

The source apportionment for the street contribution modeled with OSPM is considered to be relatively well estimated since it is only determined by the traffic source, and there is fairly good knowledge about road emissions. However, this does not apply to the number of particles where the distribution on vehicle categories is based on a good correlation between the measured street contribution of NOx and particle numbers. However, the distribution on individual vehicle categories is more uncertain as it is assumed that particle numbers are proportional to calculated NOx emissions for the individual vehicle categories. A sensitivity analysis was carried out assuming that particle numbers are proportional to modeled particulate emissions, giving a somewhat different picture of the relative contributions of the different vehicle categories.