Future Air Quality in Danish cities

4. Regional Air Quality Levels

4.1 Scenario Emission data
4.2 Validation of DEM-Predictions for 1995
4.3 Future Regional Air Quality

Introduction

This chapter describes the future regional background levels as predicted by the Danish Eulerian Model (DEM). For the present study, the model predicts hourly air quality levels of NO_x, NO₂, O₃ and CO on a 50 x 50 km² grid. The model is not able to predict benzene levels. The levels represent average levels in the grid due to long-range transport of air pollution and the influence of local emission sources are not taking into account. The development in European emissions is based on EMEP data and ECE proposed developments in future national emission ceilings. The output of the DEM model is applied as input to the Urban Background Model (UBM) for prediction of urban background levels.

4.1 Scenario Emission data

The development in European emissions is based on EMEP data for 1990 and proposed national reductions in 2010 for all European countries under ECE. The national reductions are taking from an analysis by the International Institute for Applied Systems Analysis (IIASA) in Austria that carries out the preparatory work that leads to ECE protocols (IIASA 1999). In Autumn 1999, the ECE has proposed national emission ceilings for 2010 in a new multi-effect, multi-pollutant protocol on nitrogen oxides and related substances addressing photochemical pollution, acidification and eutrophication. This protocol is also referred to as the draft Protocol to Abate Acidification, Eutrophication and Ground-level Ozone which was approved in Gothenburg (Sweden) on 29 November - 3 December 1999 (ECE 1999).

The difference between the emissions by EMEP and IIASA applied in the project and the emission ceilings in the new ECE protocol for the reference year 1990 and the scenario year 2010 is given in Table 4.1. NO_x and VOCs are the main substances that form ozone in the atmosphere. The difference in reference data for 1990 is due to slightly difference emission data for some countries but also due to inclusion of emission from sea areas (ships) in the EMEP/IIASA data. There are also minor differences in the pro cent reduction assumed for the different countries. It is seen that the difference in emissions between the two scenarios is less than about 10 per cent. Therefore, all scenario studies have been carried out with the EMEP/IIASA data that was available to the project at an earlier stage than the new draft ECE protocol.

The national emissions in 1990 and the proposed emission ceilings in 2010 are given in Appendix 1 for all countries for EMEP/IIASA and for the new ECE protocol including the percentage reduction for each country.

For the reference year 1995 and the scenario years 2000 and 2005 it is assumed that the known reductions between 1990 and 2010 can be transferred to scenario years by linear interpolation. For the scenario years 2015 and 2020 the emissions for 2010 have been assumed since no data is available for these scenario years.

Table 4.1
The Difference Between the European Emissions (EMEP/IIASA) Applied in the Project and the New ECE Protocol Emissions (thousands of tonnes per year)

The predictions by the DEM model in 1995 have been validated against measurements at two regional background stations in the rural areas of the Greater Copenhagen Area: Frederiksborg and Lille Valby. A regional remote background stations in Jutland is also shown (Ulborg). Comparisons between modelled and measured values have been carried out for annual means, and seasonal and diurnal variation.

The differences in observed ozone levels in Denmark are minor since ozone formation is a large-scale phenomenon. The levels are slightly higher in Ulborg compared to Lille Valby and Frederiksborg since Ulborg is not influenced by ozone depletion due to local NO_x emissions. Levels are slightly lower at the forest station of Frederiksborg compared to the rural Lille Valby station probably due to a higher dry deposition of ozone on forest compared to agricultural land at Lille Valby (Jensen 1998).

Table 4.2
Comparisons Between Modelled and Measured Annual Means at Regional Stations in 1995 (ppb)

*Years with limited observations are given in brackets.

The DEM model overestimates O₃ levels in the regional background areas of Copenhagen compared to measurements. However, the model gives average predictions on a 50 x 50 km² grid and it is not able to reflect the influence of local NO_x emissions. The modelled levels are in better agreement with the remote station of Ulborg that is not influenced by local NO_x emissions.

The DEM model underestimates NO₂ levels for the regional Copenhagen area.

Few measurements are available for CO on Danish rural areas. Data from the Dutch monitoring programme shows that rural CO levels are about half of urban levels. The same ratio between urban and rural levels have been applied for the Danish rural background corresponding to 0.17 ppm in 1995 (Jensen 1998). The DEM model predicts about 0.5 ppm which is greatly overestimated. However, the DEM model has a crude estimation of CO emissions based on a ratio of VOC emissions. It is possible to obtain CO emission from EMEP for improving predictions but it has not been possible within the time frame of the present project. Therefore, the regional background annual level has been assumed to be 0.17 ppm in 1995. Measurements of CO at Frederiksvęrk in the background of Sealand during 1995 showed 0.33 ppm for a limited record during the year. This level seems to be overestimated since 0.34 ppm is measured in the urban background of Copenhagen. The monitor station of Frederiksvęrk is operated by the Greater Copenhagen Air Monitoring Unit.

Since the DEM model does not predict benzene the annual level of benzene is assumed to be the same ratio between CO and benzene as measured in the urban background of Copenhagen. The method for prediction of regional benzene is depending on urban background benzene described in greater details in Chapter 5.

The predicted seasonal variation of ozone is compared to measurements at Frederiksborg and Lille Valby in Figure 4.1 and 4.2. There is a good agreement between modelled and measured levels during spring and summer but the model overestimates levels in February and in autumn.

The overestimation in February reflects that there are predicted relatively few low values during this month. It does not reflect that the highest values are predicted during February, see Figure 4.3. Therefore, the monthly mean becomes relatively high. Within the time frame of the present project it was not possible to modify the DEM model to obtain better predictions for February. The average overestimation in February will have little impact on the estimation of the highest NO₂ concentrations at street level in Copenhagen since the highest ozone levels are not predicted in February.

Figure 4.1    [Look here]
Validation of seasonal variation of ozone at Frederiksborg

Figure 4.2    [Look here]
Validation of seasonal variation of ozone at Lille Valby

Figure 4.3    [Look here]
DEM predicted seasonal variation of hourly regional ozone levels for the regional background of Copenhagen in 1995

The predicted seasonal variation of NO₂ is compared to measurements at Frederiksborg and Lille Valby in Figure 4.4 and 4.5. There is generally a fair agreement although predicted levels are underestimated during spring and summer months and overestimated during November and December. As expected, the seasonal variation of predicted NO_x shows similar results as NO₂ just with higher levels.

Figure 4.4    [Look here]
Validation of seasonal variation of NO₂ at Frederiksborg.

Figure 4.5    [Look here]
Validation of seasonal variation of NO₂ regional background levels for Copenhagen at Lille Valby.

The predicted seasonal variation of CO showed almost no variation which is unlikely. Therefore, it is assumed that the seasonal variation is similar to the seasonal variation in the urban background in Copenhagen where the only available measurements are carried out.

Since the DEM model does not predict benzene the seasonal variation of benzene is assumed to be similar to CO.

The predicted diurnal variation of ozone is only compared to measurements at Frederiksborg in Figure 4.6 since the diurnal variation of ozone at Lille Valby is similar to Frederiksborg (Jensen 1998). There is generally a good agreement between modelled and measured levels although predicted levels show less relative difference between night and day time compared to measurements.

Figure 4.6    [Look here]
Validation of diurnal variation of regional background concentrations of ozone for Copenhagen at Frederiksborg.

The predicted diurnal variation of NO₂ is only compared to measurements at Lille Valby in Figure 4.7 since measurements are not available for Frederiksborg where only 24 hour samples are collected. NO₂ measurements at Lille Valby show a distinct diurnal variation with high levels in the morning and in the evening. The variation in measurements shows that Lille Valby is influenced by local traffic NO_x emissions from the nearby city of Roskilde and from the Copenhagen area. The predicted diurnal variation of NO₂ is smother since the DEM model is not able to take into account the influence of local emissions. As expected, the predicted diurnal variation of NO_x shows similar results although levels are slightly higher than NO₂.

Figure 4.7    [Look here]
Validation of diurnal variation of regional background concentrations of NO₂ for Copenhagen at Lille Valby.

The predicted diurnal variation of CO showed almost no variation which is unlikely. Therefore, it is assumed that the diurnal variation is similar to the diurnal variation in the urban background in Copenhagen where the only available measurements are carried out (Jensen 1999).

Since the DEM model does not predict benzene the diurnal variation of benzene is assumed to be similar to CO.

4.3 Future Regional Air Quality

Copenhagen Regional Background

Table 4.3 sums up the DEM model runs for the different scenario years for the future regional background in the rural Copenhagen area.

The table gives the predicted development in annual levels in ppb/ppm, µg/m³/mg/m³ and as an index. The index is defined as the level in the scenario years divided by the level in 1995.

Future predicted levels are also given for the regional background station Ll. Valby about 40 km outside Copenhagen using observed levels from 1995 as a base.

Since the DEM model does not predict benzene levels, the development for benzene is assumed to be similar to CO. CO and benzene in 1995 are not DEM predictions but estimated based on measurement in the urban background of Copenhagen. Levels of CO and benzene in a scenario year are estimated based on the 1995 level and the index determined by DEM calculations.

Table 4.3
DEM Predictions of Development in Future Regional Air Quality forthe Rural Copenhagen Area

From 1995 to 2010, NO_x and NO₂ levels are predicted to decrease about 40%. Ozone levels will only decrease by 2%, and CO and benzene by about 30% and 70%, respectively. Levels are the same from 2010-2020 since no information is available about future European emission ceilings beyond 2010.