Assessment of urban air quality

− ∆ [PM_2.5] CLE−>UFR

∆ [PM2.5] UFR−>CLE

non−DE grids DE grids

∆ emis ratio

Figure 5.13: Differences in annual mean concentrations of ambient PM2.5 resulting from a change of German SO2, NOx, VOC, NH3 and primary PM2.5 emissions from UFR to CLE with all other European emissions at UFR versus a change of the German emissions from CLE to UFR with all other European emissions at CLE

5.3.4 Assessment of urban air quality

The underlying epidemiological studies detected and quantified relationships between mortality and characteristic background concentrations of fine particulate matter. Thus, in order to accurately apply their findings for a health impact assessment, it will be necessary to target the characteristic exposure for the urban and rural population in Europe. The EMEP Eulerian model provides estimates of regional air quality with a 50*50 km resolution, which can be considered as representative for rural air sheds. However, the majority of the European population lives in urban areas, and the health impact assessment needs to reflect the higher concentrations of particles in urban air.

5.3.4.1 The City-Delta project

To acquire deeper insight into the factors that deteriorate air quality in cities, IIASA together with the Institute for Environment and Sustainability of the Joint Research Centre (Ispra), MET.NO, EUROTRAC-2 and CONCAWE, has initiated the City-Delta model intercomparison (http://rea.ei.jrc.it/netshare/thunis/citydelta/). The aim of this exercise is to conduct a systematic comparison between regional scale and local scale dispersion models to identify and quantify the factors that lead to systematic differences between air pollution in urban background air and rural background concentrations.

City-Delta explores

• systematic differences (deltas) between rural and urban background AQ,

• how these deltas depend on urban emissions and other factors,

• how these deltas vary across cities, and

• how these deltas vary across models.

Based on the findings of City-Delta, functional relationships will be developed that allow the estimation of urban levels of pollution (PM2.5) as a function of rural background concentrations and

local factors. The City-Delta analysis addresses the response of health-relevant metrics of pollution exposure (i.e., long-term concentrations with or without thresholds) towards changes in local and regional precursor emissions, including the formation of secondary aerosols. This enables the generic analysis of urban air quality for a large number of European cities based on information available in the RAINS model framework.

City-Delta provides harmonized emission inventories, meteorological conditions and observational data and explores the changes in air quality for seven emission control scenarios in eight European cities. It is important that the participating models apply different spatial resolutions, i.e., some of the models operate with a 50 km resolution and are thus directly comparable to the EMEP model (which is also participating), while others use finer resolution in the urban model domains. Thus, it is now possible to directly compare results from European scale models with finer resolved urban dispersion models and to search for systematic differences.

In its first phase, 20 modelling teams participated in City-Delta and produced for PM in total 143 model runs, each of them covering the full 12-months period of 1999 (Table 5.4). The focus of the City-Delta project on health-relevant air quality metrics gave a major impetus to many participating models that used to operate for selected short-term episodes only to extend their modelling capacities to full 12 months calculations. As a result, there is now a large ensemble of dispersion models available in Europe that can be used for health impact assessment according to the WHO recommended methodology.

The first phase of City-Delta clearly revealed serious deficiencies in the present scientific ability to accurately model observed PM mass concentrations. All chemistry models that simulate the fate of the various chemical components of PM result in serious underestimates of observed PM mass concentrations. However, models agree to a large extent on the fate of anthropogenic primary particles and secondary inorganic aerosols. It should be noted that also the lack of quality controlled monitoring data, especially for PM2.5, puts a serious limit to the validation of the model results.

Table 5.4: Models participating in City-Delta (O .. ozone calculations, P .. particle calculations) City Berlin Copenhagen Katowice London Milan Paris Prague Resolution 5

While models do not perform well in reproducing the observed total PM mass, the limited available monitoring data strengthen their credibility for modelling anthropogenic fractions of PM, at least from primary emissions and from secondary inorganic aerosols. Based on this assertion, City-Delta 1 compared the responses of regional and urban scale dispersion models towards changes in regional precursor emissions (Figure 5.14). Models also agree that for urban areas there seems to be in the year

2010 less practical scope for further reductions of PM than the expected improvements between 1999 and 2010.

Figure 5.14: Summary of City-Delta Phase 1 model responses of urban PM10 levels (annual mean concentrations) towards changes in the various precursor emissions for Berlin, Milan and Paris. The left panel shows responses of regional scale models (50 km resolution), while the right panel shows responses of fine scale models. Four “deltas” are presented: (1) the change between emissions of 1999 and CLE 2010; (2) the change from CLE 2010 to MFR NOx; (3) the change from CLE 2010 emissions to MFR for VOC, and (4) the change from CLE 2010 to MFR for NOx and VOC. Different models are displayed in different colours.

As a general finding it can be shown that PM concentrations modelled for cities are systematically higher than those in rural background air. The increase depends on the size fraction considered (there is a larger difference for PM10 than for PM2.5) and on the emission densities within the cities. All models participating in Phase 1 of City-Delta agree that a large fraction of fine particles present in urban background air (especially PM2.5) originate from outside the cities, and in many cases from long-range transport sources several hundred kilometres away. Thus, the boundary conditions of urban model calculation dominate the results also within the cities.

This finding, which is also supported by monitoring data (Putaud et al., 2002), is important for the integrated assessment of European emission control strategies, because it underlines the relevance of Europe-wide emission control efforts also for the improvement of air quality in the cities.

Phase 1 of City-Delta has identified a number of factors that lead to differences in model results.

Some of them are related to different model concepts and are therefore interesting from a scientific standpoint, but also a number of mistakes in input data (especially emission inventories), data handling or model routines have been identified. Many of the originally large discrepancies could be reduced by thorough quality control. Despite these improvements, the numerical results from City-Delta 1 for particulate matter were not considered sufficiently robust to serve as a basis for serious policy analysis. Thus, a second phase of City –Delta has been launched, which focuses specifically on particulate matter. This phase is expected to be completed in spring 2004 and will provide input to the RAINS modelling of urban air.