1. Urban-scale air quality modelling with AURORA
Karen Van de Vel, Koen De Ridder, Filip Lefebre, Clemens Mensink, Jo Vliegen
Flemish Institute for Technological Research (VITO), Mol, Belgium
Karen Van de Vel from VITO.
Give presentation on behalf of Koen de Ridder.
Next 15 minutes, I will show results from the urban-scale AQ model AURORA

2. Monitoring ambient air quality for health impact assessment.
EXPOSURE
FORECASTS
SCENARIOS
Why use modelling in order to asses air quality ?
Extract from a WHO publication, shows in which cases ‘monitoring’ is best way to assess problems, and in which cases ‘modelling’.
For 3 types of studies, modelling highly relevant, this type of studies can be performed with models such as the AURORA model
Monitoring ambient air quality for health impact assessment.
WHO regional publications, European series, no. 85, 1999.

3. AURORA (VITO) 3-D chemistry-transport model
different scales (continental  urban scale)
meteorology from ARPS model
emissions ~ traffic, industry, households, …
advanced gas and aerosol chemistry
output: hourly (typically) 1-km gridded pollutant concentrations (O3, PM10, NO2, benzene, …)
street-box sub-model
I shortly summarize some features of the AURORA model : 3D
Can be applied on different scales ranging from … to …
It employs meteorological fields from the ARPS meteorological model
Emissions of pollutants from various sectors
Chemistry
Output
When zooming in down to the street level : a street-box model can be plugged in.

4. projects involving AURORA
BUGS (EU FP5, )
Paris (F), Flanders (B), Ruhr-area (G)
(aBB, )
Budapest (H)
GSE-PROMOTE (ESA GMES, )
various European cities
(FIT, )
Yangzhou (China)
AMFIQ (EU FP6 – ESA DRAGON, )
Shenyang (China)
Environment & Health (AMINAL, ) >> poster session
Flanders (B)
SANAIR (aBB/AMINAL, )
Santiago de Chile (Chile)
EU COST 728 ( )
AURORA is a model which can be used very flexibly all over the world.
Overview of recent and present projects carried out.
Studies performed for Flemish Government, also EU and ESA projects
Area of application is indicated : regional, urban scale
In the following i will show results obtained in these projects

5. benzene simulations (Antwerp)
We start in Antwerp, the biggest city in the North of Belgium.
Left :
City of antwerp, 20x20 km showing the harbour (river Schelde), center and near surroundings of Antwerp.
Map of benzene concentrations, comparison of modelling results with measurements of Life project (MacBeth), typical error of 25%.
Right :
zoom of the left picture, 10x10 km.
It shows the results of a scenario study : what happens if we put a roof on the ring road of Antwerp ie. Change benzene emissions.
Change in benzene concentrations (in %), ranging from a decrease with 25% very locally to -5%
RMSE  0.69 µg m-3 (~ 25 %)
% concentration change (roof on Ring)

6. vegetation versus O3 (Antwerp)
vegetation decreased by 50 %
vegetation increased by 50 %
Again simulation for the greater surroundings of Antwerp.
What happens if we change the vegetation in the center of the city of Antwerp. Change in land use has an effect on the air temperature which in turn affects the formation of e.g. ozone.
Left : vegetation decreased by 50%. We get an increase in O3 concentrations in the city centre by at most 4%.
Right : vegetation increased by 50%. We get a decrease in 03 concentrations n the city centre by at most 4%
Remark : 4% is not a lot, but major part of the population lives in the area where the changes are the largest. So when looking to exposure, great effect.
(colour code ~ % O3 concentration change)

7. detailed terrain parameters (Paris)
Some images reflecting the degree of detail we use to run AURORA simulations.
Input parameters : land use (from CORINE), terrain height, vegetation cover from SPOT-VEGETATION.

8. model vs obs O3 (Ruhr area)
simulation
observation
Bottrop
Essen
Case study of the Ruhr area.
Simulated vs measured O3 concentrations, for 2 measurement stations, for a period of 3 weeks
Bottrop (traffic station, NO2 emissions) : titration effects, decrease of O3 concentrations to zero during night
Essen (background station), O3 levels stay high
After 13 days decrease in O3 concentrations well captured by model, passage of front resulting in cooler temperature and more clouds
Air quality simulations performed with the AURORA model for the reference situation were validated by comparing model results with available observations from two stations in the area that measured ozone.
Even though the simulations overestimate the ozone peak concentrations somewhat, the diurnal cycle as well as the behaviour of the model over the entire three-week period is rather satisfactory.
In particular the difference of night time concentrations between the two locations, which is due to the titration effect (reduction of ozone by traffic-related NO emissions) caused by the more intense traffic at Bottrop, is well captured by the model, meaning that the spatial distribution of traffic emissions as well as the chemical processes accounted for in the model perform correctly. Also, the abrupt decrease of ozone concentrations towards the end of the simulation period, caused by a frontal passage, was relatively well simulated.

9. impact urban sprawl (Ruhr area)
O3 concentration change
land use of reference state 
land use of sprawl scenario
Area of 200x200 km (cities of Bocchum, Essen, )
What if people migrate from the city centre towards the more rural areas = urban sprawl
People travel greater distances (e.g. to go to work, to shops, …), hence higher NO2 concentrations, resulting in lower O3 concentrations in original city centers
But upwind of simulation domain (upper left corner) higher O3 concentrations
Compact city forms are associated with minimal consumption of land and energy, hence they are often promoted as being the more sustainable thus preferred mode of urban development.
In this context, numerical simulations were performed to evaluate the effect of urban sprawl on air quality and associated human exposure.
Working on a highly urbanised area in the German Ruhrgebiet, models dealing with satellite data processing, traffic flows, pollutant emission and atmospheric dispersion were applied, under conditions representative of the urbanised area as it is today.
An urban sprawl scenario was defined and implemented using spatial modelling techniques, relocating 10 % of the urban population to the cities’ periphery.
The land use of the reference state : green (= vegetated areas, pasture, forests), red : current urban surface cover, including industrial areas.
The land use of the urban sprawl scenario : dark red = locations of residential land use added in the urban sprawl scenario
The resulting updated land cover and population density maps of the area were then used as input for the traffic and atmospheric dispersion simulations,
Total traffic kilometres, associated emissions, and domain-average pollutant concentrations increased by approximately 10 to 15 %.
However, despite the overall concentration increases, an analysis of human exposure to atmospheric pollution revealed that the urban-sprawl scenario leads to lower rather than higher exposure values in the case of PM10, while for ozone the exposure remains almost unchanged.
The associated simulated percentage change of ground-level ozone is shown in the right figure. Owing to the dominant south-easterly wind direction during this episode, increased ozone values are simulated in a plume extending at the north-west side of the urban agglomeration. The titration effect, on the other hand, which is the consequence of increased traffic emissions, slightly depresses ozone concentrations in the central portion of the domain, i.e., where the highest population densities occur.
Impact analysis :
In terms of land consumption the urban-sprawl scenario induces an increase of built-up areas by 75 %. Owing to the increased average distance between people’s homes and working places, as well as by increased number of car trips, the total amount of daily passenger traffic kilometres increases by 17 %. As a direct consequence of the increased number of vehicle kilometres, the related overall energy consumption and associated emissions increase. In particular, CO2 emissions increased by 12.7 %, thus having an adverse effect on global warming. Likewise, emissions of pollutants that are harmful to health (PM10, O3 precursors, ...) increase, and so did the domain-average concentrations of these pollutants. However, the exposure of people to PM10 decreased. Even though this may seem surprising initially, there is a straightforward explanation. Indeed, notwithstanding the increase of domain-average PM10 concentrations, in both scenarios a significant share of the population is relocated to areas outside the dense conurbation, to areas where pollutant concentrations are lower, hence exposure is reduced. Ground-level ozone being a non-local pollutant (unlike primary particulate matter), its effect on human exposure is not so clear. On the one hand the ozone concentrations in the urban plume increase, but the effects of that are partly compensated by a decrease (owing to the ozone titration effect) in the urbanized portions of the area. As a result, population exposure to ozone increases slightly, by 0.3 %.

10. PAH version of AURORA (Flanders)
Polycyclic Aromatic Hydrocarbons :
gas – particle partitioning
degradation
formation of secondary PAHs
Case study of the Ruhr area. Simulated vs measured O3 concentrations, 17days
Bottrop (traffic station, NO2 emissions) : titration effects, decrease of O3 concentrations to zero during night
Essen (background station), O3 levels stay high
After 13 days decrease in O3 concentrations well captured by model, passage of front resulting in cooler temperature and more clouds

11. PAH version of AURORA (Flanders)
2NPyrene (p), Aarschot
B(b)F (p), Borgerhout
Case study of the Ruhr area. Simulated vs measured O3 concentrations, 17days
Bottrop (traffic station, NO2 emissions) : titration effects, decrease of O3 concentrations to zero during night
Essen (background station), O3 levels stay high
After 13 days decrease in O3 concentrations well captured by model, passage of front resulting in cooler temperature and more clouds

12. under development…
data assimilation (AirBase in-situ data, SCIAMACHY & OMI satellite data)
mapping of urban air quality indicators (“exceedance of air quality limit values in cities” – EEA)
extension to urban micro-scale, focus on ultra-fine particles (~ traffic-related)