Influence of Reducing the Highway Speed Limit to 80 km/h
PAUL SCHERRER INSTITUT
Influence of Reducing
the Highway Speed Limit to 80 km/h
on Ozone in Switzerland
Johannes Keller, Sebnem Andreani-Aksoyoglu,
Michel Tinguely, André Prévôt
Laboratory of Atmospheric Chemistry
Paul Scherrer Institut, CH-5232 Villigen PSI
This project was financially supported by the
Swiss Agency for the Environment, Forests and Landscape, SAEFL
(Bundesamt für Umwelt, Wald und Landschaft, BUWAL)
The exceptionally hot and dry summer in 2003 led to ozone levels exceeding the ambient
air quality standards in many parts of Europe. This preliminary study investigates how
some measures such as reducing the speed limit on highways would have effected the
ozone concentrations in Switzerland during such conditions. A 4-day period in August
2003 was studied by means of a 3-dimensional photochemical model (CAMx4) with 2
nested domains. The coarse domain covered Switzerland and a large part of the
neighbouring countries with a horizontal resolution of 27 km x 27 km. The resolution of the
fine domain was 9 km x 9 km covering all Switzerland and parts of the surrounding
countries including the agglomeration of Milano. Meteorological data such as 3dimensional wind fields, temperature, pressure, water vapour, vertical diffusivity and
clouds/rainfall were obtained from MM5 Meteorological Model. The emission inventory
was prepared by compiling European and Swiss anthropogenic emissions from various
sources. Reference year was 2000. Biogenic emissions (isoprene and monoterpenes
from trees, NO from soil) were calculated with temperature, irradiance dependent
algorithms using land use and meteorological data. Initial and boundary conditions were
adjusted from a European model (REM-3) output. Base case calculations were performed
for 4-7 August 2003. For a scenario case, speed limit on the highways in Switzerland was
set to 80 km/h and corresponding emission rates (calculated by INFRAS) were used in the
model simulations. The decrease in emissions was about 4% for NOx emissions and VOC
emissions were not significantly affected by the reduction of the speed limit. The effect of
reducing the speed limit to 80 km/h on ozone was very small. The decrease in afternoon
ozone levels was less than 1%. In the morning, an increase along the highways can be
identified. This is due to the VOC-limited ozone formation during morning hours. The
evaluated local and short-term emission reductions are not large enough to reduce ozone
concentrations effectively. More important for the ozone reductions in Switzerland will be
the long-term emission developments in Switzerland, the surrounding countries, and to
some extent even in the whole northern hemisphere. The influence of past and possible
future emission changes in Switzerland and in the near surroundings will be studied in a
follow-up study. Due to the strict time constraints in this project, the sensitivity of the model
to uncertainties in the model set-up and the emissions could not yet be tested in all
respects. These uncertainties should be addressed within the follow-up study. Future
model studies should also focus on aerosols.
Summer 2003 was an extraordinarily hot and dry season with ozone levels frequently
exceeding the legal thresholds. (BUWAL, 2003, 2004). The number of hours with ozone
levels > 120 µg m-3 was 2 to 3 times the values recorded in previous years. The most
frequent exceedances of the NABEL network were measured at the southern stations
Lugano and Magadino and at the elevated locations Laegern, Chaumont and Rigi. As an
emergency action to reduce ozone levels in southern Switzerland, a general speed limit of
80 km/h on the freeways A2 and A13 of canton Ticino was set from 12 to 17 August 2003.
This measure, however, was not based on scientific findings. Ozone levels on 12 and 13
August measured at Lugano and Magadino were not affected by the speed reduction. Due
to a cold front moving slowly towards the Alps, an increased cloud cover associated with
isolated precipitation was observed leading to a decrease of the ozone level on 14 August.
Hence it was not possible to assess the effectiveness of the speed limit.
In the view of a possibly increased frequency of hot summers, the following questions
Is a general speed limit for whole Switzerland during more than 1 week suitable for
reducing peak ozone?
What is the spatial pattern of ozone change?
What is the maximum decrease if emissions are reduced in Switzerland only?
What is the maximum decrease if emissions are reduced abroad as well?
Answers to these questions can be given only by applying air quality models. At PSI the
model package CAMx together with the meteorological pre-processor MM5 is in use. PSI
was asked by BUWAL to perform a preliminary study as a basis for political decisions in
summer 2004. For this study, the models have been adapted to 2 domains that include
Switzerland. A new emission inventory for 2000 (reference year) taking into account
changes of traffic emissions due to the reductions had to be set up in a very short time. In
Sections 2 and 3 the meteorology and the emission inventory are described. Section 4 is
devoted to the air quality model and the most important findings regarding the impact of
road traffic speed reductions.
The dispersion of air pollutants is predominantly controlled by the prevailing
meteorological conditions. On the one hand, the wind pattern determines the advective
transport. Heavily polluted air from high emission regions may increase the pollution level
in downwind areas. Conversely, air from large regions with low human activity is mostly
clean and may improve air quality. For instance, the low air quality in the Po basin
influences the pollutants’ level in southern Switzerland during south wind conditions,
whereas cleaner air is observed when the wind is blowing from the north (Weber and
Prevot, 2002). On the other hand, a stable atmosphere confines the pollutants to a small
air volume leading to high concentrations. A persistent inversion layer prevents the air
from being vertically mixed.
Eulerian air quality models compute the spatial and temporal distributions of atmospheric
species on a 3-dimensional rectangular grid. The meteorological data must be available
for the same coordinates. Gridded data can be derived from numerical forecasts or from
monitored data provided by the national weather services. The grid and the resolution of
this data, however, are usually different from those required by the air quality model.
To solve this problem, the air quality model is usually coupled with a meteorological
forecast model that can be tailored in order to match its output parameters and grids to the
needs of the air quality model. At the Laboratory of Atmospheric Chemistry (LAC) we
apply the meso-scale model MM5 (PSU and NCAR, 2004) as meteorological driver for the
air quality model CAMX. Up to 4 nested grids can be defined. The most suitable projection
for mid-latitude domains is the Lambert Conformal one, which transforms the earth’s
surface to a cone. This cone is defined by the longitude and the latitude of the coarse
domain’s centre and the “true” latitudes where the cone intersects the earth’s surface. The
position of the cone was selected in such a way that coordinate system is close to the
oblique Mercator grid used by the Federal Office of Topography, SWISSTOPO.
For the present study 2 nested domains were defined:
Coarse grid (domain 1):
centre longitude : 8 deg E
centre latitude : 47.5 deg N
grid cell size: 27 km x 27 km
number of cells: 35 x 29
Fine grid (domain 2):
grid cell size: 9 km x 9 km
number of cells: 72 x 54
The positions of these domains are shown in Figure 2.1 together with an example of the
The current version of MM5 uses 25 pressure levels. The thickness of a given layer varies
with surface altitude. At the surface pressure of 950 hPa the bottom layer thickness is
about 40 m.
MM5 is initialized by data of the “alpine model” (aLMo) of MeteoSwiss. aLMo is a nonhydrostatic model operational at MeteoSwiss since April 2001. It is based on the Local
Model (LM) developed in the frame of COSMO (COnsortium for Small scale MOdelling of
the five national weather services of Germany, Switzerland, Italy, Greece and Poland
(COSMO, 2002)). The model runs in a configuration of 385x325 grid points with a
horizontal grid mesh of 7 km. It has 45 levels up to 23 km a.s.l., whereof 19 layers being
within the first 2 km. Since January 2002, hourly aLMo outputs are available both as
forecasts and as analyses, the latter being assimilated with soundings and surface
measurements (wind only from stations below 100 m a.s.l.). For the present study
assimilated date were used as first guess data. The simulations were nudged towards
surface level measurements (ANETZ data), balloon soundings and aLMo upper level data
using the 4-dimensional data assimilation (FDDA) option of MM5 to obtain as realistic
meteorological fields as possible.
Weather conditions during the 4 – 6 August period were characterized by low pressure
gradients over Central Europe. A persistent anticyclone was located over the North Sea,
and weak cyclonic regions over the Alps. On 7 August the pressure field flattened. Noon
temperatures varied roughly between 30 and 35 deg C in the Swiss Plateau.
Due to the small pressure gradients, aLMo wind fields were mostly very irregular, the wind
speeds often being small. Surface wind velocities measured by the ANETZ stations often
exceeded the model data, especially at mountain locations. Moreover, the monitored wind
directions differ substantially from the aLMo values because of local topographic effects.
As mentioned above, MM5 is nudged by both aLMo and ANETZ data. Far from surface
stations, the MM5 output is similar to the aLMo analysis. Within the Swiss boundaries,
however, the two data sources compete, leading to a MM5 output that is often neither
close to the aLMo results nor match the experimental data in a satisfactory manner. It is
left to future investigations to figure out, under which conditions FDDA actually improves
the model results.
As examples, Figures 2.1 and 2.2 show the surface wind fields on 5 August 2003 at 12 h
UTC (13 h CET) for the domains 1 and 2, respectively.
Figure 2.1 : aLMo (black), ANETZ (red) and MM5 (green) wind fields of model domain 1 on
5 August 2003 at 12 h UTC (13 h CET). The boundaries of the CAMx domains 1 and 2 are
depicted in grey. Only every 9th aLMo vector and every 3rd MM5 vector is shown.
Figure 2.2 : aLMo (black), ANETZ (red) and MM5 (green) wind fields of model domain 2 on
August 5, 2003 at 12 h UTC (13 h CET). The boundary of the CAMx domain 2 is depicted
in grey. Only every 9th aLMo vector and every 3rd MM5 vector is shown.
Gridded emission rates for regular time intervals (usually 1 h) are required as input for the
air quality model CAMx. The organic compounds must be converted to species defined in
the Carbon Bond Mechanism (CBM-IV) (Gery et al., 1989). This mechanism is based on
chemical bonds. Examples are the single (paraffinic, PAR) and the double (olefinic, OLE)
bond. Molecules are split into a specified number of PARs and OLEs according to their
number of reactive bonds. For instance, n-butane is represented by 4 PAR, propene by 1
OLE + 1 PAR.
European Anthropogenic Emissions
Annual emissions and time functions for Europe were kindly provided by the Freie
Universitaet Berlin (FBU). This inventory was jointly developed with the Umweltbundesamt
(UBA) and The Netherlands Organisation for Applied Scientific Research (TNO) in the
frame of the CITY-DELTA project (Stern, 2003, Builtjes et al., 2002). The inventory
includes TSP, PM10, PM2.5, CH4, CO, NH3, NMVOC, NOx and SO2 for 12 source
categories following the SNAP classification (Table 3.1):
Table 3.1 SNAP categories of the European emissions
public power, cogeneration and district heating plants
commercial, institutional and residential combustion
industrial combustion and processes with combustion
non-combustion production processes
extraction and distribution of fossil fuels
road transport gasoline
road transport diesel
road transport evaporation
other mobile sources and machinery
waste treatment and disposal
The spatial resolution of the inventory is 0.25 deg latitude (~28 km) and 0.5 deg longitude
(~38 km at 47 deg lat).
Reference year is 1995. Factors to extrapolate the data to 2000 are given for each
country. Seasonal, weekly and diurnal variations are available as well. TNO provided
factors for each SNAP category to convert NMVOCs to CBM-IV species.
In Figure 3.1 the emission inventory of NOx is given.
Figure 3.1 : Annual NOx emissions for Europe (in kt NO2 / grid cell). Data source: FUB /
UBA / TNO (Stern, 2003)
Swiss Anthropogenic Emissions
Annual road traffic emissions of NOx, CO, NMVOC, toluene, benzene and xylene were
prepared by INFRAS. Data are split into link and zone emissions. The spatial resolution is
250m, the co-ordinates are based on the Swiss co-ordinate system. Reference year is
2000. An average diurnal variation was provided as well.
Data sets of 2 scenarios were provided:
reference scenario: speed limit on freeways according to current legislation.
general speed limit of 80 km/h on all freeways
Figure 3.2 shows the annual NOx emissions for the reference case. The relative difference
of the V80 and the reference scenario is given in Figure 3.3. It is obvious that there are
differences only on freeways because of the general speed limits of 80 km/h or less on
other roads. Due to the speed limit, NOx emissions decrease typically by 10 to 20 %, on
certain freeway sections by up to 35 %. Averaged over Switzerland, the reductions relative
to road traffic and relative to total NOx emissions are 7.7% and 4.3%, respectively.
Conversely, NMVOC emissions are not significantly affected by the speed reduction. This
different behaviour is also evident from measurements taken at the exhaust pipes of
individual cars. Speed dependent emission factors of NOx and VOC were found (Figure
3.4, Keller et al., 1995b). The emission rates of road traffic used in this study are already
based on new emission factors. Their speed dependences are similar to those given in the
Figure 3.2 : Annual NOx emissions of road traffic resampled to 1 km resolution
(t NO2 / km2). Data source: INFRAS
Figure 3.3 : Relative difference (ev80-eref)/eref of annual NOx emissions (%).
Industrial and residential NOx
Annual NOx emissions from residential activities, heating, industry, off-road traffic and
agriculture / forestry on a 200 m resolution were provided by Meteotest. Reference year is
2000. For Figure 3.5 the data set was resampled to 1km.
Residential and Industrial VOC
In the frame of the air quality project TRACT an emission inventory was developed for
September 1992 (Kunz et al., 1995). The resolution is 5 km. From this inventory residential
and industrial VOC emissions were extracted. The inventory includes 32 species
according to the chemical mechanism RADM. The species are grouped according to their
reactivity with OH radicals. For CAMx this data has to be converted to the CBM-IV
Using the time projections given in BUWAL, 1995 the emissions were converted to 2000.
As an example, Figure 3.6 depicts the daily NMVOC emissions for a weekday.
Ammonia is released mainly by manure, followed by waste treatment and road traffic.
Meteotest provided annual NH3 emissions for 2000 on a 1 km grid (Figure 3.7)
Figure 3.4 : Speed dependence of the emission factors for NOx and VOC for various
vehicle types (Keller et al., 1995b)
Figure 3.5 : Annual NOx emissions from residential industrial and off-road activities and
from agriculture / forestry, resampled to 1 km resolution (t NO2 / km2). Data source:
Figure 3.6 : Daily NMVOC emissions from residential and industrial activities (kg / grid cell
developed for TRACT. Resolution 5 km. Reference year 2000. Source of original data:
Meteotest (Kunz et al., 1995)
Figure 3.7 : Annual NH3 emissions (t NH3 / km2). Resolution 1km. Data source: Meteotest
The most abundant species are monoterpenes, which are released mainly by spruce and
fir. Less abundant, but much more reactive is isoprene emitted by oak trees and pasture.
NO emissions are caused by bacteriological decomposition in soils. Monoterpene and NO
emissions are temperature dependent, whereas the isoprene release is a function of
temperature and shortwave irradiance. In Andreani-Aksoyoglu and Keller, 1995 and Keller
et al., 1995a a methodology for the estimation of biogenic emissions is given. Gridded
biogenic emissions were calculated directly for the CAMx domains. Land use and
meteorological data are required for each domain.
Global land use data on a 30’’ grid were downloaded and converted by the MM5 preprocessors to the domains of interest. Inside the Swiss border the global data were
replaced by data of the “Arealstatistik” (100m resolution) issued by the Federal office of
Statistics (BFS, 1999) and the by forest data (1km resolution) taken from the
“Landesforstinventar” (Mahrer and Vollenweider, 1983). The latter includes the land cover
of 10 different tree species, in particular spruce, fir and oak. About 24% of the Swiss area
is covered with forests, 71% thereof are conifers. Norway spruce and fir are the most
abundant species (67 and 20 % of the conifers). Conversely, oak trees contribute only 8%
to deciduous trees. In the global land use data there is only a split of the forested areas
into conifers and deciduous trees. Due to the lack of information, the Swiss proportions of
Norway spruce, fir and oak were assumed to be valid for the foreign countries as well.
Gridded temperature and shortwave irradiance data were extracted from the MM5 output.
Conversion of the emissions to the CAMx grids
As mentioned in Sec. 2, the resolutions of the 2 model domains are currently 27 and 9 km.
The anthropogenic emissions of a given CAMx grid cell are calculated by computing the
geographic (or Swiss) co-ordinates of the 4 corners and the totals of the European (or
Swiss) emission rates within the respective polygons. For obvious reasons the biogenic
emissions do not need to be converted.
Figures 3.8 and 3.9 show the gridded emissions of NO and PAR of the CAMx coarse
domain 1 resampled to 9 km resolution. In the area of the fine domain 2 the data are
replaced by the domain 2 data.
Figure 3.8 : NO emissions in CAMx domain 1 and 2 calculated for 5 August 2003, 12:00
UTC (14:00 CEST) (kmol / h). The data were resampled to a common grid cell size of 9
km. Reference year of the original emission data is 2000. The boundaries of the CAMx
domains 1 and 2 are depicted in grey.
Figure 3.9 : PAR emissions in CAMx domain 1 and 2 calculated for 5 August 2003, 12:00
UTC (14:00 CEST) (kg / h). The data were resampled to a common grid cell size of 9 km.
Reference year of the original emission data is 2000.
4 Photochemical Modelling
The Comprehensive Air Quality Model with Extensions (CAMx) is an Eulerian
photochemical dispersion model that allows for an integrated “one-atmosphere“
assessment of gaseous and particulate air pollution over many scales ranging from urban
to regional (ENVIRON, 2003). CAMx simulates the emission, dispersion, chemical
reactions, and removal of pollutants in the lower troposphere by solving the pollutant
continuity equation for each chemical species on a system of nested three-dimensional
grids. CAMx incorporates two-way grid nesting, which means that pollutant concentration
information propagates into and out of all grid nests. CAMx carries concentrations at the
centre of each grid cell volume, representing the average concentration over the entire
cell. Horizontal advection is performed using the advection solvers of Bott, 1989 or the
PPM method of Colella and Woodward, 1984.
There are five gas-phase mechanisms supported in CAMx4 (version CAMx v4.03). These
are four different versions of Carbon Bond Mechanism (CBM-IV, Gery et al., 1989) - one
with reactive chlorine chemistry, two with different isoprene chemistry, one with the
extensions for aerosol modelling - and SAPRC99 chemical mechanism (Carter, 2000). In
this study, CBM-IV mechanism with the extensions for aerosol modelling (mechanism 4)
was used. It includes condensable organic gas species and a second olefin species to
account for the biogenic olefins (representing terpenes). Table 4.1 shows the precursor
gas species for condensable organic gases and the secondary organic aerosol products.
Photolysis rates are derived for each grid cell assuming clear sky conditions as a function
of five parameters: solar zenith angle, altitude, total ozone column, surface albedo and
atmospheric turbidity. Since the photolysis rates are significantly affected by the presence
of clouds, a cloud input file is required in case of cloudy conditions. The model provides an
option to adjust photolysis rates for the presence of clouds using the approach developed
for the Regional Acid Deposition Model (RADM, Chang et al., 1987). This approach
provides a realistic impact on photolysis rates by accounting for cloud optical depth.
Besides reducing photolysis below clouds, it enhances photolytic rates above clouds due
In CAMx4, aerosol processes are linked to the CBM-IV gas-phase mechanism. The gasphase photochemistry forms aerosol precursors via the OH initiated oxidation of SO2 to
sulphate, production of nitric acid, and formation of condensable organic gases. The CBMIV precursor TOL (mostly toluene) produces two different condensable species CG1 and
CG2 (Table 4.1). The same condensable species are also produced from the oxidation of
CBM-IV precursor XYL (mostly xylene). In the model, there are two more condensable
organic gases (CG3 and CG4) produced by the oxidation of cresol and terpenes.
The aerosol precursors are supplied to the aerosol chemistry module, which performs the
-aqueous sulphate and nitrate formation in resolved cloud water using RADM aqueous
chemistry algorithm (Chang et al., 1987).
-partitioning of condensable organic gases (CG1-CG4) to secondary organic aerosols
(SOA1-SOA4) to form a condensed organic solution phase using a semi-volatile
equilibrium scheme called SOAP (Strader et al., 1998).
-partitioning of inorganic aerosol constituents (sulphate, nitrate, ammonium, sodium,
and chloride) between the gas and particle phases using the ISORROPIA
thermodynamic module (Nenes et al., 1998).
Particle sizes are static but vary by chemical constituent. The aerosol species calculated
by CAMx4 include sulphate, nitrate, ammonium, organic carbon, sodium, chlorine, and
primary inert PM (particulate matter). In this study, the particle size range for the aerosol
species was chosen as 0.04 - 2.5 m. In other versions of the model, several size modes
may be used.
For this study no primary emissions were considered.
Table 4.1. Representation of secondary aerosols in CAMx4. PAR : paraffinic carbon bond,
OLE : olefinic carbon bond (anthropogenic), TOL: toluene, XYL: xylene, CRES: cresol,
OLE2: olefinic bond (biogenic).
CBM-IV VOC Precursor
Condensible Gas Species
Secondary Aerosol Species
4.1.2 Pollutant Removal
Trace gases and small particles are removed from the atmosphere via deposition to the
surface. Dry deposition refers to the direct sedimentation and/or diffusion of material to
various terrestrial surfaces and uptake into biota. Dry deposition of gases is based on the
resistance model of Wesely, 1989. Surface deposition of particles occurs via diffusion,
impaction and/or gravitational settling. The resistance approach used in UAM-AERO
(Kumar et al., 1996) has been adopted in CAMx4.
Wet deposition refers to the uptake of material via chemical absorption (gases) or
nucleation/impaction (particles) into cloud water, and subsequent transfer to the Earth’s
surface by precipitation. The efficiency of wet and dry deposition processes to remove
pollutants from the air depends on the physical and chemical properties of the pollutants,
local meteorological conditions, the type of surface on which they are being deposited, and
on the frequency, duration, and intensity of precipitation events. The wet scavenging
model implemented in CAMx4, calculates the following processes: wet scavenging of
gases within and below precipitating clouds, wet scavenging of gases dissolved in cloud
water, and in-cloud aerosols, and wet scavenging of dry particles.
4.1.3 Input and Output Files
CAMx requires inputs to describe photochemical conditions, surface characteristics, initial
and boundary conditions, emission rates, and various meteorological fields over the entire
modelling domain (Table 4.2). Preparing this information requires several
preprocessing/premodelling steps to translate raw data to final input files for CAMx4. The
model produces hourly average concentration output files containing entire 3-dimensional
fields of user-selected species. There are also output files for deposition parameters, and
4.1.4 Model Setup
The size of the CAMx4 coarse domain is 35 grid cells in the east-west direction and 29
grid cells in the north-south direction with a resolution of 27 km x 27 km. The fine domain
contains 68 and 50 grid cells in the east-west and north-south direction, respectively, with
a resolution of 9 km x 9 km. There are 10 layers in a terrain-following coordinate system,
the first being 30 m above ground. The coordinate system used in this study was Lambert
Conic Conformal system. The model top was set at about 4000 m above ground.
Meteorological input files were prepared using the MM5 mesoscale Model as described in
Section 2. Initial and boundary conditions were calculated using the REM-3 European
model output data for the same time period provided by the Meteorological Institute of
Freie Universität Berlin. An example of the REM-3 output for ozone on 7 August is shown
in Figure 4.1. REM-3 domain covers almost whole Europe with a resolution of 0.5 x 0.25
degree. This model has another coordinate system (geographic) and a different vertical
structure than CAMx4. The thickness of the lowest layer is 20 m. Above the lowest layer,
there are 3 layers with varying heights. One layer is always above the mixing layer and the
other two are within the mixing layer. Therefore, the output of REM-3 model had to be
resampled to the vertical structure of CAMx4 before being used for boundary and initial
Ozone column densities were extracted from TOMS data. Photolysis rates were calculated
using the preprocessors provided together with the model. Simulations started on 4
August 2003 at 0000 UTC and ended on 7 August at 2400 UTC. Calculations were
performed using the base case and scenario emissions with a highway speed limit as
described in section 3.
Table 4.2. Data requirements of CAMx
3-dimensional gridded fields
(supplied by a meteorological model)
-horizontal wind components
gridded initial concentrations
(obtained from either measured ambient gridded boundary concentrations
data or from other models)
time/space constant top concentrations
gridded sources (anthropogenic and
(supplied by an emission model)
elevated point sources
gridded land use/surface cover
gridded surface UV albedo codes
atmospheric radiative properties
(ozone column from TOMS data, photolysis
rates from radiative model)
-gridded haze opacity codes
-gridded ozone column codes
-photolysis rates lookup table
chemical information for the simulation and
Figure 4.1: Ozone mixing ratio (ppb) in the lowest layer (0-20 m) calculated by REM-3
model (Freie Universität Berlin), 7 August 2003, 13:00-14:00 UTC.
4.2.1 Base case
Only the part of fine domain covering Switzerland will be discussed in this section. The
highest ozone mixing ratios in the lowest layer were predicted generally in the afternoon
between 13:00 and 16:00 UTC. As seen in Figures 4.2-4.5, ozone levels on 4 August are
relatively lower than the other days in Switzerland. Concentrations increased on the 5th
especially in southern Switzerland, around Lugano under the influence of southerly winds
with polluted Po Basin air (see Figures 4.6-4.9 for the wind fields). The wind speed on 5
August was stronger than on the other days. The wind direction north of the Alps, on the
other hand, changed from west wind to north wind during the studied period.
Figure 4.10 provides an example of the diurnal evolution of the vertical ozone profile at a
location about 30 kilometres west of Zurich. The highest concentrations are usually found
in the afternoon up to 1000 to 1500 meters above ground. Concentrations in the upper
layers increased between 4 August and 7 August. Ozone is most depleted during the
night in the lowest 300 meters above ground. However, the depletion is not as strong as
often found at low altitude stations. Boundary layer parameterization was not the focus of
this project but it needs to be optimized in the future to avoid too strong mixing at night.
Model results for the lowest layer were compared with the measurements at some NABEL
stations (Figure 4.11). Resolution of the model (9 km x 9 km) and the location of the
measurements have to be kept in mind when comparing model results and
measurements. Peak ozone concentrations increased with time during the first 3 days.
Model results show a similar trend as the measurements north of the Alps. NOx mixing
ratios measured at urban stations such as Lugano and Zurich are higher than the
predicted ones because of the rather short distance of the measurement station to the
emissions but also due to too high mixing during the night. Higher night-time ozone
concentrations in the model are also found in more rural areas like Payerne and Tänikon
(not shown). Comparison of Ox concentrations (O3 + NO2) provide a better evaluation of
the model by eliminating the local NO titration effects. In rural areas such as Laegern and
at the elevated location Rigi, agreement between model predictions and measurements is
satisfactory. Within the time constraints of the project the model was not evaluated in
4.2.2 Scenario case
In the scenario case, emissions were adjusted to a speed limit of 80 km/h on the freeways.
Changes in peak ozone concentrations in the lowest layer due to reduced speed limit are
shown as percentages in Figures 4.12 - 4.15. In Figures, blue color indicates a decrease
in ozone mixing ratios in percentage, and red color shows the increase.
Ozone concentrations increased in the morning along the highways due to the fact that
ozone production is usually VOC-limited during the morning hours. On the other hand,
ozone decreased in the afternoon when ozone formation becomes more NOx-limited. This
is the case for every day of the studied period.
The decrease in ozone levels due to speed reduction however, is very small, lower than
1%. Keeping in mind that NOx emissions decreased only by about 4% and VOC emissions
were not significantly affected by speed reduction, the small influence on ozone
concentrations is not surprising. Other studies with similar measures indicated the need for
emission reductions larger than 50% to achieve considerable decreases of ozone
Figure 4.2: Predicted O3 mixing ratios (ppb) on 4 August 2003, at 13:00-14:00 UTC.
Figure 4.3: Predicted O3 mixing ratios (ppb) on 5 August 2003, at 13:00-14:00 UTC.
Figure 4.4: Predicted O3 mixing ratios (ppb) on 6 August 2003, at 13:00-14:00 UTC.
Figure 4.5: Predicted O3 mixing ratios (ppb) on 7 August 2003, at 13:00-14:00 UTC.
Figure 4.6: Modelled wind fields in the lowest layer (0-30 m) on 4 August 2003, at 13:0014:00 UTC shown over the topography (m asl).
Figure 4.7: Modelled wind fields in the lowest layer (0-30 m) on 5 August 2003, at 13:0014:00 UTC shown over the topography (m asl).
Figure 4.8: Modelled wind fields in the lowest layer (0-30 m) on 6 August 2003, at 13:0014:00 UTC shown over the topography (m asl).
Figure 4.9: Modelled wind fields in the lowest layer (0-30 m) on 7 August 2003, at 13:0014:00 UTC shown over the topography (m asl).
4 August 2003
5 August 2003
6 August 2003
7 August 2003
Figure 4.10: Vertical distribution of O3 (ppb) as a function of time (UTC) in one grid cell
about 30 km west of Zurich (x=99 km, y=135 km) for the period 4 - 7 August 2003.
Figure 4.11: Diurnal variation of measured (plus symbol) and predicted (solid line) mixing
ratios (ppb) for O3, NOx and Ox (O3+NO2) during 4-7 August 2003 in Laegern, Lugano, Rigi
Figure 4.12: Predicted change in O3 mixing ratios (%) due to speed limit, on 4 August
2003, at 06:00-07:00 UTC (top), 13:00-14:00 UTC (down). Blue color shows the decrease,
red color shows the increase.
Figure 4.13: Predicted change in O3 mixing ratios (%) due to speed limit, on 5 August
2003, at 06:00-07:00 UTC (top), 13:00-14:00 UTC (down). Blue color shows the decrease,
red color shows the increase.
Figure 4.14: Predicted change in O3 mixing ratios (%) due to speed limit, on 6 August
2003, at 06:00-07:00 UTC (top), 13:00-14:00 UTC (down). Blue color shows the decrease,
red color shows the increase.
Figure 4.15: Predicted change in O3 mixing ratios (%) due to speed limit, on 7 August
2003, at 06:00-07:00 UTC (top), 13:00-14:00 UTC (down). Blue color shows the decrease,
red color shows the increase.
5 Discussion and Conlusions
The influence of the traffic speed reductions to maximum 80 km/h in Switzerland on ozone
concentrations is not very high, typically lower than 1%. This can be backed up by a backof-the-envelope calculation. In the beginning of the 90s, results from the POLLUMET
(Pollution and Meteorology) project in Switzerland suggested that around 30% of the
ozone concentrations could be controlled by Swiss emissions during summer smog
situations (BUWAL, 1996). Due to lower emissions nowadays, a lower amount, around
25%, can be estimated to be controllable today. The Swiss NOx emissions in the traffic
scenario was calculated to be around 4% lower than in the base case. If the response of
the ozone production is linear to the NOx emissions, 4% of 25% yielding 1% of the ozone
concentrations could be reduced. This is true for NOx-limited conditions. The response can
be opposite if the ozone production is VOC-limited. This could be shown for the morning
hours when the ozone production is usually VOC-limited and the ozone increased in the
traffic scenario near the highways. Overall, the traffic speed reduction alone is not enough
to significantly reduce the ozone levels. However it should be noted that the NOx
concentrations and the aerosol concentrations also decrease in such a traffic scenario.
This leads to a relief in addition to the small ozone reduction during these summer smog
Larger scale (Central Europe) emission decreases would yield better results than just
decreases within Switzerland, but first model runs indicate that also in this case larger
emission reductions than speed limitations on highways are necessary to make a
detectable difference in ozone. A study carried out in Germany with the emissions based
on the year 1990, showed that temporary and locally restrictive measures such as speed
limit, a ban on non-cat motor vehicles or the burning of reformulated fuels are not very
effective in decreasing high ozone levels, because attainable emission reduction is small
(http://www.umweltdaten.de/ozone). For example, setting the speed limit in BerlinBrandenburg area to 80 km/h for passenger cars and to 60 km/h for trucks on the
freeways and to 60 km/h for all vehicles on the other roads led to 10% reduction in NOx
emissions and 2% reduction in VOC emissions. The decrease in ozone concentrations
was only up to 4%, and there were small increases in the city center. These results led to
the conclusion that temporary, local restrictive measures are not very effective, if the
emissions reduction potential is below 50%.
More important for the ozone reductions in Switzerland will be the long-term emission
developments in Switzerland, the surrounding countries, and to some extent even in the
whole northern hemisphere. The influence of past and possible future emission changes in
Switzerland and in the near surroundings will be investigated in a follow-up study.
Due to the strict time constraints of this project, the sensitivity of the model to uncertainties
in the model set-up and the emissions could not yet be tested in all respects. These
uncertainties should be addressed within the follow-up study. This includes
the use of a longer time period to get more representative results,
the assessment of the influence of the model grid resolution (vertical and
the sensitivity to different VOC emissions (the quality of the current VOC emission
inventory for industry is not satisfactory because it relies on the spatial variations
of the TRACT emission inventory of 1991),
the assessment of the sensitivity of different boundary layer parameterizations (at
the moment, the mixing during the night is obviously too strong).
Future model studies should also focus on aerosols. Much less is known about the limiting
factors of the aerosol formation compared to the ozone production. For this task, up-todate SO2 emissions, size resolved particulate matter emissions, and up-to-date VOC
emissions would be necessary.
Bundesamt fuer Umwelt, Wald und Landschaft
California Grid Model
Comprehensive Air Quality Model with EXtensions
Carbon Bond Mechanism, version 4
Freie Universitaet Berlin
Meso-scale Model 5
non-methane volatile organic compounds
Paul Scherrer Institut
Regional Euleriam Model with 3 different chemistry scemes
Selected Nomenclature form Air Pollution
The Netherlands Organisation for Applied Scientific Research
Total Ozone Mapping Spectrometer
Andreani-Aksoyoglu S., Keller J., (1995) Estimates of monoterpene and isoprene
emissions from the forests in Switzerland. J. Atmospheric Chemistry 20 71-87.
BFS, (1999) GEOSTAT Benuetzerhandbuch, Bern.
Bott A., (1989) A positive definite advection scheme obtained by nonlinear renormalization
of the advective fluxes. Monthly Weather Review 117 (5), 1006–1016.
Builtjes P., van Loon M., Schaap M., Teeuwisse S., Visschedijk A. J. H., Bloos J. P.,
(2002) The development of an emission data base over Europe and further contributions
of TNO-MEP. Freie Universitaet Berlin / Institut fuer Meteorologie und Troposphaerischer
BUWAL, (1995) Vom Menschen verursachte Luftschadstoffemissionen in der Schweiz von
1900 bis 2010. Schriftenreihe Umwelt Nr.256, BUWAL, Bern.
BUWAL, (1996) Troposphaerisches Ozon. Aktuelle Forschungsergebnisse und ihre
Konsequenzen fuer die Luftreinhaltung. BUWAL, Bern.
BUWAL, (2003) Ozonsommer 2003 im Vergleich mit 1993-2002, http://www.umweltschweiz.ch/buwal/de/fachgebiete/fg_luft/luftbelastung/publikat/nabel/index.html.
BUWAL, (2004) Luftbelastung 2003, http://www.umweltschweiz.ch/buwal/de/fachgebiete/fg_luft/luftbelastung/publikat/nabel/index.html.
Carter W. P. L., (2000) Programs and files implementing the SAPRC-99 mechanism and
its associates emissions processing procedures for Models-3 and other regional models,
Chang J. S., Brost R. A., Isaksen I. S. A., Madronich S., Middleton P., Stockwell W. R.,
Walcek C. J., (1987) A three-dimensional eulerian acid deposition model : Physical
concepts and formulation. JOURNAL OF GEOPHYSICAL RESEARCH 92 14681-14700.
Colella P., Woodward P. R., (1984) The piecewice parabolic method (PPM) for gasdynamical simulations. Journal of Computational Physics 54 174-201.
COSMO, (2002) Newsletter No. 2. Deutscher Wetterdienst, Offenbach (D).
ENVIRON, (2003) User's Guide, Comprehensive Air Quality Model with Extensions
(CAMx). Version 4.00. ENVIRON International Corporation, Novato.
Gery M. W., Whitten G. Z., Killus J. P., Dodge M. C., (1989) A photochemical kinetics
mechanism for urban and regional scale computer modeling. JOURNAL OF
GEOPHYSICAL RESEARCH 94 925-956.
Keller J., Andreani-Aksoyoglu S., Joss U., (1995a) Inventory of natural emissions in
Switzerland. In Air Pollution III, Air Pollution Engineering and Management, 339-346.
Computational Mechanics Publications, Southhampton.
Keller M., Evéquoz R., Rellstab J., Kessler H., (1995b) Luftschadstoffemissionen des
Strassenverkehrs 1950-2010. Schriftenreihe Umwelt Nr.255, BUWAL, Bern.
Kumar N., Lurmann F. W., Wexler A. S., Pandis S. N., Seinfeld J. H., (1996) Development
and application of a three dimensional aerosol model. A&WMA Specialty Conference on
Computing in Environmental Resource Management, Research Triangle Park, NC (USA),
Kunz S., Kuenzle T., Rihm B., Schneider A., Schlaepfer K., (1995) TRACT
Emissionsmodell Schweiz. Schlussbericht Maerz 1995. METEOTEST / Carbotech, Bern.
Mahrer F., Vollenweider C., (1983) Landesforstinventar LFI, Eidgenössische
Forschungsanstalt für Wald, Schnee und Landschaft (WSL), Birmensdorf.
Nenes A., Pandis S. N., Pilinis C., (1998) A new thermodynamic equilibrium model for
multiphase multicomponent inorganic aerosols. Aquatic Geochemistry 4 (1), 123-152.
PSU, NCAR, (2004) MM5 Version 3 Tutorial Presentations,
Stern R., (2003) Erstellung einer europaweiten Emissionsdatenbasis mit Bezugsjahr 1995
und die Erarbeitung von Emissionsszenarien für die grossraeumige
Ausbreitungsrechnungen mit REM/CALGRID. Freie Universitaet Berlin / Institut fuer
Meteorologie und Troposphaerischer Umweltforschung, Berlin.
Strader R., Gurciullo C., Pandis S. N., Kumar N., Lurmann F. W., (1998) Development of
gas-phase chemistry, secondary organic aerosol, and aqueous-phase chemistry modules
for PM modeling. STI-97510-1822-FR, Sonoma Technology, Inc., Petaluma, CA,,
Umweltbundesamt, (2004) Programme of Control Concepts and Measures for Ozone,
Weber R. O., Prevot A. S. H., (2002) Climatology of ozone transport from the free
troposphere into the boundary layer south of the Alps during North Foehn. JOURNAL OF
GEOPHYSICAL RESEARCH 107 (D3),
Wesely M. L., (1989) Parameterization of surface resistances to gaseous dry deposition in
regional-scale numerical models. Atmospheric Environment 23 1293-1304.
We are grateful to the following people and institutions for providing weather, emission and
air quality data within very restricted time limits: F.Schubiger and C. Voisard (MeteoSwiss)
for aLMo analysis data, R.Stern (FUB), A.Graff (UBA) and M. van Loon (TNO) for
European emissions, R. Zbinden, M. Keller and J. Heldstab (INFRAS) for Swiss road
traffic data, Th. Kuenzle and B.Rihm (METEOTEST) for updated Swiss NOx and NH3 data
and for the TRACT emission inventory, and J. Flemming (FUB) for REM-3 model output
data. We thank also R.Weber (BUWAL) for the fruitful co-operation during the project.