Atmospheric Environment 39 (2005) 2759–2773 Analysis and evaluation of selected local-scale PM 10 air pollution episodes in four European cities: Helsinki, London, Milan and Oslo Jaakko Kukkonen a, , Mia Pohjola a , Ranjeet S Sokhi b , Lakhu Luhana b , Nutthida Kitwiroon b , Lia Fragkou b , Minna Rantama¨ki a , Erik Berge c , Viel Ødegaard d , Leiv Ha˚vard Slørdal e , Bruce Denby e , Sandro Finardi f a Air Quality Research, Finnish Meteorological Institute, Sahaajankatu 20 E, Helsinki, Finland b University of Hertfordshire, ASRG, Science and Technology Centre, Hatfield, UK c Kjeller Vindteknikk, P.O.Box 122, 2007 Kjeller, Norway d Norwegian Meteorological Institute, Postboks 43, Oslo, Norway e Norwegian Institute for Air Research, Postboks 100, Norway f ARIANET, via Gilino 9, 20128 Milano, Italy Received 18 February 2004; received in revised form 8 September 2004; accepted 20 September 2004 Abstract We have analysed in detail four selected episodes involving substantially high concentrations of PM 10 that occurred in Oslo on 4–10 January 2003, in Helsinki on 3–14 April 2002, in London on 18–27 February 2003 and in Milan on 14–19 December 1998. We have also utilised a more extensive dataset containing relevant information regarding 21 episodes from seven cities in six countries. The four episodes analysed in detail were recently occurring cases that were at least partly caused by various local emission sources. In particular, we have addressed the evolution of the measured concentrations in terms of the measured, meteorologically pre-processed and predicted (using numerical weather prediction models and a meso-scale meteorological model) meteorological variables. All the four episodes addressed were associated with areas of high pressure (Oslo, Helsinki and London) or a high-pressure ridge (Milan). The best meteorological prediction variables were found to be the temporal evolution of the temperature inversions and atmospheric stability and, in some of the cases, wind speed. Strong ground-based or slightly elevated temperature inversions prevailed in the course of the episodes in Oslo, Helsinki and Milan, and there was a slight ground-based inversion also in London; their occurrence coinciding with the highest PM 10 concentrations. The same result was also obtained by considering an additional set of seven PM 10 episodes from the larger dataset. The inversions in Oslo and Milan were mainly caused by the advection of warmer air above a relatively colder surface, and that in Helsinki by radiation cooling of snow-covered ground. It was also found that a low wind speed is not necessarily a good indicator of episodes; this is the case, e.g., in the Po valley, due to the frequently occurring calm and low wind speed conditions there. r 2005 Elsevier Ltd. All rights reserved. Keywords: Episode; Urban; PM 10 ; Inversion; Forecasting ARTICLE IN PRESS www.elsevier.com/locate/atmosenv 1352-2310/$ - see front matter r 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.atmosenv.2004.09.090 Corresponding author. Tel.: +358 9 1929 5450; fax: +358 9 1929 5403. E-mail address: jaakko.kukkonen@fmi.fi (J. Kukkonen).
15
Embed
Analysis and evaluation of selected local-scale PM10 air pollution episodes in four European cities: Helsinki, London, Milan and Oslo
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
ARTICLE IN PRESS
1352-2310/$ - se
doi:10.1016/j.at
�CorrespondE-mail addr
Atmospheric Environment 39 (2005) 2759–2773
www.elsevier.com/locate/atmosenv
Analysis and evaluation of selected local-scale PM10 airpollution episodes in four European cities: Helsinki, London,
Milan and Oslo
Jaakko Kukkonena,�, Mia Pohjolaa, Ranjeet S Sokhib, Lakhu Luhanab,Nutthida Kitwiroonb, Lia Fragkoub, Minna Rantamakia, Erik Bergec,Viel Ødegaardd, Leiv Havard Slørdale, Bruce Denbye, Sandro Finardif
aAir Quality Research, Finnish Meteorological Institute, Sahaajankatu 20 E, Helsinki, FinlandbUniversity of Hertfordshire, ASRG, Science and Technology Centre, Hatfield, UK
cKjeller Vindteknikk, P.O.Box 122, 2007 Kjeller, NorwaydNorwegian Meteorological Institute, Postboks 43, Oslo, NorwayeNorwegian Institute for Air Research, Postboks 100, Norway
fARIANET, via Gilino 9, 20128 Milano, Italy
Received 18 February 2004; received in revised form 8 September 2004; accepted 20 September 2004
Abstract
We have analysed in detail four selected episodes involving substantially high concentrations of PM10 that occurred in
Oslo on 4–10 January 2003, in Helsinki on 3–14 April 2002, in London on 18–27 February 2003 and in Milan on 14–19
December 1998. We have also utilised a more extensive dataset containing relevant information regarding 21 episodes
from seven cities in six countries. The four episodes analysed in detail were recently occurring cases that were at least
partly caused by various local emission sources. In particular, we have addressed the evolution of the measured
concentrations in terms of the measured, meteorologically pre-processed and predicted (using numerical weather
prediction models and a meso-scale meteorological model) meteorological variables. All the four episodes addressed were
associated with areas of high pressure (Oslo, Helsinki and London) or a high-pressure ridge (Milan). The best
meteorological prediction variables were found to be the temporal evolution of the temperature inversions and
atmospheric stability and, in some of the cases, wind speed. Strong ground-based or slightly elevated temperature
inversions prevailed in the course of the episodes in Oslo, Helsinki and Milan, and there was a slight ground-based
inversion also in London; their occurrence coinciding with the highest PM10 concentrations. The same result was also
obtained by considering an additional set of seven PM10 episodes from the larger dataset. The inversions in Oslo and
Milan were mainly caused by the advection of warmer air above a relatively colder surface, and that in Helsinki by
radiation cooling of snow-covered ground. It was also found that a low wind speed is not necessarily a good indicator of
episodes; this is the case, e.g., in the Po valley, due to the frequently occurring calm and low wind speed conditions there.
Local-scale episodes are presented in the upper table, while episodes that are caused mainly by regional and long-range transport appear in the lower table.
J.
Ku
kk
on
enet
al.
/A
tmo
sph
ericE
nviro
nm
ent
39
(2
00
5)
27
59
–2
77
32761
ARTICLE IN PRESSJ. Kukkonen et al. / Atmospheric Environment 39 (2005) 2759–27732762
the episodes in four European cities in relation to
prevailing meteorological conditions, local emissions,
and regionally and LRT background concentrations.
We have primarily analysed PM10 concentrations
instead of those of PM2.5, due both to the better
availability of experimental data, and the fact that the
European Union has only issued PM air quality
directives for the PM10 fraction. However, the PM2.5concentrations are included in the analysis wherever
such measurements were available.
The literature on urban air pollution episodes is
scarce, especially that regarding their systematic char-
acterisation in terms of the prevailing meteorological
conditions. Furthermore, in previous literature usually
specific episodes have been analysed separately, or, at
best, a few episodes within one specific country. In the
present study, we have aimed at a structured and
homogeneous analysis in all the four cities. A particular
aim of this study is to gain a better insight of the
influence of various meteorological variables on the
evolution of high pollutant concentrations in European
cities.
2. Material and methods
The cities considered in this study are located in
geographic and climatic regions of Northern (Oslo and
Helsinki), Northwestern (London) and Southern (Mi-
lan) Europe. These areas represent maritime climate
(London and Oslo), a partly maritime-influenced and
partly continental climate (Helsinki), and a mainly
continental climate (Milan). Two of the cities are located
in or surrounded by mountainous terrain (Oslo and
Milan, respectively) and two cities are situated in fairly
flat areas (Helsinki and London). London and Milan are
amongst some of the largest cities in Europe (their
populations are 7.5 and 3.5 million, respectively), while
the metropolitan areas of Oslo and Helsinki are
relatively smaller conurbations (the populations of both
of these are approximately 1.0 million).
For the analysis we have selected recently occurring
(1998–2003) episodes that were predominantly caused
by various local emission sources. Study by Valkama
and Kukkonen (2004) has indicated that episodes having
similar levels of concentrations and belonging to the
same episode category (described above) tend to occur
regularly every year, or at least every other year. Hence,
the episodes considered in this paper may be considered
as being characteristic of each region.
Corresponding spring dust episodes have occurred in
Helsinki every year during the last few years, sometimes
several times in the same spring. Spring dust episodes
are the most common category of PM10 episodes in
Helsinki. In Oslo, both ‘‘wood-burning’’ and ‘‘road-
dust’’ episodes of varying strengths are to be expected
every winter or spring. The wood-burning episodes
characteristically occur in cold, calm and stable winter
conditions, in which the ground is covered with ice and
snow, and the traffic- induced suspension of road dust is
therefore of minor importance. In Milan, from one to
two corresponding PM10 episodes have occurred every
winter during recent years; these events are characterised
by simultaneous infringements of PM10 and NO2 limit
values, and by similar meteorological conditions (Fi-
nardi and Pellegrini, 2004).
Some published studies on UK episodes (Ryall et. al.,
2002; Malcolm et. al., 2000) have shown that they can
arise because of local emissions (such as traffic or
industrial emissions) under stagnant conditions as well
as because of contributions from long-range transport of
polluted air masses including Saharan dust storms.
However, these studies have considered the UK as whole
and have not focused on urban areas. An analysis of the
occurrence of elevated 24 hour mean concentrations of
PM10 in London (that is, concentrations greater than
50mgm�3 lasting for two or more days) for 2000–2003
has shown that five out of eight episodes occurred
during late winter or spring periods.
2.1. Topography of the cities and the main emission
sources
The city of Oslo is located at the northern end of the
Oslo fjord, surrounded by a large topographical pot
formation. The topographical features of the area tend
to worsen the dispersion conditions, capturing pollu-
tants emitted within the urban airshed. The most
important local sources of PM in Oslo are domestic
wood-burning in stoves that are used for wintertime
house heating, and vehicular traffic (Laupsa and
Slørdal, 2002). The influence of wood-burning PM
emissions is most dominant in the densely populated
central city area, where a large fraction of the flats are
equipped with strongly polluting old stoves.
The city of Helsinki and its surrounding regions are
situated in a fairly flat coastal area. The PM10concentrations in street level air are dominated by the
combustion, non-combustion and suspension emissions
originating from vehicular traffic (e.g., Kukkonen et al.,
2001b). The influence of small-scale local sources, such
as that of domestic wood burning, is negligible.
London is situated in relatively flat terrain with
shallow hills to the west and south. London is one of
the most congested cities in Europe; most of the CO and
more than half of NOx emissions result from road
transport (GLA, 2002). In central London, approxi-
mately 80% of road-transport-related PM10 originates
from diesel vehicles, such as taxis, buses and goods
vehicles (GLA, 2002).
The city of Milan and its surrounding urban area are
located in the central part of the Po river basin, in a flat
ARTICLE IN PRESSJ. Kukkonen et al. / Atmospheric Environment 39 (2005) 2759–2773 2763
area. The atmospheric circulation of the Po valley is
characterised by the strong modification of synoptic
flow due to the high mountains (Alps and Apennines)
that surround the valley on three sides. Calm conditions
(defined as conditions in which the hourly-averaged
wind speed at a height of 10m is lower than 1.0m s�1)
and light winds occur frequently. The most severe winter
episodes are commonly associated with elevated tem-
perature inversions. According to the regional emission
inventory, road traffic is mostly responsible for the PM10emissions in the Milan Province.
2.2. The selected air quality and meteorological
measuring stations
The locations of the air quality and meteorological
stations utilised in this study are presented in Figs. 1a–d.
In Oslo, the stations of Kirkeveien, Alna, Furuset,
Løren and Manglerud are in urban traffic environments,
while Iladalen and Birkenes are urban and rural
background stations, respectively. In Helsinki, the
stations of Toolo and Vallila are located in urban traffic
Fig. 1. (a–d) The location in the four cities (Oslo, Helsinki, London a
in this study.
environments, while Kallio 2 and Luukki are urban and
rural background stations, respectively. In London, the
station of Bloomsbury is in an urban centre, the station
of Marylebone Road is in an urban kerbside environ-
ment, the station of Bexley is in a suburb, the stations of
North Kensington and Brent are urban background
stations, and the stations of Harwell and Rochester are
rural background stations. In Milan, the station of
Zavattari is in an urban traffic environment, the stations
of Juvara and Limito are urban background stations.
The station of Limito is not located in Milan, but in a
smaller town in the Milan Province. The classifications
of the air quality stations have been made according to
EU directives, except in the case of London, where
national definitions are used.
The classifications of the meteorological stations used
in this study are as follows. In Oslo, Blindern and
Tryvann synoptic stations are in an urban and a rural
area, respectively; the station of Hovin is an urban
station. Helsinki–Vantaa is a synoptic station in a
suburban area, Helsinki–Isosaari is a synoptic station
located on a sea island and Kivenlahti is a World
nd Milan) of the air quality and meteorological stations utilised
ARTICLE IN PRESSJ. Kukkonen et al. / Atmospheric Environment 39 (2005) 2759–27732764
Meteorological Organisation (WMO) station that is on
a 327-m-high radio tower situated in a partly rural,
partly suburban environment. London Weather Center
is located in central London, Heathrow airport is a
synoptic station in a suburban area, and Herstmonceux
is a WMO station in a rural area. Milano Linate Airport
is a WMO station in a suburban area, and Juvara is an
urban background station.
2.3. Experimental methods for measuring PM10 and
PM2.5 concentrations
In Oslo, a Tapered Element Oscillating Micro-balance
(TEOM) was used at all of the stations where both PM10and PM2.5 are measured. At two stations, Alna and
Iladalen, the Eberline FH 62 IR (based on b-attenua-tion) was employed. The measurement height at all
stations was 3.5m. These instruments are maintained on
a three-monthly basis, and the flow rates are calibrated
on a six-monthly basis against the BIOS DC-2 primary
flow meter. The Partisol 2025 sequential air-sampling
instrument was utilised at the rural background station
of Birkenes. A national quality control system is now
implemented in Oslo based on criteria developed in
EUROAIRNET (Larssen et al., 1999) and EU direc-
tives.
In Helsinki, at the station of Toolo, the concentration
of PM10 was measured with a TEOM, and at the
stations of Vallila, Kallio 2 and Luukki, with Eberline
FH 62 I-R. The concentration of PM2.5 was measured at
the stations of Vallila and Kallio 2, with Eberline FH 62
I-R instruments. Sillanpaa et al. (2002) compared the
results of these continuously-operating monitors (Eber-
line and TEOM) with the reference gravimetric sampler
(Standard EN 12341, LVS-PM10). The sampling was
carried out in autumn for 7 weeks, and in winter and
spring for 8 weeks. The correction factors k (defined as:
reference sampler ¼ k�test instrument) for the concen-trations measured by the Eberline and TEOM varied in
autumn from 0.95 to 1.07, and in winter and spring from
0.84 to 0.86. These factors are near unity, probably
caused by the smaller semi-volatile proportion of aerosol
mass in Helsinki, compared with that in most other
European areas (Sillanpaa et al., 2002).
In London, the PM10 measurements are conducted
with TEOM instruments. The network in London
incorporates comprehensive QA/QC procedures (Ste-
venson, 2003). As part of the process, a six-monthly
audit is conducted; this involves flow checks, sample
inlet checks, as well as checks of site infrastructure and
location classification. The filters are changed when they
approach a mass loading of 80% or more.
In Milan, at the monitoring stations utilised here,
PM10 is measured with TEOM instruments. The QA/QC
of the air quality monitoring network of the Lombardia
Region is performed by the European Reference
Laboratory for Air Pollution. These procedures have
been evaluated in comparison with the EUROAIRNET
scheme for a complete QA/QC by the National
Environmental Protection Agency (Desiato et al., 2000).
2.4. Modelling methods
At present, there are no generally applicable theore-
tical schemes for the interpretation of the data measured
in episodic conditions, both due to the related extremely
stable atmospheric conditions and to the characteristics
of the urban meteorology. However, some improve-
ments have recently been achieved in the urbanisation of
NWP, MM and meteorological pre-processing (MPP)
models (e.g., Baklanov et al., 2002).
The synoptic-scale meteorological analyses are based
on the results computed by the national versions of the
NWP model HIRLAM in the case of Norway and
Finland, and on the ECMWF model in the case of UK
and Italy. The HIRLAM and ECMWF models are used
only for the synoptic analysis. The meso- and micro-
scale meteorological conditions are analysed using
locally measured data and sounding data for all the
cities. In addition, we have utilised the predictions of the
two MPP models (for Helsinki and Milan), and the
MM5 model (for London).
The MM5 model has been used to predict the
conditions in London, as additional meso- and micro-
scale data was especially needed in order to interpret
that episodic case. We have used five nesting levels in the
MM5 computations with the horizontal resolutions of
81, 27, 9, 3 and 1 km; the finest grid covers the entire
London area. We utilised the land cover information
from the US Geology Survey, and the reanalysis data of
the National Centers for Environment Prediction and
Atmospheric Research (NCEP and NCAR) as meteor-
ological input data. The physical schemes were as
follows: Simple Ice Explicit Moisture Scheme and
Multi-Layer Soil Surface Scheme (Dudhia, 1996),
MRF boundary layer scheme (Hong and Pan, 1996),
and Grell Convective Parameterisation (Grell et al.,
1994).
In Helsinki, the Monin–Obukhov length was eval-
uated using a MPP model, MPP-FMI, that has been
adapted for an urban environment (Karppinen et al.,
2000a); this model was originally based on the energy
budget method of van Ulden and Holtslag (1985). The
model utilises meteorological synoptic and sounding
observations, and its output consists of estimates of the
hourly time series of the relevant atmospheric turbulence
parameters and the boundary layer height. The compu-
tation is based on a combination of the data from the
stations at Helsinki–Vantaa and Helsinki–Isosaari. The
performance of the integrated modelling system contain-
ing the MPP-FMI model has been evaluated, e.g., by
Kousa et al. (2001).
ARTICLE IN PRESSJ. Kukkonen et al. / Atmospheric Environment 39 (2005) 2759–2773 2765
In Milan, the Monin–Obukhov length was evaluated
using a meteorological pre-processor, SURFPRO (Fi-
nardi et al., 1997 and Arianet, 2002); the model is based
on the energy budget method of van Ulden and Holtslag
(1985), and Paine (1988). The model utilises surface-
based data and vertical profiles of temperature, and