Professional Documents
Culture Documents
DOI 10.1007/s11157-010-9227-2
R EV I E W S
Abstract National directives on air quality oblige the other, the lack of a common strategy which allows
nations to monitor and report on their air quality, to compare the state of the air for cities that follow
allowing the public to be informed on the ambient different directives. The main differences between the
pollution levels. The last is the reason for the always indices will be also described.
increasing interest, demonstrated by the number of
publications on this topic in recent years, in air Keywords Air quality indices · Pollution indices
quality/ pollution indices: since the concentration of
individual pollutants can be confusing, concentration
measure- ments are conveniently transformed in
terms of an air quality index. In this way, complex 1 Introduction
situations are summarized in a single figure, letting
comparisons in time and space be possible. In this Clean air can be considered a basic requirement of
paper we will give an overview about the Air human health and well-being. Neverthless, according
Quality/Pollution Indices proposed in literature and/or to the World Health Organization (WHO) assessment
adopted by countries, trying also to categorize them of the burden of disease due to air pollution, more
into homogeneous groups. For the classification than 2 million premature deaths each year can be
different approaches can be followed. Since in real attributed to the effects of air pollution (WHO 2006)
life exposure to mixtures of chemicals occurs, with Worldwide, many cities continuously assess air
additive, synergistic or antagonistic effects, here we quality using monitoring networks that record air
will distinguish between indices that consider the pollution concentrations at several points. Pollutant
conjoint effect of pollutants and indices only based on concentrations may be conveyed to the public through
the actual most dangerous pollutant. This brief review periodic reports and an approach widely used is to
on air pollution indices shows, on one side, the wide communicate the health risk by using an air pollution/
interest in the problem, on quality index. Actually, air pollution data are com-
plex. By using environmental synthetic indices we are
able to summarize complex situations in a single
A. Plaia (&) · M. Ruggieri
figure, allowing for comparisons in time and in space
Department of Statistical and Mathematical Sciences,
University of Palermo, Italy, Viale delle Scienze ed. 13, (Bruno and Cocchi 2007).
90128 Palermo, Italy Many different definitions of AQI can be found in
e-mail: plaia@unipa.it literature; nevertheless, an index ranging from good to
M. Ruggieri bad seems to be the most widespread way to interpret
e-mail: mariantonietta.ruggieri@unipa.it
123
air quality, as it allows to make the concentrations in called, both in literature and in this paper, Air
micrograms/m3 more understandable for the general Pollution Indices—APIs). The former are based on an
public. At this aim, the definition in (Shooter and arithmetic summation of relative concentrations of air
Brimblecombe 2009) seems very suitable: ‘‘Air pollutants or relative numbers of exceedences of air
quality indices aim at expressing the concentration pollutant specific short-term standards. They can be
of individual pollutants on a common scale where considered a summary assessment of the ambient air
effects, usually health effects, occur at a value that is pollution and have not direct relation to the well-
common to all pollutants’’. In other words, AQIs can being and health of human beings. Their typical
be considered as a simple and understandable way to structure is:
measure the air quality with respect to its effects on XP f l
human health. 1 C
ASI ¼ ;
Of course, AQIs can be use with different aims/ P p¼1 R p
purposes: to communicate air quality to citizens, to
assess the success/unsuccess of pollution reduction where C is the mean concentration (mostly over one
strategies, to monitor medium and long term trends, year) and R is a reference value for the air pollutant
. . ., and at these aims, even if AQIs’ basic concepts p. If a short-term air pollution stress is being assessed,
are similar, they can show large differences in C is the annual number of actual exceedences of air
practical implementation (de Leeuw and Mol 2005). pollutant specific standards and R is the correspond-
ing annual number of exceedences permitted in
directives or guidelines, e.g. of the European Union
(EU).
2 Methods AQIs, on the other side, are impact-related with
respect to people well being and actually they
A literature review can be structured in different represent the most diffused approach.
ways, ranging from a chronological organization to a ‘‘In general, the guidelines address single pollu-
topical or thematic one. We prefer considering this tants, whereas in real life exposure to mixtures of
last approach and, in the following sections, we will chemicals occurs, with additive, synergistic or antag-
try to categorize into homogeneous groups most of onistic effects. In dealing with practical situations or
the Air Pollution Indices proposed in literature and/or standard-setting procedures, therefore, consideration
adopted in some countries. All the indices here should be given to the interrelationships between the
reviewed follow a data-driven approach, the most various air pollutants’’ (WHO 2000). That is why we
common in literature, but a model based approach do not think the distinction between ASI and AQI is
that, modeling the data by a stochastic process, the most appropriate and useful. In our opinion, a
allows for missing value imputation and forecasting more useful approach consists in distinguishing
could be followed as well. Lagona (2005), for between indices that consider the conjoint effect of
example, proposes to follow a Hidden Markov Model pollutants and indices that are based only on the
approach that, even if could appear a promising way actual more dangerous pollutant (Sect. 4) This is the
to deal with the problem, lacks of simplicity and kind of classification we will follow in this review;
interpretability. nevertheless, Table 1 will classify the indices that
Even if air quality indices are updated every now will be described in this paper according to other
and then published material risks to be outdated very important statistical issues like spatial aggregation,
soon, we think that a review on this subject can be the availability of uncertainty measures for the index,
useful for both scientists and local governments. The being/not-being health based, purpose, accounting for
period spanned by the reviewed articles is low level exposure. Moreover, we will try to give an
1999–2009. overview about some actual AQI systems and
For the classification different approaches can be worldwide guidelines (Sect. 3) Even if they are not
followed. One could be that proposed by Makra et al. the main object of this review, Sect. 5 will introduce
(2003), who divides indices into Air Stress Indices the more general context of sustainability indices,
(ASIs) and Air Quality Indices (AQIs, sometimes while Sect. 6 will consider a brief discussion of AQIs
123
Re
v
Table 1 AQIs reviewed En
vir
Author Index Uncertainty Spatial Pollutant Health Purpose Low level on
measures aggregation aggregation based exposure Sci
Bi
Kassomenos et al. (1999) See Table 3 No No No Yes Uniform indexing of air pollution over large Yes ote
metropolitan areas ch
nol
Bruno and Cocchi (2002) I1g*,f, g or I2g*,f, g Yes Yes No Yes Recovering information from air quality NO (20
indices 11)
van den Elshout et al. (2008) CAQI No No No No Comparing urban air quality in real time Yes 10:
Mayer et al. (2002) DAQxp No No No Yes Information of people in the Internet Yes 16
5–
Mayer and Kalberlah (2009) LAQx No No No Yes To assess daily and long term air pollution Yes 17
Swamee and Tyagi (1999) PP 1=q
q
No No Yes Yes Ambiguity-free ecipsicity-free air quality No 9
p¼1 sp
function
Kyrkilis et al. (2007) PP q
1=q No Yes Yes Yes Estimating citizen exposure No
p¼1 ðAQIp Þ
Cheng et al. (2004) RAQI No No Yes Yes To produce an objective result in the long Yes
term
Makra et al. (2003) ASI1 ; ASI2 ASIBW No No Yes No To consider mean and short term air stress No
indices
Cogliani (2001) I, Ic No No Yes Yes To evaluate and rank air quality in megacities Yes
Gurjar et al. (2008) MPI No Yes Yes Yes Air pollution forecasting accounting for Yes
meteorological variables
Chelani et al. (2002) PP b No No Yes Yes To assess air quality status in metropolitan No
ha p¼1 Ip i
cities
Bishoi et al. (2009) NAQI No No Yes No To define the state of air in relative terms No
Zhou et al. (2006) hP i1=q
P q q Yes No Yes Yes To quantify the loss of information due to NO
p¼1 wp rps aggregating function
1 Zujic et al. (2009)
PK
s¼1 AQIps · Wps No Yes No Yes To reflect effective population exposure Yes
2
3 16
7
Rev Environ Sci Biotechnol (2011) 10:165–179 1
and health-effects. Finally some conclusions will be considered harmful to public health and environment.
illustrated in Sect. 7. The Clean Air Act established two types of national
Along the whole paper, whenever it will be air quality standards. Primary standards set limits to
necessary, we will refer to a data matrix X, whose protect public health, including the health of ‘‘sensi-
generic element, xtps, represents the concentration of tive’’ populations such as asthmatics, children and
the pollutant p recorded at a site s at time t (usually a elderly. Secondary standards set limits to protect
daily synthesis obtained by aggregating hourly data). public welfare, including protection against visibility
All the reported URLs have been accessed on impair- ment, damage to animals, crops, vegetation
September 5th 2010. and build- ings. The Clean Air Act requires periodic
reviews of the science upon which the standards are
based and the standards themselves.
3 Some current AQI systems and worldwide EPA AQI (http://www.epa.gov/air/urbanair/) is an
guidelines index for reporting daily air quality. It tells how clean
or polluted air is, and what associated health effects
A fundamental reference is WHO ‘‘Air quality might be a concern for you. The AQI focuses on
guidelines for Europe’’, published for the first time health effects you may experience within a few hours
in 1987, and updated in 2000 (WHO 2000) and 2006 or days after breathing polluted air.
(WHO 2006). ‘‘The WHO air quality guidelines EPA has set NAAQS for six principal pollutants,
(AQGs) are intended for worldwide use but have which are called ‘‘criteria’’ pollutants: CO, Pb (not
been developed to support actions to achieve air included in the computation of AQI), NO2 ; O3 , PM
quality that protects public health in different and SO2 (http://epa.gov/air/criteria.html). Six
contexts’’ (WHO 2006). categories corresponding to different level of health
Even if WHO air quality guidelines define a concerns (and symbolized by different colors) are
common and worldwide aim—i.e. protecting public considered.
health—, the Organization itself asserts that Subindices are calculated according to a table
‘‘National standards will vary according to the (http://www.epa.gov/ttn/naaqs/standards/nox/fr/2010
approach adopted for balancing health risks, 0209.pdf, p. 65) by linear interpolation between the 6
technological feasibility, economic considerations class borders:
and various other political IH IL
and social factors, which in turn will depend, among AQIp ¼ ðCp BPL Þ þ IL
BPH BPL
other things, on the level of development and national
capability in air quality management. The guideline where:
values recommended by WHO acknowledge this
• AQIp is the index for pollutant p;
heterogeneity and, in particular, recognize that when
• Cp is the concentration (daily synthesis) of the
formulating policy targets, governments should con-
pollutant p;
sider their own local circumstances carefully before
• BPH is the breakpoint 2' Cp ;
adopting the guidelines directly as legally based
• BPL is the breakpoint ::: Cp ;
standards’’ (WHO 2006).
• IH is the AQI value corresponding to BPH;
The United States Environmental Protection
Agency
(EPA) started to use an Air Quality Index (AQI) in for wide- spread pollutants from numerous and
1976 (the original name was Pollutant Standard Index diverse sources
—PSI) for use by States and local agencies on a
voluntary basis. The aim was to create a certain
homogeneity among the
14 different indices used by more than 50 urban areas
in USA and Canada at that time. The Clean Air Act
(the law that defines EPA’s responsibilities for
protecting and improving the nation’s air quality),
which was last amended in 1990, requires EPA to set
National Ambient Air Quality Standards (NAAQS)
123
2 Rev Environ Sci Biotechnol (2011) 10:165–179
• IL is the AQI value corresponding to BPL. corresponds to the national air quality standard for
The final index, ranging in [0, 500], is obtained as each pollutant.
the largest or ‘‘dominant’’ AQIp value and usually Canadian government, through the Meteorological
Service of Environment Canada, provides an AQI
commu- nicated (http://www.airnow.gov/) together
with the computed in the same way as EPA’s one, but consid-
‘‘dominant’’ pollutant; an AQI value of 100 generally ering 4 categories only. Recently, a new Air Quality
123
Health Index (AQHI) has been added to AQI. The accordance with its public health objectives, pub-
AQHI is based on the relative risks of a combination lishes the Bulletin de l’Air based on Atmo indices
of common air pollutants which are known to harm calculated by certified air-quality monitoring agen-
human health: O3, PM and NO2 (Stieb et al. 2008). cies (Associations Agre´e´es de Surveillance de la
The AQHI is measured on a scale ranging from 1 to Qualite´ de l’Air—AASQA). The Atmo index, sym-
10?. The AQHI values are also grouped into 4 health bolised by a giraffe, represents the mean urban air
risk categories helping to easily and quickly identify quality using a single figure scale. The Atmo ranges
the actual level of risk (http://www.ec.gc.ca/cas-aqhi/ from 1 to 10 (1 = very good air quality, 10 = very bad
default.asp?lang=En&n=065BE995-0). air quality) and bases its calculation on four subin-
At European level, the last air quality directives dices characterising the four pollutants NO2 ; SO2 ; O3
came into force in June 2008 and will be transposed and PM. The thresholds used to define subindex
into national legislation by June 2010 (Directive levels were set basing on regulatory criteria on air
2008/50/EC of the European Parliament and of the quality (http://www.atmoauvergne.asso.fr/en/index/
Council of 21 May 2008 on ambient air quality and calculation.htm). The final Atmo index is equal to
cleaner air for Europe, http://eur-lex.europa.eu/ the highest of the four subindices.
LexUriServ/LexUriServ.do?uri=OJ:L:2008:152:0001: In Germany, at our knowledge, no Air Quality
0044:EN:PDF). This directive lays down measures Index has been introduced. One-hour-averages of
aiming at, among others, ‘‘assessing the ambient air NO2 ; SO2 ; O3 ; CO and PM10 are available at http://
quality in Member States on the basis of common www.env-it.de/umweltbundesamt/luftdaten/index.html?
methods and criteria’’. Actually this directive has not setLanguage=en and the values are hourly updated.
been applied yet, and each country follows (when it is For each pollutant 11 classes (10 for CO and 7 for
done) its own method. The European Environment O3), symbolized by different colors, are considered.
Agency (EEA, http://www.eea.europa.eu/) is the Infor- mation on air quality can also be found at
agency of the European Union (that came into force in http://db.eurad.uni-koeln.de/index_e.html?/prognose/
late 1993) whose task is to provide sound, indepen- index_e.html. An AQI, defined as:
dent information on the environment; it represents a
major information source for who is involved in SO2 ð24 hÞ NO2 ð24 hÞ PM10 ð24 hÞ
AQI ¼ Max ; ; ;
developing, adopting, implementing and evaluating 125 90 50
environmental policies, and also for the general pub- !
lic. Currently, the EEA has 32 member countries. O3 ð24 hÞ COð24 hÞ
; p 50;
The Department for Environment, Food and Rural 100 10; 000
Affairs (DEFRA) is the government department
responsible for environmental protection in the is computed by the Rhenish Institute for Environ-
United Kingdom. Here, most of air pollution mental Research at the University of Cologne. The
information services use the index and banding index assumes values in 6 classes, from Very Good
system approved by the Committee on Medical (AQI \ 10) to Very Poor (AQI [ 80).
Effects of Air Pollution Episodes (COMEAP). The The Air Monitoring Networks of Flanders, Brus-
system uses an index ranging from 1 to 10, divided sels and Wallonia compute and communicate air
into four bands to provide more details about air pollution through an index scaled from 1 (excellent
pollution levels in a simple way. The overall air air quality) to 10 (awful quality). The computation is
pollution index for a site or region is calculated from performed by using data obtained from a telemetric
the highest concentration of the five pollutants NO2 ; networks that measure continuously the air quality in
SO2 ; O3 ; CO and PM10. The sub index for each the 3 Regions. The index is based on the concentra-
pollutant again is computed by linear interpolation tions of NO2 ; SO2 ; O3 and PM10. A ‘‘characteristic
between the 10 class borders, according to a table value’’ is computed every day for these 4 pollutants
(http://www.airquality.co.uk/ standards.php). and then compared to a concentration scale. The
The French Environment and Energy Management concentration scales are based on the European
Agency (ADEME), in cooperation with the Ministry guidelines concerning the assessment and manage-
of Ecology and Sustainable Development and in ment of the ambient air quality and reported in a table
(http://www.irceline.be/*celinair/english/homee n_ hourly data. Very often, to obtain a daily synthesis for
java.html) that shows, for each pollutant, the rela- each pollutant at each monitoring site, a researcher
tion among the measured concentrations, the index can refer to the guidelines of the
value and the corresponding scale, symbolized also national/international environmental agencies (that
by different colors. usually, given a pollu- tant, consider the same
In Italy air monitoring should be coordinated by function as time synthesis). A second step, necessary
ISPRA (Istituto Superiore per la Protezione e la at least before aggregating by pollutants, concerns
Ricerca Ambientale). Actually, through its site (http:// standardization: a first possibility lies in a simple ratio
www.apat.gov.it/site/it-IT/Servizi_per_l’Ambiente/ (eventually multiplied by 100) between the pollutant
Dati_di_Qualita’_dell’aria/), it is possible to see concentration and its specific threshold value.
what some cities or some regional agencies do. Alternatively, a segmented linear function can be
Some of them communicate only the observed daily considered. In the more recent paper (Bruno and
concentrations of pollutants, others (Puglia, Emilia- Cocchi 2007) the first possibility is preferred,
Romagna, Piemonte, Toscana) transform concentra- considering the choice as more coherent with the
tions in subindices, even if usually the number of recent EU directives.
categories is not the same from a region to another. After standardizing, a choice is necessary: if
Subindices for each pollutant are usually obtained by aggregating first by pollutant and then by space or
dividing observed concentration by a reference value vice versa. The results usually depend on this choice.
and multiplying by 100. The final index is equal to the The authors consider both the possibilities, compar-
highest of the determined subindices. ing the results:
• Monitoring site—Pollutants aggregation.
4 Literature review Here the spatial dimension is first eliminated
(through a function g()), then the standardized
4.1 Single-pollutant indices values f(g()) are aggregated by pollutants (func-
*
tion g ), getting the final index:
A very simple, but widely applicable indicator is
proposed by Kassomenos et al. (1999), who consider I1gp;f ;g ¼ gppf ðgs ðxps ÞÞ
neither an aggregation by pollutant nor by space. The
• Pollutant—Monitoring sites aggregation.
authors define a set of limit values, pollutant specific
Here the standardized values (f()) need to be
and sometimes also site-specific (Table 2).
computed as a first step, then the aggregation by
The choice of one of the alternatives in Table 2 *
pollutant (through a function g ()) and finally the
depends on the domain where monitoring stations are
spatial dimension is eliminated (function g),
located, since in urban domain population risk should
getting the final index:
be mainly addressed, while in a semi-urban or rural
domain the effects on flora and fauna should be p
p ðf ðx
I2gp;f ;g ¼ gs ðg ps
accounted for. ÞÞ
According to these limits, for each site and each
pollutant an air quality indicator, that assumes values
in 7 classes, is considered (Table 3), where C is the Both I1 and I2 will depend on the three functions
*
observed concentration. f, g, g .
Bruno and Cocchi (2002, 2007), consider explicite- *
ly the three dimensions upon which air quality data About the two functions g() and g () the authors
are defined: time, space and type of pollutant. The propose two order statistics: the median (m) and the
authors describe the aggregation steps that have to be maximum (M). In this way, given the type of standard-
followed in order to get a final index from a matrix of ization chosen, 8 different indices can be obtained, 4
elementary data xtps. The first aggregation step I1-type and 4 I2-type. Actually, two of these are the
same, I1MM and I2MM, as the result obtained by using
concerns usually time, allowing to get a daily
synthesis starting from the maximum is not influenced by the order of
aggregation. A measure of dispersion for each of the
two series of indices, I1 and I2 type, is proposed.
Table 2 Limits for the scale of air quality indicators
Value Alternative 1 (semi-urban or rural areas) Alternative 2 (urban areas)
CUL Upper protection limit (greater health risk and worse Upper protection limit (greater health risk and worse air
air quality condition or double the value of CLL) quality condition or double the value of CLL)
CLL Lower protection limit (standard limit value for Lower protection limit (standard limit value for health
health protection) protection)
CTV Target value, set by standards Short term target value
CAV Alert value, required by standards Alert value, 0.85 of CTV value
CIV Intermediate value, the limit for vegetation Intermediate value, ðCTV þ CAM Þ=2
protection when it is lower than the one for health
effects
CAM Annual mean limit specified by standards Annual mean value from recorded data
Table 3 Air quality indicator scale daily value and as an hourly index, is calculated
according to a grid in a table (5 classes heavily
Index Air quality indicator Limits inspired by EU legislation and based on a
7 Extreme C [ CUL compromise among the participating cities) by linear
6 Severe CUL 2' C 2' CLL interpolation between the class borders. The final
5 Bad CLL 2' C 2' CTV index is the highest value of the sub-indices for each
4 Critical CTV 2' C 2' CAV component. The index is computed by separating
3 Poor CAV 2' C 2' CIV
traffic monitoring sites from urban background sites
2 Moderate CIV 2' C 2' CAM
and considering NO2 ; PM10 and CO in the first case
1 Good CAM 2' C and NO2 ; PM10 ; O3 ; CO and SO2 in the second. The
EyeOnEarth site (http:// eyeonearth.cloudapp.net)
provides daily CAQI val- ues across Europe.
Among all the proposed indices, I1MM ¼ I2MM is In (Mayer et al. 2002; Mayer et al. 2004) an index
considered in (Bruno and Cocchi 2007), the choice developed and tested by the Research and Advisory
being supported by the well-known EPA AQI. Institute for Hazardous Substances, Freiburg, Ger-
Bodnar et al. (2008) present an interesting applica- many, and the Meteorological Institute, University of
tion of two of the indices proposed by Bruno and Freiburg, Germany, is reported. It is an impact-
Cocchi (2002) in the perspective of defining a Euro- related index applicable for the information of people
pean common index methodology which makes air in Internet on the daily integral air quality. For each
quality comparable in time across different countries. pollutant a linear interpolation between single index
I2MM and I2mM have been computed for a synthetic classes is computed:
communication of daily health risk related to air f( \ l
pollution 2005 data collected in Piemonte and Lom- p DAQxup DAQxlow p p
DAQx
¼ p p p ðCinst: Clow Þ
bardia (Italy), Berlin and Brandenburg (Germany) and Cup Clow
Masovian Province (Poland). A comparison of the þ DAQxlow ;
two indices, as proposed by Bruno and Cocchi
p
(2002), has
been also used to assess the spatial or network where: Cinst. is the daily maximum 1-h concentration
variability. of NO2 ; SO2 , and O3, the daily highest 8-h running
The paper by van den Elshout et al. (2008) presents mean concentration of CO or the daily mean concen-
p p
the results of a European project, CITEAIR, initiative a
of the cities of Leicester, Paris, Prague, Rome and
Rotterdam (http://citeair.rec.org). The paper proposes
a Common Air Quality Index (CAQI) which allows to
compare air quality for different cities in different
countries in real time. CAQI, that is computed both as
tration of PM10; Cup and Clow are the upper and lower DAQxlow are the index values corresponding to C
p
pollutant exceeds its threshold. The index for a region 0 if xps ::: xpmin 8p; s
s ps
s is defined as I ¼ f ðAs Þ, where As ¼ gðx Þ is min min
a p p where xp = 50% of NAAQS and x p is, for each
particular attribute (air quality, water quality …) for a pollutant, the concentration corresponding to AQI =
region depending on P indicators (individual sources 500. The final index will be:
of pollution). That is: 8 9ð 1=qÞ
8 < X =
f ðgðxps ; xp0 s ÞÞ [ AQI ¼ xp Dp ðxps Þ
q
:
0>
if x [ X min ; xp0 s [ X min ; p 6¼ : ;
p:D ð Þ [ 0
p0 < ps
> p p0 p
Is ¼ f ðgðxps ÞÞ [ 0
min
>if xps [ X ;p0 s ::: X ;
min
p 6¼ The pollutant space is divided into 5 areas charac-
0
>: x p p p 0