Environ Monit Assess (2017) 189:635
https://doi.org/10.1007/s10661-017-6327-2
Particulate matter levels in a South American megacity:
the metropolitan area of Lima-Callao, Peru
Jose Silva & Jhojan Rojas & Magdalena Norabuena &
Carolina Molina & Richard A. Toro &
Manuel A. Leiva-Guzmán
Received: 5 June 2017 / Accepted: 24 October 2017
# Springer International Publishing AG 2017
Abstract The temporal and spatial trends in the vari-
ability of PM10 and PM2.5 from 2010 to 2015 in the
metropolitan area of Lima-Callao, Peru, are studied and
interpreted in this work. The mean annual concentra-
tions of PM10 and PM2.5 have ranges (averages) of 133–
45 μg m−3 (84 μg m−3) and 35–16 μg m−3 (26 μg m−3)
for the monitoring sites under study. In general, the
highest annual concentrations are observed in the east-
ern part of the city, which is a result of the pattern of
persistent local winds entering from the coast in a south-
southwest direction. Seasonal fluctuations in the partic-
ulate matter (PM) concentrations are observed; these
can be explained by subsidence thermal inversion.
There is also a daytime pattern that corresponds to the
peak traffic of a total of 9 million trips a day. The PM2.5
value is approximately 40% of the PM10 value. This
proportion can be explained by PM10 re-suspension due
Electronic supplementary material The online version of this
article (https://doi.org/10.1007/s10661-017-6327-2) contains
supplementary material, which is available to authorized users.
J. Silva: J. Rojas : M. Norabuena
National Meteorology and Hydrology Service, Jr. Cahuide 785,
Jesús María, Lima, Peru
J. Silva
e-mail: jsilva@senamhi.gob.pe
C. Molina : R. A. Toro : M. A. Leiva-Guzmán (*)
Center for Environmental Science and Department of Chemistry,
Faculty of Science, University of Chile, Las Palmeras 3425,
Ñuñoa, Santiago 7800003, Chile
e-mail: manleiva@uchile.cl
e-mail: manleiva@me.com
to weather conditions. The long-term trends based on
the Theil-Sen estimator reveal decreasing PM10 concen-
trations on the order of −4.3 and −5.3% year−1 at two
stations. For the other stations, no significant trend is
observed. The metropolitan area of Lima-Callao is
ranked 12th and 16th in terms of PM10 and PM2.5,
respectively, out of 39 megacities. The annual World
Health Organization thresholds and national air quality
standards are exceeded. A large fraction of the Lima
population is exposed to PM concentrations that exceed
protection thresholds. Hence, the development of pollu-
tion control and reduction measures is paramount.
Keywords Particulate matter . Air pollution assessment .
Long-term trend . Metropolitan area of Lima-Callao,
Peru . Megacity
Introduction
At present, there are a total of 3.3 million premature
deaths globally from exposure to air pollution (WHO
2016a). This figure could double by 2050 if pollution
continues to rise at the current rates (Lelieveld et al.
2015). The high levels of air pollution are mainly due to
the excess emissions released into the atmosphere from
vehicular, industrial, and residential emissions as well as
power energy generation, among others. One of the air
pollutants that negatively impacts public health is par-
ticulate matter (PM) (EPA-US 2009).
Particles with a diameter of less than 10 μm (PM10)
have the greatest ability to access—and thus have the
635 Page 2 of 18
greatest deleterious effect on—airways. Within the
PM10 fraction, smaller particles (less than 2.5 μm,
PM2.5) are deposited in the alveoli, the deepest part of
the respiratory system, and become trapped, possibly
having severe health effects (Stanek et al. 2011; Tager
2012; Valavanidis et al. 2013). Epidemiological studies
have revealed positive associations between PM con-
tamination and serious harm to human health (EPA-US
2009). The harm caused by PM10 and PM2.5 to human
health is manifested as cardiac and respiratory mortality,
decreased lung capacity in children and adults with
asthma, and an increased frequency of chronic obstruc-
tive pulmonary disease (EPA-US 2009).
To protect human health, organizations such as the
World Health Organization (WHO) have established
acceptable concentration thresholds, below which the
harm to humans is sufficiently minimal (WHO 2006).
However, particle pollution has health effects even at
very low concentrations, and it has not been possible to
identify any threshold below which no health damage
has been observed. Several governments around the
world have issued PM air quality standards (Vahlsing
and Smith 2012), such as those established by the Unit-
ed States Environmental Protection Agency (EPA-US
2013) and by the Ministry of the Environment of the
Government of Peru (PE-MINAM 2001).
Some of the greatest impacts of PM contamination on
people are evident in urban environments (Baklanov et al.
2016; Baldasano et al. 2003; Molina and Molina 2002,
2004). Fifty-two percent (52%) of the world’s population
lives in cities (UN 2016), and 31 cities around the world
that surpass 10 million inhabitants, which in total hold
approximately 700 million inhabitants (UN 2016). Five
of these so-called megacities are in Latin America, name-
ly, São Paulo, Brazil (top 12, 21.297 million inhabitants);
Mexico City (top 10, 22.157 million inhabitants); Buenos
Aires, Argentina (top 22, 15.334 million inhabitants); Rio
de Janeiro, Brazil (top 27, 12.981 million inhabitants);
and the metropolitan area of Lima-Callao (MALC) in
Peru, which now has 10.072 million inhabitants and is
currently positioned at number 31 of the most populated
cities in the world (UN 2016). These megacities of Latin
America have high PM levels (Bell et al. 2006; Reich
et al. 2008).
The ambient air pollution estimated annual health
impact of urban air pollution from particulate matter in
MALC caused the premature death of approximately
3900 Peruvians and accounted for the loss of approxi-
mately 66,000 disability-adjusted life years per year,
Environ Monit Assess (2017) 189:635
attributable to mortality (44%) and chronic bronchitis
(13%), restricted activity days (20%), and respiratory
symptoms (16%) (Awe et al. 2015).
In the MALC, the PM10 and PM2.5 levels in certain
areas in the city often exceed the established safety
thresholds for environmental concentrations recom-
mended by the WHO (2006) and the Ministry of the
Environment (MINAM by its acronym in Spanish) (PE-
MINAM 2001). The main source of PM is associated
with the growing size of the automotive fleet (1,195,353
in 2010 to 1,674,145 in 2015) (MTC 2016) and the use
of fossil fuels (Dawidowski et al. 2014). Some of the
older public transport vehicles have been removed from
Lima’s roads, but 50% of Lima’s busses and minivans
are over 18 years old (Carbajal-Arroyo et al. 2007;
Underhill et al. 2015). A gas reformulation program
has also been implemented to reduce the sulfur content
and to convert part of the automotive fleet to natural gas
(Dawidowski et al. 2014). Other sources are the indus-
trial activity (brickworks, cement production, smelting,
the fishing industry, and power generation) (PE-
MINAM 2014). Additionally, 12% of municipal waste
is open burned. Another factor is the prevailing anticy-
clonic meteorological conditions throughout the year,
which added to an abrupt topography and led to a
permanent subsidence and thermal inversion layer
above the city, thus providing a stable atmospheric
gradient that reduces the dispersion of air pollutants.
In this work, we analyze PM10 and PM2.5 measure-
ments taken in the period of 2010–2015 to characterize
the spatial and temporal distribution of the concentra-
tions of particles on different time scales. Air pollution
trends for PM10 and PM2.5 are analyzed. The results are
evaluated by comparing the mean annual and mean
daily concentrations of PM10 and PM2.5 with the corre-
sponding WHO and MINAM standards. The study aims
to understand PM air pollution trends to assess measures
implemented to improve air quality and to determine the
level of exposure of the MALC population over the past
6 years.
Materials and methods
Area of study
Lima is located at 77° W and 12° S in the middle of the
western side of the South American continent (Arellano
Rojas 2013; IGN 2014). It is the capital of the Republic
Environ Monit Assess (2017) 189:635
of Peru, of the region and of the province of Lima
(Fig. 1). It forms an extensive and populous urban area
known as the MALC. From the coastline, in a north-
easterly direction, the steep landscape ranges from a
large alluvial plain formed by the valleys of the Chillón,
Rímac, and Lurín rivers to a landscape of hills and
finally mountains in the extreme east, which are located
at more than 1000 m above sea level (and are the first
foothills of the western slopes of the central Andes). It
has an area of 2819.3 km2. In the summer (December
through March), the average temperature (relative hu-
midity) oscillates between 24 and 26 °C (65 and 68%) in
the mornings, whereas at night, it fluctuates between 18
and 20 °C (87 and 90%). In winter (June through Sep-
tember), the average daytime temperature (relative hu-
midity) oscillates between 18 and 19 °C (85 and 87%),
and during nights, it ranges between 18 and 20 °C (90
and 92%). The average annual precipitation is 10 mm
(Arellano Rojas 2013; SENAMHI 2016). The MALC is
under the influence of persistent atmospheric stability
caused by the interaction of the South Pacific Anticy-
clone, the Von Humboldt Current, and the presence of
the Andes mountain range, which runs parallel to the
coast. The thermal inversion reaches average altitudes in
summer from its base of approximately 500 m above sea
level, whereas in winter, its base reaches 1500 m above
sea level (Arellano Rojas 2013). A detailed description
of meteorological variables is provided in Fig. S1 in the
supplementary material.
PM10 and PM2.5 pollutant measurements
Since 1999, the MALC has had an air quality monitor-
ing program, which is operated by the General Direc-
torate of Environmental Health (DIGESA by its acro-
nym in Spanish) of the Ministry of Health (MINSA by
its acronym in Spanish); however, it was not until 2010
that the city of Lima had a real-time automatic air quality
monitoring network (AAQMN) (LR 2010). Station and
analyzer operation is the responsibility of the National
Service of Meteorology and Hydrology (SENAMHI by
its acronym in Spanish) of MINAM and is supervised
by the network manager. There are currently ten stations
that measure atmospheric aerosols (SENAMHI 2016)
(Fig. 1).
The sites in the network are selected to represent
where the population lives, plays, and works. The sta-
tion meets regulations for occupational health and safety
under national regulations. The electrical power system
Page 3 of 18 635
should meet electrical codes. In general, the shelter
temperature should be maintained between 20 and
30 °C. Operation of the station includes regularly sched-
uled station visits, instrument zero and span verifica-
tions, calibrations, detection of leaks, preventive main-
tenance, and documentation. The verifications are con-
ducted on a monthly or weekly basis. The data are
transmitted by telemetry to SENAMHI headquarter
where the data are validated to correct for null entries,
duplicates, and/or anomalies in the data. The database is
a 1-h time resolution. More information about the PM
instrumentation and GPS coordinates and station oper-
ations is provided in Table S1.
Data analysis
A descriptive statistical analysis was performed using
MS-Excel © (Microsoft Corporation, Redmond, WA,
USA) in the R programming language (R Development
Core Team 2012) using the Openair package (Carslaw
2013) under the open-source software RStudio: Inte-
grated Development Environment for R (RStudio
2016). Temporal trends were estimated using the
Theil-Sen approach (Sen 1968; Theil 1992). The
Theil-Sen test calculates the slopes among all pairs of
points, and the median slopes are selected as the Theil-
Sen estimate, which is used as the pollutant trend for the
given period (Carslaw and Ropkins 2012).
Assessment of standards and thresholds
To assess the possible health impact, the WHO thresh-
olds (GD-WHO) and PE-MINAM air quality standards
are compared on an annual and daily basis. The GD-
WHO values on an annual basis are 20 and 10 μg m−3
for PM10 and PM2.5, respectively, expressed as the
annual mean (WHO 2006). On a daily basis, they cor-
respond to values of 50 μg m−3 for PM10 and 25 μg m−3
for PM2.5 (WHO 2006), in both cases expressed as a 24-
h moving average. The PE-MINAM standards for PM10
concentrations on an annual basis are 50 μg m−3,
expressed as an arithmetic mean; on a daily basis, they
correspond to a value of 150 μg m−3 as a 24-h arithmetic
average (PE-MINAM 2001). To date, the PE-MINAM
has not established an annual PM2.5 standard but has
done so for the 24-h period (25 μg m−3 for the 24-h
arithmetic mean) (PE-MINAM 2001). It should be not-
ed that the WHO thresholds are stricter than those of PE-
MINAM, which is because the former consider only
635 Page 4 of 18
Fig. 1 Map of the MALC and the station locations of the Lima air
quality monitoring network (right: modified after Wikimedia.
https://goo.gl/n8t4MX, accessed 2017-03-10; left: google earth
v7.1.5.1557, 2016-03-02, DigitalGlobe 2016. http://www.earth.
health risks, whereas the other standards are established
by including cost-benefit considerations.
Results and discussion
PM10 and PM2.5 annual concentration levels
Table 1 presents a summary of the annual concentra-
tions, 98th percentile concentrations, and percentage of
annual data availability for the PM10 and PM2.5 mea-
surements at the MALC monitoring stations. For the
purposes of this study, only when there is data availabil-
ity greater than 60% are the data associated with a given
Environ Monit Assess (2017) 189:635
google.com, accessed 2017-03-10). For more detail about the air
quality monitoring network, see Table S1 in the supplementary
material
station considered valid values and representative of the
annual concentration.
The average PM10 concentrations for monitoring
stations (see Table 1(a)) are highest for VMT
(133 μg m−3), ATE (113 μg m−3), PPD (111 μg m−3),
and HCH (96 μg m−3). They are followed by lower
concentrations at the SJL (85 μg m − 3 ), STA
(82 μg m−3), and CRB (79 μg m−3) stations. The
stations with the lowest annual PM10 concentrations
are SBJ (52 μg m−3), SMP (48 μg m−3), and CDM
(45 μg m−3). The 98th percentile of environmental
concentrations of PM10 is between 1.6 and 2.1 times
the average annual value at the stations under study
and follows a pattern similar to the mean annual PM10
Environ Monit Assess (2017) 189:635 Page 5 of 18 635

Table 1 Annual PM10 and PM2.5 concentrations in μg m−3 and annual percentages of data availability by the stations under study (see Fig.
1)

Year Variable ATE SBJ CDM STA VMT HCH SJL SMP CBR PPD

a) PM10
舃

2010
舃 Annual Conc.
舃 123
舃 47a
舃 42a
舃 –
舃 –
舃 –
舃 –
舃 –
舃 –
舃 –
舃

(μg m−3)
P98 Conc. (μg m−3)
舃 189
舃 68a
舃 76a
舃 –
舃 –
舃 –
舃 –
舃 –
舃 –
舃 –
舃

Available data (%)

舃 74
舃 23
舃
a
32
舃
a
–
舃 –
舃 –
舃 –
舃 –
舃 –
舃 –
舃

2011
舃 Annual Conc.
舃 130
舃 51
舃 45
舃 76a
舃 –
舃 –
舃 –
舃 –
舃 –
舃 –
舃

(μg m−3)
P98 Conc. (μg m−3)
舃 203
舃 75
舃 75
舃 118a
舃 –
舃 –
舃 –
舃 –
舃 –
舃 –
舃

Available data (%)

舃 76
舃 88
舃 82
舃 15a
舃 –
舃 –
舃 –
舃 –
舃 –
舃 –
舃

2012
舃 Annual Conc.
舃 113
舃 49
舃 45
舃 80
舃 113
舃 –
舃 –
舃 –
舃 –
舃 –
舃

(μg m−3)
P98 Conc. (μg m−3)
舃 198
舃 76
舃 78
舃 143
舃 227
舃 –
舃 –
舃 –
舃 –
舃 –
舃

Available data (%)

舃 78
舃 73
舃 82
舃 80
舃 76
舃 –
舃 –
舃 –
舃 –
舃 –
舃

2013
舃 Annual Conc.
舃 109
舃 55
舃 48
舃 98
舃 134
舃 –
舃 –
舃 –
舃 –
舃 –
舃

(μg m−3)
P98 Conc. (μg m−3)
舃 199
舃 91
舃 88
舃 185
舃 298
舃 –
舃 –
舃 –
舃 –
舃 –
舃

Available data (%)

舃 94
舃 89
舃 86
舃 62
舃 88
舃 –
舃 –
舃 –
舃 –
舃 –
舃

2014
舃 Annual Conc.
舃 112
舃 51
舃 44
舃 73a
舃 90a
舃 101
舃 87
舃 49
舃 80
舃 107
舃

(μg m−3)
P98 Conc. (μg m−3)
舃 179
舃 98
舃 85
舃 135a
舃 200
舃 163
舃 145
舃 79
舃 137
舃 175
舃

Available data (%)

舃 94
舃 62
舃 85
舃 56a
舃 59a
舃 70
舃 71
舃 75
舃 69
舃 69
舃

2015
舃 Annual Conc.
舃 91
舃 46a
舃 42
舃 68
舃 153
舃 91
舃 82
舃 47
舃 77
舃 115
舃

(μg m−3)
P98 Conc. (μg m−3)
舃 162
舃 76a
舃 75
舃 114
舃 270
舃 145
舃 143
舃 80
舃 124
舃 188
舃

a a
Available data (%)
舃 74
舃 25
舃 70
舃 75
舃 32
舃 76
舃 82
舃 68
舃 79
舃 75
舃

Avg
舃 Annual Conc.
舃 113
舃 52
舃 45
舃 82
舃 133
舃 96
舃 85
舃 48
舃 79
舃 111
舃

(μg m−3)
P98 Conc. (μg m−3)
舃 188
舃 85
舃 80
舃 147
舃 249
舃 154
舃 144
舃 80
舃 131
舃 182
舃

Available data (%)

舃 82
舃 78
舃 81
舃 72
舃 65
舃 73
舃 77
舃 72
舃 74
舃 72
舃

b) PM2.5
舃

2014
舃 Annual Conc.
舃 44a
舃 18a
舃 16a
舃 21a
舃 23a
舃 38
舃 34
舃 19
舃 32
舃 35
舃

(μg m−3)
P98 Conc. (μg m−3)
舃 68a
舃 31a
舃 36
舃 30a
舃 39a
舃 67
舃 59
舃 33
舃 50
舃 55
舃

a a a a a
Available data (%)
舃 25
舃 25
舃 16
舃 8
舃 8
舃 71
舃 66
舃 74
舃 69
舃 69
舃

2015
舃 Annual Conc.
舃 35
舃 18
舃 16
舃 27
舃 26a
舃 24
舃 28
舃 17
舃 23
舃 29
舃

(μg m−3)
P98 Conc. (μg m−3)
舃 56
舃 30
舃 27
舃 47
舃 44a
舃 41
舃 46
舃 28
舃 40
舃 45
舃

Available data (%)

舃 65
舃 73
舃 64
舃 78
舃 32a
舃 79
舃 82
舃 69
舃 69
舃 76
舃

Avg
舃 Annual Conc.
舃 35
舃 18
舃 16
舃 27
舃 –
舃 31
舃 31
舃 18
舃 28
舃 32
舃

(μg m−3)
P98 Conc. (μg m−3)
舃 56
舃 30
舃 32
舃 47
舃 –
舃 54
舃 53
舃 31
舃 45
舃 50
舃

Available data (%)

舃 65
舃 73
舃 64
舃 78
舃 –
舃 75
舃 74
舃 72
舃 69
舃 73
舃

(−) No available data

a
Concentrations with low data availability. Only calculated for reference

concentrations. It is noteworthy that the stations with eastern part of the city, that is, the highest part,
the highest PM10 concentrations are located in the whereas those with a lower concentration are found
635 Page 6 of 18
in the western part, which is the lowest part of the city
(see Fig. 1); this trend is due to the pattern of persis-
tent local winds entering from the coast with a south-
southwest direction, causing the pollution loads to be
transferred to the eastern and northeastern areas,
which are critical deposition areas. The annual PM10
concentrations for the following groups of stations
are not significantly different at the 95% confidence
level (p value > 0.05): [VMT, ATE, PDA, and HCH],
[STA, SJL, and CRB], and [SBJ, CDM, and SMP].
For PM2.5 (see Table 1(b)), a similar pattern is gen-
erally observed, even though the average concentrations
between the stations are more similar and are in the
range of 35 to 16 μg m−3. The highest annual concen-
trations of PM2.5 are observed at ATE (35 μg m−3), PPD
(32 μg m−3), HCH (31 μg m−3), SJL (31 μg m−3), CRB
(28 μg m−3), STA (27 μg m−3), SBJ (18 μg m−3), SMP
(18 μg m−3), and CDM (16 μg m−3). The 98th percentile
value, in general, follows a pattern similar to the annual
concentrations of PM2.5. The 98th percentile of the
environmental concentration of PM2.5 is between 1.6
and 2.0 times the average annual value at the stations
under study. Two groups of stations in terms of the
magnitude of the concentration of PM2.5 are clearly
observed. The first group, formed by the stations of
ATE, PPD, HCH, SJL, CRB, and STA, exhibits the
highest concentrations; the second group, formed by
the SBJ, SMP, and CDM stations, has lower concentra-
tions. Among these groups of stations, an analysis of
variance (ANOVA) indicated that there are no signifi-
cant differences at the 95% confidence level (p value
> 0.05).
PM10 and PM2.5 monthly concentration level
Figure 2 shows the temporal variability of the monthly
(see Fig. 2a), daily (see Fig. 2b) and hourly (see Fig. 2c)
averages of the PM10 and PM2.5 concentrations obtained
from hourly measurements available at the monitoring
stations under study.
The average monthly variability (see Fig. 2a) exhibits
the highest PM 1 0 concentrations at the VMT
(168 μg m−3), PPD (138 μg m−3), ATE (136 μg m−3),
HCH (117 μg m −3 ), SJL (116 μg m −3 ), CRB
(108 μg m−3), STA (93 μg m−3), SBJ (57 μg m−3),
SMP (56 μg m−3), and CDM (52 μg m−3) stations. In
general, the maximum concentrations are observed in
the months of February through April, whereas the
minimum concentrations are observed between May
Environ Monit Assess (2017) 189:635
and September and are in the range of 90 to
36 μg m−3. In addition, a greater amplitude of the
variation of the monthly concentrations (expressed as
the difference between the maximum and minimum
monthly concentration) at the VMT station (98 μg m−3)
is observed, followed by the SJL, CRB, ATE, and PPD
(approximately 48 μg m−3) and HCH (34 μg m−3) sta-
tions, whereas the lowest annual variability is observed
at the STA, SMP, CDM, and SBJ stations, with values
between 24 and 10 μg m−3. The month-to-month vari-
ability in the PM10 concentration (Fig. 2a) exhibited
significant differences at the 95% confidence level (p
value < 0.05) between the maximum and minimum
values.
In the case of the PM2.5 average monthly variability
(Fig. 2a), the maximum concentrations observed for
PM2.5 per station from greatest to least rank as ATE
(43 μg m−3), PPD (38 μg m−3), SJL (37 μg m−3),
HCH (36 μg m−3), STA (33 μg m−3), SBJ (23 μg m−3),
CDM (21 μg m−3), and SMP (20 μg m−3). The maxi-
mum concentrations are generally evident between May
and September. The minimum concentrations of PM2.5
vary from season to season between 11 and 30 μg m−3,
with the absolute minima generally observed between
October and April. The amplitude of the variability
(expressed as the difference between maximum and
minimum monthly concentrations) is in the range of 7
to 14 μg m−3, according to the season. The stations with
the highest variability are VMT (14 μg m−3), ATE
(13 μg m−3), PPD (12 μg m−3), CRB (11 μg m−3),
HCH (10 μg m −3 ), CDM (9 μg m −3 ), and SMP
(7 μg m−3). Significant differences between the maxi-
mum and minimum values are observed at 95% confi-
dence (p value < 0.05).
PM10 and PM2.5 daily and hourly concentration levels
In the case of the behavior of the daily averages within
the weekly cycle (Fig. 2b) for PM10, there is an ampli-
tude of variation among the days of the week (expressed
as the difference between days with maximum and
minimum concentrations), with ranges between 8 and
24 μg m−3. The maximum amplitudes are observed at
the ATE (24 μg m−3) and SJL (20 μg m−3) stations, and
the lowest values are observed at the CRB (11 μg m−3),
CDM (10 μg m−3), and SMP (8 μg m−3) stations. There
is a significant decrease in concentrations on Sundays.
According to a 95% confidence-level ANOVA, there is
a significant decrease in concentrations on Sundays
Environ Monit Assess (2017) 189:635 Page 7 of 18 635

Fig. 2 Temporal variations in the PM2.5 and PM10 concentrations on a monthly, daily, and hourly basis for monitoring stations in the MALC
(see Fig. 1)
635 Page 8 of 18
compared with Tuesday through Friday (p value
< 0.05), when the concentrations are higher. This is
due to the variability in vehicular emissions because,
during weekends, vehicular traffic decreases, and the
emission contributed by mobile sources to the PM is
more than 68% (SENAMHI 2014). In the case of PM2.5,
the amplitude of variation between the days of the week
for the stations under study ranges between 2 and
6 μg m−3, which explains why there is no statistically
significant difference between the average concentra-
tions on different days of the week (p value > 0.05).
This result occurs because PM2.5, due to its smaller size
and mass, can remain for a longer time in the atmo-
sphere (Konstantinos 2008). The hourly variations for
PM10 and PM2.5 at the stations under study are shown in
Fig. 2c. It is observed that the maximum concentrations
occur at a peak generally between 6:00 and 10:00 in the
morning and a second peak between 18:00 and 23:00 h.
The average hourly maximums are observed at the
VMT (PM10: 154 μg m−3, PM2.5: 35 μg m−3), ATE
(PM10: 152 μg m−3; PM2.5: 56 μg m−3), HCH (PM10:
119 μg m−3; PM2.5: 47 μg m−3), and PPD (PM10:
140 μg m−3; PM2.5: 39 μg m−3) stations, followed by
the SBJ (PM10: 65 μg m−3; PM2.5: 24 μg m−3), CDM
(PM10: 62 μg m−3; PM2.5: 22 μg m−3), and SMP (PM10:
62 μg m−3; PM2.5: 24 μg m−3) stations. These trends can
be understood as follows: the MALC has a vehicular
fleet of 2.2 million; at an average age of 18 years old, the
residents of the MALC make a total of 10 million trips a
day in Lima and Callao (SENAMHI 2014). This com-
muter traffic in the metropolis results in peak hours of
traffic and therefore emissions, which would occur in
the morning and evening hours.
Relationships between PM2.5 and PM10
The relationship between the concentrations of PM2.5
and PM 10 at the same station is evaluated using
Pearson’s correlation coefficient. The results are pre-
sented in Table 2 (bold font), where it can be observed
that the correlation coefficients of (PM 2.5) St-i vs.
(PM10)St-i are in the range of 0.72 to 0.49. The greatest
values of the (PM2.5)St-i vs. (PM10)St-i correlation coef-
ficient are observed at the SMP (0.72) and ATE (0.70)
stations, whereas the least are found at the CRB (0.54),
VMT (0.57), and PPD (0.49) stations. In general, these
results are consistent because the emission sources of
PM10 and PM2.5 at the stations should be similar. In the
cases in which the correlation coefficients are smaller
Environ Monit Assess (2017) 189:635
(CRB, VMT, and PPD stations), the stations are located
in zones that have an important component of industrial
emissions, few green areas, and an intense vehicular
activity; consequently, the sources of PM10 and PM2.5
emissions are different. For example, there is a cement
plant at VMT that, as part of its production process,
emits PM10, whereas the emissions of PM2.5 in the zone
primarily originate from vehicles.
The Pearson correlation coefficients for each fraction
of particles at paired sites (PM10)St-i vs. (PM10)St-j and
(PM2.5)St-i vs. (PM2.5)St-j are presented in Table 2 (in
normal font and italic font, respectively). The values for
the correlations cover the range of 0.33 to 0.72 for the
coefficients of (PM10)St-i vs. (PM10)St-j and 0.37 to 0.78
for the coefficients of (PM2.5)St-i vs. (PM2.5)St-j (see
Table 2). Correlations with Pearson coefficients greater
than 0.6 can be considered strong correlations and dem-
onstrate that the PM emissions have the same emissions
pattern in the considered sites (Molina et al. 2017; Toro
et al. 2014). According to the results shown in Table 2,
the coefficients are categorized into three groups based
on the degree of correlation. Thus, there are higher
correlations among three groups of sites: the first con-
sists of the ATE, STA, and HCH sites; the second the
SMP, CDM, and SBJ stations; and the third the PPD,
CRB, and SJL stations. In the first group, the stations are
located relatively close to each other, and their environ-
ment has a high residential component. In the second
group, the stations are located in the lower part of the
city, and their environments have green areas and
asphalted streets, with a strong component of vehicular
emissions. Finally, the third group is located in Lima
North, where the stations located in the upper parts of
the city have similar patterns of PM10 and PM2.5 emis-
sions; this area of the city is characterized by large
industrial and commercial areas.
PM2.5/PM10 ratio
The annual average of the ratio for the stations under
study is 0.4 ± 0.1 (± 1σ), with a range of 0.21 to 0.44.
This indicates that in the MALC, the PM2.5 concentra-
tion represents between 44 and 21% of the total PM10
and on average 40%. In the literature, it has been esti-
mated that ratios higher than 0.7 indicate significant
vehicular emissions (Xu et al. 2017), which could be
assumed in the MALC because of its large vehicle fleet
(1.7 million); however, in the MALC, the ratio is lower
than this threshold, so there must be additional PM10
Environ Monit Assess (2017) 189:635 Page 9 of 18 635

Table 2 Site and inter-site Pearson correlation coefficients of (PM10)St-i vs. (PM10)St-j (normal font ), (PM2.5)St-i vs. (PM2.5)St-j (italics
font ) and (PM25)St-i vs. (PM10)St-j (bold font ) between paired monitoring stations

(PM2.5)St-i vs. (PM10)St-j (bold font) ATE SBJ CDM STA VMT HCH SJL SMP CBR PPD

(PM10)St-i vs. (PM10)St-j (normal font)

ATE
舃 (PM2.5)St-i vs. (PM2.5)St-j (italics font)
舃 0.70
舃 0.50
舃 0.42
舃 0.69
舃 0.32
舃 0.76
舃 0.60
舃 0.43
舃 0.53
舃 0.37
舃

SBJ
舃 0.45
舃 0.66
舃 0.74
舃 0.66
舃 0.40
舃 0.47
舃 0.57
舃 0.66
舃 0.30
舃 0.41
舃

CDM
舃 0.38
舃 0.61
舃 0.74
舃 0.52
舃 0.27
舃 0.49
舃 0.50
舃 0.78
舃 0.37
舃 0.39
舃

STA
舃 0.64
舃 0.53
舃 0.54
舃 0.62
舃 0.34
舃 0.71
舃 0.65
舃 0.50
舃 0.48
舃 0.46
舃

VMT
舃 0.36
舃 0.43
舃 0.40
舃 0.53
舃 0.57
舃 0.43
舃 0.48
舃 0.28
舃 0.57
舃 0.58
舃

HCH
舃 0.72
舃 0.35
舃 0.33
舃 0.45
舃 0.48
舃 0.69
舃 0.75
舃 0.52
舃 0.65
舃 0.53
舃

SJL
舃 0.67
舃 0.50
舃 0.45
舃 0.63
舃 0.59
舃 0.67
舃 0.65
舃 0.56
舃 0.65
舃 0.56
舃

SMP
舃 0.53
舃 0.62
舃 0.70
舃 0.58
舃 0.47
舃 0.50
舃 0.58
舃 0.72
舃 0.49
舃 0.39
舃

CBR
舃 0.51
舃 0.33
舃 0.41
舃 0.41
舃 0.57
舃 0.65
舃 0.62
舃 0.51
舃 0.54
舃 0.5
舃

PPD
舃 0.53
舃 0.37
舃 0.37
舃 0.49
舃 0.58
舃 0.58
舃 0.61
舃 0.45
舃 0.66
舃 0.49
舃

emission sources in the city. In the north, south, and east
areas of the MALC, there are a large number of unpaved
roads and sidewalks, which, undoubtedly, in addition to
the re-suspension of PM by vehicular traffic and the
presence of marine aerosols, can account for this
proportion.
The month-to-month, day-to-day, and hour-by-hour
variations in the PM2.5/PM10 ratio are illustrated in
Fig. 3. With respect to the month-to-month variation in
the PM2.5/PM10 ratio, Fig. 3a shows that the highest
values are recorded in the months of May to August,
and they are significantly different at the 95% confi-
dence level with respect to the minima, which is asso-
ciated with the variability of the meteorological vari-
ables (wind speed, wind direction, temperature, and
relative humidity) throughout the year. This observed
pattern is explained in the next sections. The day-to-day
variations, which are shown in Fig. 3b, indicate an
increase in the PM2.5/PM10 ratio on the weekend (main-
ly on Sundays), which can be attributed to a decrease in
the re-suspension of coarse particles and a decrease in
emissions resulting from less vehicular traffic. These
patterns could be explained by the most intense vehic-
ular activity occurring on working days, which would
support the high concentrations of coarse particles and
the fine particles that remain in the atmosphere for a
longer time.
In the case of daytime variation (Fig. 3c), a pro-
nounced variation in the PM2.5/PM10 ratio is observed
at most of the study sites, with a peak in the morning
between 02:00 and 05:00 h. Then, a second peak of
lower intensity is observed between 22:00 and 24:00 h.
The hourly variation is explained by the fact that during
the first hours of the day (2:00 and 4:00 h), vehicle
activity is minimal, leading to decreased coarse PM, as
opposed to fine PM, which remains relatively constant
given its smaller size. In the morning hours, vehicular
emissions during rush hour are associated with signifi-
cant emissions of PM2.5, which cause the ratio to rise. In
the afternoon, the development of activities leads to
increases in re-suspension emissions, which contribute
to PM10 emissions, thus reducing the ratio. The noctur-
nal peak (19:00 and 22:00) can be explained by the
deposition of PM10 (Bernardoni et al. 2011).
PM2.5/PM10 ratio and meteorological variables
Regarding the monthly variation, the relationship is
influenced primarily by the meteorological conditions
of the MALC. Figure 4 shows the effect of the meteo-
rological variables: wind direction (wd), wind speed
(ws), temperature (T), and relative humidity (RH) on
the PM2.5/PM10 ratio for all sites under study by month.
A bivariate polar plot of the ws and wd and the
PM2.5/PM10 ratio with all available data for the period
under study is shown in Fig. 4a. The figure shows that,
despite having higher ws values, such as those observed
in the months of October to February, PM 10 re-
suspension and the environmental concentration of this
fraction are favored; thus, the observed PM2.5/PM10
ratio should decrease in value (blue to green color in
Fig. 4a). In contrast to lower ws values observed in the
635 Page 10 of 18 Environ Monit Assess (2017) 189:635

Fig. 3 Temporal variation in the (a) Monthly variation

concentrations of the PM2.5/PM10 VMT ATE PPD HCH SJL STA CBR SBJ CDM SMP
ratio at monthly, daily, and hourly 0.60
scales

Ratio PM2.5/PM10
0.50
0.50

0.40

0.40
0.30

0.20 0.30

Ju y

Ju y
Se g

Se g
Mb

Mb
A r

ov

A r

D v
n

p
n

n
pr

ec

pr

ec
N t

N t
A l

A l
a

a
c

c
Ju

Ju
a

o
u

u
Fe

Fe
Ja

Ja
O

O
M

M
D
Month (Mmm)

(b) Daily variation

VMT ATE PPD HCH SJL STA CBR SBJ CDM SMP

Ratio PM2.5/PM10
0.45

0.40
0.45
0.35

0.30
0.40
0.25

0.20

ed

n
on
ed

e
n
on

t
e
e

t
e

i
i

Sa

Su
Fr
Sa

Th
Su
Fr

Tu
Th
Tu

W
W

M
M

Weekday (Day)

(c) Hourly variation

VMT ATE PPD HCH SJL STA CBR SBJ CDM SMP

0.60
Ratio PM2.5/PM10

0.50

0.50
0.40

0.30 0.40

0.20
0.30
0

0
0

:0

:0

:0

:0

:0
:0

:0

:0

:0

:0

00

06

12

18

22
00

06

12

18

22

Local Time (HH:mm)

months of March to September, the PM10 re-suspension
is unfavorable, and therefore, the PM2.5/PM10 ratio ob-
served should increase in value (green to red color in
Fig. 4a). These plots account for the possible transport
phenomena. In this sense, the maximum PM2.5/PM10
ratio is observed when ws is lower, which is indicative,
in general, of local emission sources for PM2.5 with
respect to PM10. Furthermore, stagnant wind conditions
allow air pollutants to accumulate, resulting in elevated
and localized concentrations of air pollutants
(DeGaetano and Doherty 2004).
The correlation among T, RH vs. PM2.5/PM10 ratio
for all sites under study by month is shown in Fig. 4b, c.
In general, an inversely proportional relationship be-
tween T and the PM2.5/PM10 ratio (see Fig. 4b) is
observed; i.e., at a lower temperature, there is a tendency
for an increase in the PM2.5/PM10 ratio. Conversely, a
direct relationship between RH and the PM2.5/PM10
ratio (in Fig. 4c) is observed; i.e., an RH increase is
coincident with an increase in the PM2.5/PM10 ratio.
These behaviors of T and HR vs the PM2.5/PM10 ratio
are related to where higher T and lower HR (in summer
compared to autumn-winter) enhance the transport pro-
cess of coarse particles due to strong winds that re-
suspend and keep the coarse particles in the atmosphere
generated by vehicular activity or dust lifted by the wind
Environ Monit Assess (2017) 189:635 Page 11 of 18 635

(a) Jan Feb Mar Apr May Jun

N N N N N N
8 8 8 8 8 8
Bivariate polar plot

4 4 4 4 4 4
W W W W W W
E E E E E E

S S
Jul Aug
N N
8 8
4 4
W W
E E
S S
Sep Oct
N N
8 8
4 4
W W
E E
S S
Nov Dic
N N
8 8
4 4
W W
E E

S S S S S S
PM2.5/PM10 Ratio
(b) Jan Feb Mar Apr May Jun

0.8
PM2.5/PM10 Ratio

0.6
0.4
0.2

Jul Aug Sep Oct Nov Dic

0.8
0.6
0.4
0.2
Counts (Nº)
15 20 25 30 15 20 25 30 15 20 25 30 15 20 25 30 15 20 25 30 15 20 25 30

Temperature (T, ºC)

(c)
Jan Feb Mar Apr May Jun

0.8
PM2.5/PM10 Ratio

0.6
0.4
0.2

Jul Aug Sep Oct Nov Dic

0.8
0.6
0.4
0.2
Counts (Nº)
50 60 70 80 90 50 60 70 80 90 50 60 70 80 90 50 60 70 80 90 50 60 70 80 90 50 60 70 80 90

Relative humidity (RH, %)

Fig. 4 Effect of the meteorological variables (wind direction, ratio. b The relationship between relative humidity and c temper-
wind speed, temperature, and relative humidity) on the PM2.5/ ature versus the PM2.5/PM10 ratio for all sites under study by
PM10 ratio. a Bivariate polar plot of the wind speed (ms−1) and month
direction for the 1-h mean mass concentrations of the PM2.5/PM10

(Jelić and Bencetić 2010). On the one hand, the period removal processes of the coarse particles due to wet
of lower T and higher HR (autumn-winter) favors the deposition (Molina et al. 2017; Toro et al. 2015). These
formation of fine particles by gas-particle conversion patterns of variability for T and HR are associated with a
processes, and on the other hand, this period favors the phenomenon of inversion by subsidence and marine
635 Page 12 of 18
thermal inversion because of the influence of the Cor-
dillera de los Andes, the Humboldt cool oceanic current,
and the South Pacific anticyclone. During the summer,
smaller and less permanent marine thermal inversion
occurs mostly below 500 m due to the combined effects
of subsidence and oceanic cooling. This weak marine
thermal inversion causes a density decrease in stratiform
clouds, with the consequent increase of solar irradiation,
which leads to a decrease in HR and an increase of T at
the surface, resulting in turbulent processes that lead to
re-suspension of coarse PM, and at the same time pre-
vents the formation of secondary particulate material
(Arellano Rojas 2013). In contrast, the stratiform cloud-
iness increases during the winter, as does the HR, and
the occurrence of drizzle is more frequent; these condi-
tions favor a significant decrease in PM10 concentrations
due to wet deposition and an increase in PM2.5 concen-
trations due to the formation of secondary aerosols via
the process of converted gas-particulate (more details
are provided in Fig. S1).
Trend analysis
The temporal trends were calculated for the stations with
the highest number of PM10 measurement years, i.e., the
ATE, VMT, STA, SBJ, and CDM stations. The results
show that at the 95% confidence level, there is a signif-
icant trend of decreasing PM10 concentrations at the
ATE and STA stations. The trend indicates a decrease
on the order of −4.3 (CI −5.3, −3.2)% year−1 (equivalent
to −5.51 (CI −7.1, −4.0) μg m−3 year−1) for the ATE
station and on the order of −5.3 (CI −8.1, −1.6)% year−1
(equivalent to −4.7 (CI −7.5, −1.3) μg m−3 year−1) for
the STA station. On the other hand, at the other stations
at which it was possible to calculate the Telson estimator
(i.e., the VMT, SBJ, and CDM stations), the results
indicate that the trend is not significant at the 95%
confidence interval, which means that the confidence
interval exhibits a slope equal to zero. The results ob-
tained for each of the stations are as follows: VMT: 6.3
(CI −4.5, 17.6)% year−1, SBJ: −5.4 (CI −8.1, −1.6)%
year−1, and CDM: −1.5 (CI −3.5, 0.9)% year−1.
These results indicate that control measures such as
street paving, exclusive roads for mass public transport,
incentives for the use of natural gas, afforestation pro-
grams, restrictions on the import of used vehicles, and
required vehicular technical inspections have led to
reductions in PM10 concentrations in high-elevation
areas of the city, mainly residential sectors, where the
Environ Monit Assess (2017) 189:635
ATE and STA stations are located. In the rest of the city,
there was no evidence of improvement in terms of trends
of reductions in the PM10 concentration. Here, it should
be noted that in the MALC, there are approximately 2.2
million vehicles, almost 70% of the total number of
vehicles in the nation, which explains why the greatest
contribution of this pollutant is associated with the au-
tomobile fleet (SENAMHI 2014).
Air pollution assessment
Figure 5 shows a comparison of the situation of the
MALC relative to other world cities. Figure 5 shows
the PM10 and PM2.5 concentrations of different cities
around the world according to the 2014 WHO database
(WHO 2016b), the PM concentration in the MALC, and
a statistical summary of concentrations (maximum, min-
imum, mean, and standard deviation); see Fig. 5a. A
comparison with megacities around the world is also
presented. Figure 5b shows the list of megacities listed
in the WHO database and the gross domestic product of
the respective country to which the megacity belongs
(WB 2017).
When comparing the MALC with the cities listed in
the WHO database, it can be observed that it ranks in
positions 202 and 397 among more than 1500 cities in
terms of PM10 and PM2.5, respectively (see Fig. 5a). The
PM10 levels observed are approximately 1.95 times the
average concentration of all cities listed in the database.
The PM10 concentrations are similar to those observed
in Linfen, China (84 μg m −3 ); Changzhi, China
(83 μg m−3); Muscat, Oman (82 μg m−3); Nalbari, India
(82 μg m−3); Mianyang, China (82 μg m−3); Tripoli,
Libya (81 μg m−3); Bhilai, India (81 μg m−3); Kolhapur,
India (81 μg m−3); Jilin, China (81 μg m−3); Toluca,
Mexico (80 μg m−3); Shanghai, China (79 μg m−3); and
Fresno, USA (74 μg m−3). The PM2.5 concentration in
the MALC corresponds to approximately 1.2 times the
average concentration of all cities in the database. Its
concentration is similar to those observed in Ostrow,
Poland (28 μg m−3); Verona, Italy (28 μg m−3); Shantou,
China (28 μg m−3); Napoli, Italy (27 μg m−3); Rio de
Janeiro, Brazil (27 μg m −3 ); Budapest, Hungary
(27 μg m−3); Medellin, Colombia (27 μg m−3); and
Bogota, Colombia (27 μg m−3).
When comparing the MALC with global megacities,
i.e., cities of more than 10 million inhabitants, we can
see that Lima is located in positions 12 and 16 in terms
of PM10 and PM2.5, respectively, out of 39 megacities
Environ Monit Assess (2017) 189:635 Page 13 of 18 635

(a) Ranking in WHO urban outdoor air pollution database 2014

600

500
PM10 PM25
NºCities: 1524 NºCities: 1615
Max: 540 µgm -3 Max: 153 µgm -3
PM10 (µgm -3)
PM2.5 (µgm -3)

400
Median: 25 µgm -3 Median: 16 µgm -3
Min: 6 µgm -3 Min: 2 µgm -3
300 Avg: 42 µgm -3 Avg: 22 µgm -3
SD: 44 µgm -3 SD: 18 µgm -3
LMA: 82 µgm -3 LMA: 27 µgm -3
200 Ranking LMA: 203 Ranking LMA: 397

200

0 200 400 600 800 1000 1200 1400 1600

Ranking

(b) PM concentration of mega and big cities

Country
TWN
BGD

NGA

KOR

ARG

GBN
CHN

MEX

CHN

CHN
BRA

CHN

CHN

BRA

GER
COL

USA

USA

USA
USA
CHL

PER

TUR

THA

FRA
PAK

PHL

ESP
JPN

JPN
JPN
IND

IND

IND
IND

IRN
IND

IND

IND

0
City Population
(nº x 106)

10

20

30

40
300

National Gross Domestic Product

60

per capita (USD x 1000)

PM10 (µgm -3)
PM2.5 (µgm -3)

200
40

100
20

0 0
Karachi
Dhaka
Mumbai
Lagos
Beijing
Bangalore
Kolkata
Mexico City
Tehran

Sao Paulo

Buenos Aires
Hyderabad
Shanghai
Santiago
Guangzhou
Rio de Janeiro
Lima
Taipei
Shenzhen
Chennai
Seoul
Manila
Bogotá
Istanbul

Hong Kong
Jakarta

Bangkok

Los Angeles

Rhine-Ruhr
Paris
Tokyo
New York
Madrid
London
Chicago
Dallas
Osaka
Nagoya
New Delhi

City
Fig. 5 a Concentrations rankings of particulate matter (PM10 and left) per capita GDP (bottom right) and the number of inhabitants
PM2.5) in different cities around the world (WHO 2016b). b of the city (top) for global megacities (WB 2017).
Comparison of annual PM10 and PM2.5 concentrations (bottom
635 Page 14 of 18
(see Fig. 5b). It is important to note that the city in the
13th ranking is Shanghai, China, which has more than
double the population of the MALC. The relationship
between the degree of contamination and economic
development (see Fig. 5b) shows that at lower gross
domestic product (GDP) per capita, in general, a higher
concentration of PM10 and PM2.5 is observed. This
accounts for the possibility of investing in pollution
reduction policies that countries with greater economic
development have. The MALC currently has a GDP
lower than USD 10,000, which corresponds to the
GDP of the countries in which the megacities with the
highest pollution are located.
According to the WHO thresholds (GD-WHO) and
Peruvian regulation (PE-MINAM), the average annual
concentration of PM10 in the MALC exceeds the limits
of 20 and 50 μg m−3 for PM10 (see Table 2). Thus, the
WHO recommendation is exceeded by more than a
factor of 2 at all stations and by more than a factor of
5 at some of them (ATE, VMT, and PPD). The national
standard is exceeded at a minimum of seven stations of
the MALC by factors between 1.6 and 2.4; the excep-
tions are the SBJ, CDM, and SMP stations. For PM2.5,
the WHO threshold is 10 μg m−3 on an annual basis,
which is exceeded at all stations by factors of between
1.8 and 4; the station that most exceeds the threshold is
ATE (4.0 times), with CDM (1.6 times), SBJ and SMP
(1.8 times) exceeding by a lesser degree.
Table 3 lists the percentage of days for which the PE-
MINAM and WHO thresholds for the PM10 and PM2.5
concentrations are exceeded. For this calculation, the
24-h arithmetic average is used.
The results indicate that for PM10, the GD-WHO
limits are systematically exceeded at all the stations
of the monitoring network. In general, the percent-
age of days per year in which the 24-h average daily
concentration of PM10 is greater than 50 μg m−3 is
50%. The stations with the highest average number
of days for which the PM10 concentration exceeds
the GD-WHO threshold are the ATE (97%), VMT
(91.7%), PPD and SJL (100%), HCH (96.6%), CRB
(90.8%), and STA (89.8%) stations. In the case of
the national standard for PM 1 0 (PE-MINAM
150 μg m−3), seven of the ten stations of the air
quality network studied exceed the threshold; the
exceptions are the SBJ, CDM, and SMP stations
(0% exceeded days). The number of days for which
the threshold is exceeded on average by station is,
from highest to lowest, as follows: VMT (29.5%),
Environ Monit Assess (2017) 189:635
ATE (13.2%), PPD (12.7%), HCH (4.1%), SJL
(1.54%), and CRB (0.37%).
In the case of the PM2.5 concentration, GD-WHO
and PE-MINAM both have a daily value of 25 μg m−3.
Regarding PM10, the results indicate that the GD-WHO
and PE-MINAM thresholds are systematically exceeded
at all stations under study. The percentage of days for
which the GD-WHO threshold is exceeded for each
station, from highest to lowest, is as follows: ATE
(98.3%), PPD (96.6%), SJL (89.3%), HCH (79.9%),
CRB (77.8%), STA (77.5%), SBJ (31.1%), SMP
(27.7%), and CDM (17.6%). In general, the number of
days for which the GD-WHO threshold is exceeded
corresponds to at least 50% of the days of the year at
five stations. Similarly, in the case of the PE-MINAM
threshold, it is observed that all stations exceed the daily
standard. The number of days for which the threshold is
exceeded at each station, from highest to lowest, is as
follows: ATE (99.6%), PPD (86.8%), SJL (74.4%),
HCH (65.3%), CRB (63.1%), STA (62.0%), SBJ
(11.6%), SMP (9.23%), and CDM (4.00%). At the
stations with the greatest number of days for which the
threshold is exceeded, the standard is exceeded at least
30% of the days of the year.
Finally, it is important to note that a large part of the
population of Lima is exposed to concentrations that
exceed the protection thresholds. Hence, the develop-
ment of control measures and reduction of pollution are
essential to reduce pollution by PM. The government of
Peru has developed a pollution control and prevention
plan that aims to improve air quality in the MALC and
considers measures related to reduction of vehicular
emissions, exclusive roads, road improvements, auto-
mobile fleet renewal, industrial emissions reduction,
improvement of fuels, incentive for the use of clean
fuels, increased green areas, and green taxes, among
other measures (CIAL 2010).
Conclusions
The urban atmosphere of the MALC is contaminated by
PM10 and PM2.5. The annual average PM10 concentra-
tions range from 133 to 45 μg m−3, with a mean of
82 μg m−3, for the period from 2010 to 2015. The
highest annual PM10 concentrations are observed at
monitoring stations located to the east of the city, name-
ly, ATE (113 μg m−3), VMT (112 μg m−3), PPD
(111 μg m−3), HCH (96 μg m−3), SJL (85 μg m−3),
Environ Monit Assess (2017) 189:635 Page 15 of 18 635

Table 3 Exceedances (in % of days per year) of the 24-h Peruvian air quality standards (PE-MINAM) and World Health Organization
guidelines (GD-WHO) at the monitoring sites under study

Year ATE SBJ CDM STA VMT HCH SJL SMP CBR PPD

PM10
舃

GD-WHO
舃

2010
舃 99.6
舃
a
舃
a
舃 –
舃 –
舃 –
舃 –
舃 –
舃 –
舃 –
舃

2011
舃 98.6
舃 49.1
舃 27.6
舃
a
舃 –
舃 –
舃 –
舃 –
舃 –
舃 –
舃

2012
舃 98.9
舃 45.1
舃 30.2
舃 91.8
舃 90.9
舃 –
舃 –
舃 –
舃 –
舃 –
舃

2013
舃 91.3
舃 56.6
舃 40.3
舃 96.5
舃 91.9
舃 –
舃 –
舃 –
舃 –
舃 –
舃

a a
2014
舃 99.4
舃 43.6
舃 23.7
舃 舃 舃 97.6
舃 95.3
舃 40.8
舃 89.3
舃 98.8
舃

a a
2015
舃 94.8
舃 舃 22.6
舃 82.4
舃 舃 96.1
舃 83.6
舃 35.2
舃 92.1
舃 100
舃

Avg
舃 97.0
舃 49.1
舃 29.0
舃 89.8
舃 91.6
舃 96.6
舃 100.0
舃 38.3
舃 90.8
舃 100
舃

PE-MINAM
舃

2010
舃 17.2
舃
a
舃
a
舃 –
舃 –
舃 –
舃 –
舃 –
舃 –
舃 –
舃

2011
舃 17.4
舃 0.00
舃 0.00
舃
a
舃 –
舃 –
舃 –
舃 –
舃 –
舃 –
舃

2012
舃 12.7
舃 0.00
舃 0.00
舃 1.37
舃 23.9
舃 –
舃 –
舃 –
舃 –
舃 –
舃

2013
舃 16.3
舃 0.00
舃 0.00
舃 9.29
舃 34.4
舃 –
舃 –
舃 –
舃 –
舃 –
舃

a a
2014
舃 11.6
舃 0.00
舃 0.00
舃 舃 舃 6.3
舃 1.16
舃 0.00
舃 0.79
舃 7.91
舃

a a
2015
舃 4.43
舃 舃 0.00
舃 0.00
舃 舃 2.2
舃 1.34
舃 0.00
舃 0.00
舃 16.7
舃

Avg
舃 13.2
舃 0.00
舃 0.00
舃 3.03
舃 29.5
舃 4.1
舃 1.54
舃 0.00
舃 0.37
舃 12.7
舃

PM2.5
舃

GD-WHO
舃

a a a a a
2014
舃 舃 舃 舃 舃 舃 100
舃 94.6
舃 33.8
舃 96.0
舃 99.2
舃

a
2015
舃 98.3
舃 31.1
舃 17.6
舃 77.5
舃 舃 62.3
舃 84.7
舃 20.7
舃 59.9
舃 94.2
舃

a
Avg
舃 98.3
舃 31.1
舃 17.6
舃 77.5
舃 舃 79.9
舃 89.3
舃 27.7
舃 77.8
舃 96.6
舃

PE-MINAM
舃

a a a a a
2014
舃 舃 舃 舃 舃 舃 89.1
舃 77.9
舃 10.8
舃 78.5
舃 89.6
舃

a
2015
舃 99.6
舃 11.6
舃 4.00
舃 62.0
舃 舃 41.9
舃 60.3
舃 7.17
舃 47.6
舃 84.5
舃

a
Avg
舃 99.6
舃 11.6
舃 4.00
舃 62.0
舃 舃 65.3
舃 74.4
舃 9.23
舃 63.1
舃 86.8
舃

(−) No available data

a
Concentrations with low data availability

STA (82 μg m−3), and CRB (79 μg m−3). In contrast, the
lowest annual concentrations for PM10 are observed at
the stations located west of the city, namely, SBJ
(52 μg m−3), SMP (48 μg m−3), and CDM (45 μg m−3).
For PM2.5, the average annual concentrations range
from 35 to 16 μg m−3, with a mean of 27 μg m−3, in
the period under study. In terms of annual PM2.5 con-
centration, from highest to lowest, the stations are
ranked as follows: ATE (35), PPD (32), HCH (31),
SJL (31), CRB (28), STA (27), SBJ (18), SMP (18),
and CDM (16). In general, this pattern of distribution of
PM concentrations, i.e., greater in the eastern part of the
city, is related to the characteristic wind pattern of the
city, which has prevailing winds from the south and
southwest. The variation in the ratio (PM2.5/PM10)
ranges from 0.21 to 0.44, resulting in an average con-
centration of PM2.5 representing 40% of that of PM10.
This result indicates the presence of significant emitter
sources, mainly of coarse particles, which come from
the extensive uncovered areas, where re-suspension oc-
curs owing to winds and intense vehicular activity, in
addition to activities of production, transportation, and
commercialization of cement, brick, aggregates, debris,
etc.
The highest monthly concentrations of PM10 are
observed in the summer and early autumn (February–
April). The minimum monthly concentrations are ob-
served in late autumn and early winter (May to
635 Page 16 of 18
September). On the other hand, regarding the average
monthly concentrations of PM2.5, the highest concentra-
tions are recorded between late autumn and winter (May
and September), and the minimum occurs between mid-
spring and early autumn (October–April). This temporal
pattern occurs because subsidence thermal inversion
weakens in the middle of spring and early fall and
because the humidity decreases, which is detrimental
to the formation of secondary particulate matter and
contrary to what occurs during the cold and humid
period (May–September). Regarding the hourly vari-
ability of the daily concentrations of both particle sizes,
it was possible to identify two periods of maximum
concentration mainly influenced by vehicular activity
in the MALC, one between 6:00 and 10:00 h and the
second between 18:00 and 23:00 h. This pattern can be
explained if one considers that the MALC has 2.2
million motor vehicles, which make a total of nine
million trips a day; thus, the maximum vehicular emis-
sions are generated during the peak hours of traffic in the
city in the morning and evening.
From the long-term trend analysis based on the
Theil-Sen estimator, there is a trend of decreasing
PM10 concentrations at ATE (5.51 μg m−3 year−1)
and STA (4.7 μg m−3 year−1). However, to estimate
the long-term trends, continuity in the measurement of
PM environmental concentrations is required. The
beginning of the installation of the monitoring net-
work has made it possible to perform these first eval-
uations, which will be improved by having in the
future more and better information from the stations
of the air quality monitoring network.
The MALC is one of the most polluted cities in the
world. The MALC is ranked 202 out of 1524 cities and
ranked 397 out of 1615 cities with the highest envi-
ronmental concentrations listed in the WHO database
for PM10 and PM2.5, respectively. The observed con-
centration for PM10 (84 μg m−3) in the MALC is
approximately twice the average concentration of all
cities listed in the WHO databases, whereas the aver-
age concentration of PM2.5 corresponds to 1.2 times
the average concentrations of cities listed in the
database.
According to WHO thresholds (GD-WHO) and
Peruvian regulations (PE-MINAM), the annual and
daily threshold concentrations for more PM fractions
will be systematically exceeded in the MALC. For the
WHO thresholds, the fractions of days exceeded per
year for PM10 and PM2.5 were greater than 65%. The
Environ Monit Assess (2017) 189:635
stations that achieved the greatest number of days of
exceedance of the WHO thresholds were the ATE
(287 days) and CDM (86 days) stations. For PM2.5,
these stations were PPD (256 days) and CDM
(41 days). On the other hand, the maximum rates of
exceedances for both PM fractions with respect to PE-
MINAM regulations were approximately 7% for
PM 10 and more than 70% for PM 2.5 , with VMT
(88 days) being the station with the greatest number
of days of exceedance of the national standard of air
quality for PM10. For PM2.5, all the stations under
study exceed the standard, reaching a maximum of
days of exceedance of the standard at ATE (235 days)
and a minimum number of days of exceedance at
CDM (10 days). Although the national standards and
WHO thresholds have been exceeded, a tendency has
been observed for some stations to exhibit a decrease
in the PM10 concentration (namely, the ATE and STA
stations).
In consideration of the above, it is important to
understand that the improvement in air quality is due
to different mitigation measures that must be applied
systematically over time. However, often, these mea-
sures are too late, and the problem of contamination
could be intensified. At the global level, different
governments have implemented mitigation and pre-
vention measures for decontamination, such as renew-
al of the public transit system, updating of emission
standards, enforcement of regulations to certify emis-
sions from stationary sources, mobile and vehicular
restrictions during critical episodes of pollution, af-
forestation and creation of green areas, paving pro-
grams, and street washing. In general, these measures
are applied in the MALC. For example, exclusive
roads have been developed for public transit, and a
metro is currently under construction in the city. Like-
wise, reforestation programs and the promotion of
natural gas-fueled vehicles have been developed in
the city. Notwithstanding the above in the city and
given the evaluation performed in the present work,
it is necessary to strengthen and extend the application
of policies and strategies for air quality management,
including strategic planning, design of sustainable
urban areas, periodic evaluation of past policies, and
strengthening association with health impacts and
many others. Likewise, adequate monitoring and eval-
uation of the measures applied should be sought, and
the population should be made aware of the implica-
tions of air quality in their daily lives.
Environ Monit Assess (2017) 189:635
Acknowledgments We acknowledge the financial support of
the National Meteorology and Hydrology Service of Peru–
SENAMHI-Perú, for project SNIP N° 199842 BExpansion and
Improvement of the Monitoring Network for the Forecasting of
Air Quality in the Metropolitan Area of Lima^ and Program
096—PPR096-Air Quality Management. MALG acknowledges
the support of the National Commission for Scientific and Tech-
nological Research CONICYT/FONDECYT 2016 grant no.
1160617. RAT acknowledges the partial support of the National
Commission for Scientific and Technological Research
CONICYT/FONDECYT INICIACION 2015 grant no.
11150931. The funders had no role in the study design, data
collection and analysis, and decision to publish or the preparation
of the manuscript.
Compliance with ethical standards
Competing interests The authors declared that they have no
competing interests.
References
Arellano Rojas, C. S. (2013). Meteorological conditions and pol-
lution levels in the metropolitan region of Lima - Peru (in
Portuguese). Master Thesis. Institute of Astronomy,
Geophysics and Atmospheric Sciences, Universidade de
São Paulo, São Paulo, Brazil. [WWW Document]. URL
https://goo.gl/qw9BOv (accessed 3.8.17).
Awe, Y., Nygard, J., Larssen, S., Lee, H., Dulal, H., Kanakia, R.,
(2015). Clean air and healthy lungs: enhancing the World
Bank’s approach to air quality management (No. No3),
Environment and natural resources global practice discussion
paper. Washington, DC.
Baklanov, A., Molina, L. T., & Gauss, M. (2016). Megacities, air
quality and climate. Atmospheric Environment, 126, 235–
249. https://doi.org/10.1016/j.atmosenv.2015.11.059.
Baldasano, J., Valera, E., & Jimenez, P. (2003). Air quality data
from large cities. Sci. Total Environ., 307, 141–165.
https://doi.org/10.1016/S0048-9697(02)00537-5.
Bell, M. L. L., Davis, D. L. L., Gouveia, N., Borja-Aburto, V. H.
H., & Cifuentes, L. A. A. (2006). The avoidable health effects
of air pollution in three Latin American cities: Santiago, São
Paulo, and Mexico City. Environmental Research, 100, 431–
440. https://doi.org/10.1016/j.envres.2005.08.002.
Bernardoni, V., Vecchi, R., Valli, G., Piazzalunga, A., & Fermo, P.
(2011). PM10 source apportionment in Milan (Italy) using
time-resolved data. Sci. Total Environ., 409, 4788–4795.
https://doi.org/10.1016/j.scitotenv.2011.07.048.
Carbajal-Arroyo, L., Barraza-Villarreal, A., Durand-Pardo, R.,
Moreno-Macías, H., Espinoza-Laín, R., Chiarella-Ortigosa,
P., & Romieu, I. (2007). Impact of traffic flow on the asthma
prevalence among school children in lima, Peru. The Journal
of Asthma, 44, 197–202. https://doi.org/10.1080
/02770900701209756.
Carslaw, D. C. (2013). The openair manual—open-source tools
for analysing air pollution data. Manual for version 0.9–0,
Page 17 of 18 635
King’s College London. London, UK. [WWW Document].
URL https://goo.gl/iigCI9 (accessed 3.8.17).
Carslaw, D. C., & Ropkins, K. (2012). Openair—an R package for
air quality data analysis. Environ. Model. Softw., 27–28, 52–
61. https://doi.org/10.1016/j.envsoft.2011.09.008.
CIAL. (2010). Second Integral Plan of Atmospheric Sanitation for
Lima-Callao 2011–2015 (in spanish). Management
Committee Clean Air Initiative (IAL) Lima-Callao, Lima,
Peru. [WWW Document]. URL https://goo.gl/O9eaB3.
Dawidowski, L., Sánchez-Ccoyllo, O., Alarcón, N.. (2014).
Estimation of vehicular emissions in Metropolitan Lima.
Final report (in Spanish). Lima, Peru: Servicio Nacional de
Meteorología e Hidrología del Perú (SENAMHI). Report
within the framework of the project South American
Emissions, Megacities and Climate.
DeGaetano, A. T., & Doherty, O. M. (2004). Temporal, spatial and
meteorological variations in hourly PM2.5 concentration ex-
tremes in new York City. Atmospheric Environment, 38,
1547–1558. https://doi.org/10.1016/j.atmosenv.2003.12.020.
EPA-US. (2009). Integrated Science Assessment (ISA) for
Particulate Matter (PM). U.S. Environmental Protection
Agency, Washington, DC, USA. EPA/600/R-08/139F.
[WWW Document]. URL https://goo.gl/PCnvbj (accessed
3.13.17).
EPA-US, 2013. National Ambient Air Quality Standards for
Particulate Matter; Final Rule. 78 Federal Register 3086
(Document No2012–30946), Office of Air Quality Planning
and Standards, United States Environmental Protection
Agency, Research Triangle Park, NC, USA. [WWW
Document]. URL https://goo.gl/3KB3SX.
IGN. (2014). National Geographic Institute, Republic of Peru
[WWW Document]. URL http://www.ign.gob.pe (accessed
3.13.17).
Jelić, D., Bencetić, Z. (2010). Air quality in Rijeka, Croatia 27.
Konstantinos, M. (2008). Quantification and evaluation of dust
resuspension pm10 emissions in two large urban centers in
Greece Quantifica, 543–547.
Lelieveld, J., Evans, J. S., Fnais, M., Giannadaki, D., & Pozzer, A.
(2015). The contribution of outdoor air pollution sources to
premature mortality on a global scale. Nature, 525, 367–371.
LR. (2010). Senamhi inaugurates First Air Quality Monitoring
Station in Ate | News from Peru | [WWW Document].
TheRepublica.pe. URL https://goo.gl/5k3qc8 (accessed
3.14.17).
Molina, L. T., & Molina, M. J. (2002). Air quality impacts: local
and global concern. In L. T. Molina & M. J. Molina (Eds.),
Air quality in the Mexico megacity: an integrated assessment
(pp. 1–19). Dordrecht: Springer Netherlands. https://doi.
org/10.1007/978-94-010-0454-1_1.
Molina, M. J. J., & Molina, L. T. T. (2004). Megacities and
atmospheric pollution. Journal of the Air & Waste
Management Association (1995), 54, 644–680.
Molina, C., Toro A. R., Morales, S. R. G. E., Manzano, C., Leiva-
Guzmán, M. A. (2017). Particulate matter in urban areas of
south-central Chile exceeds air quality standards. Air Qual.
Atmos. {&} Heal. 1–15. doi:https://doi.org/10.1007/s11869-
017-0459-y.
MTC. (2016). Estimated National Vehicle Park, according to
Department: 2007–2016. Lima, Peru.
PE-MINAM. (2001). Supreme Decree No. 074–2001-PCM
Regulation of National Environmental AIr Quality
635 Page 18 of 18
Standards (in spanish), Ministry of the Environment
(MINAM), Republic of Peru. [WWW Document]. doi:D.S
No 074–2001 –PCM.
PE-MINAM. (2014). National air quality report 2013–2014 (in
spanish). Ministry of the Environment (MINAM), Republic
of Peru. [WWW Document]. URL https://goo.gl/czqcNg
(accessed 3.15.14).
Reich, S., Robledo, F., Gomez, D., & Smichowski, P. (2008). Air
pollution sources of PM10 in Buenos Aires City.
Environmental Monitoring and Assessment, 155, 191.
https://doi.org/10.1007/s10661-008-0428-x.
RStudio. (2016). RStudio: Integrated development environment
for R [Computer software].
Sen, P. K. (1968). Estimates of the regression coefficient based on
Kendall’s tau. Journal of the American Statistical
Association, 63, 1379–1389. https://doi.org/10.2307
/2285891.
SENAMHI. (2014). Estimation of vehicular emissions in
Metropolitan Lima - Final report (in spanish). National
Service of Meteorology and Hydrology of Peru
(SENAMHI), Ministry of the Environment (MINAM),
Republic of Perú. [WWW Document]. Natl. Serv.
Meteorol. Hydrol. Peru (SENAMHI), Minist. Environ.
(MINAM), Repub. Perú.
SENAMHI. (2016). Air quality assessment in Lima metropolitan
area 2015 (in spanish). National Service of Meteorology and
Hydrology of Peru (SENAMHI), Ministry of the
Environment (MINAM), Republic of Peru. [WWW
Document]. URL https://goo.gl/db2JgU (accessed 3.8.17).
Stanek, L. W., Brown, J. S., Stanek, J., Gift, J., & Costa, D. L.
(2011). Air pollution toxicology—a brief review of the role
of the science in shaping the current understanding of air
pollution health risks. Toxicological Sciences, 120, S8–S27.
Tager, I. (2012). Health Effects of Aerosols. In Aerosols Handbook
(pp. 565–636). CRC Press. doi:https://doi.org/10.1201
/b12668-24.
Theil, H. (1992). A rank-invariant method of linear and polyno-
mial regression analysis. In B. Raj & J. Koerts (Eds.), Henri
Theil’s contributions to economics and econometrics:
Econometric theory and methodology (pp. 345–381).
Dordrecht: Springer Netherlands. https://doi.org/10.1007
/978-94-011-2546-8_20.
Toro, A. R., Morales, S. R. G. E., Canales, M., Gonzalez-Rojas,
C., & Leiva, G. M. A. A. (2014). Inhaled and inspired
particulates in metropolitan Santiago Chile exceed air quality
standards. Building and Environment, 79, 115–123.
https://doi.org/10.1016/j.buildenv.2014.05.004.
Toro, A. R., Córdova, J. A., Canales, M., Morales, S. R. G. E.,
Mardones, P. P., Leiva, G. M. A., 2015. Trends and threshold
Environ Monit Assess (2017) 189:635
exceedances analysis of airborne pollen concentrations in
Metropolitan Santiago Chile. PLoS One 10. doi:https://doi.
org/10.1371/journal.pone.0123077.
UN. (2016). The World’s Cities in 2016 – Data Booklet (ST/ESA/
SER.A/392). United Nations, Department of Economic and
Social A airs, Population Division, New York, NY, USA.
[WWW Document]. URL https://goo.gl/pVdtwy (accessed
3.8.17).
Underhill, L., Bose, S., Williams, D., Romero, K., Malpartida, G.,
Breysse, P., Klasen, E., Combe, J., Checkley, W., & Hansel,
N. (2015). Association of Roadway Proximity with indoor air
pollution in a Peri-Urban Community in lima, Peru.
International Journal of Environmental Research and
Public Health, 12, 13466–13481. https://doi.org/10.3390
/ijerph121013466.
Vahlsing, C., & Smith, K. R. (2012). Global review of national
ambient air quality standards for PM(10) and SO(2) (24 h).
Air Quality, Atmosphere and Health, 5, 393–399. https://doi.
org/10.1007/s11869-010-0131-2.
Valavanidis, A., Vlachogianni, T., Fiotakis, K., & Loridas, S.
(2013). Pulmonary oxidative stress, inflammation and can-
cer: Respirable particulate matter, fibrous dusts and ozone as
major causes of lung carcinogenesis through reactive oxygen
species mechanisms. International Journal of Environmental
Research and Public Health, 10, 3886–3907. https://doi.
org/10.3390/ijerph10093886.
WB. (2017). GDP Online Database. World Bank, Washington,
DC, USA. [WWW Document]. URL https://goo.gl/Bgn2X
(accessed 3.13.17).
WHO. (2006). Air quality guidelines. Global update 2005.
Particulate matter, ozone, nitrogen dioxide and sulfur diox-
ide. World Health Organization Copenhagen, Denmark.
[WWW Document]. URL https://goo.gl/hz79bI (accessed
1.23.15).
WHO. (2016a). Fact sheet: Ambient (outdoor) air quality and
health. World Health Organization, Media Centre,
Copenhagen, Denmark. [WWW Document]. URL
https://goo.gl/iJomX (accessed 1.23.15).
WHO. (2016b). Ambient (outdoor) air pollution database, by
country and city (xlsx file). World Health Organization
(WHO), Department of Public Health, Environmental and
Social Determinants of Health. Geneva, Switzerland.
[WWW Document]. URL https://goo.gl/nPsdIq (accessed
3.8.17).
Xu, G., Jiao, L., Zhang, B., Zhao, S., Yuan, M., Gu, Y., Liu, J., &
Tang, X. (2017). Spatial and temporal variability of the
PM2.5/PM10 ratio in Wuhan, Central China. Aerosol and
Air Quality Research, 17, 741–751. https://doi.org/10.4209
/aaqr.2016.09.0406.