Atmospheric Environment 79 (2013) 299e307

Contents lists available at SciVerse ScienceDirect

Atmospheric Environment
journal homepage: www.elsevier.com/locate/atmosenv

Contribution of long and mid-range transport on the sodium and

potassium concentrations in rainwater samples, São Paulo megacity,
Brazil
Marcelo S. Vieira-Filho a, Jairo J. Pedrotti b, Adalgiza Fornaro a, *
a
Department of Atmospheric Sciences, Institute of Astronomy, Geophysics, and Atmospheric Sciences, University of São Paulo (IUSP),
Rua do Matão 1226, São Paulo 05508-090, Brazil
b
Escola de Engenharia, Universidade Presbiteriana Mackenzie, Rua Consolação 896, São Paulo 01302-907, Brazil

h i g h l i g h t s

The wet-only deposition was continuously measured for four consecutive years.

The long-range transport potassium compounds contribution was evaluated by backward trajectories.

The bias in the estimate of non-sea salt fraction for rainwater samples is discussed regarding the Cl depletion and pH.

The potassium and sodium ions in rainwater had a significant contribution from the local anthropogenic sources.

a r t i c l e i n f o a b s t r a c t

Article history: The surface winds and backward trajectories were analyzed to investigate sources of sodium and po-
Received 2 January 2013 tassium ions measured in rainwater in the metropolitan area of São Paulo (MASP), Brazil, from July-2002
Received in revised form to December-2005. In this study, we evaluated the chemical composition of 315 rainwater samples with
17 May 2013
focus on precipitation events with high concentrations of sodium and potassium, with cations concen-
Accepted 19 May 2013
trations, at least 1.5 times higher than the sulfate and nitrate concentrations. In the results from the total
samples, the Naþ and Kþ volume weighted mean (VWM) concentrations were 10.4 and 3.2 mmol L1,
Keywords:
respectively, while for the excess events, the concentrations were 32.4 and 11.1 mmol L1. From the 31
Rainwater
Sea-salt
sodium excess events, 45% showed an influence from ocean air masses by the wind direction evaluation
Biomass burning while 57% of backward trajectories for 7 potassium excess events indicated a contribution of biomass
Backward trajectories burning air masses from the northwestern region of the São Paulo State. Although some excess events
HYSPLIT pointed the pattern of long-range transport, they represented less than 15% of total samples. The results
Long-range transport suggest that the megacity itself is predominantly involved in the process of emission and removal of the
compounds and the influence of external components are limited to circumstantial events.
Ó 2013 Elsevier Ltd. All rights reserved.

1. Introduction deposition, studies have linked the rainwater chemical composition

The physical and chemical processes of the atmospheric wet
removal cycle include not only the synoptic systems but also cloud
microphysics and “below-cloud” processes (Seinfeld and Pandis,
1998; Gonçalves et al., 2010). Other factors such as topography,
mesoscale circulations and local sources differ significantly from
region to region and, therefore, also on the precipitation chemical
composition. In order to identify possible ions origins in wet
* Corresponding author. Tel.: þ55 11 3091 2819; fax: þ55 11 3091 4714.
E-mail addresses: marcelovieira@model.iag.usp.br (M.S. Vieira-Filho), jpedrotti@
mackenzie.br (J.J. Pedrotti), fornaro@model.iag.usp.br (A. Fornaro).
with weather conditions by monitoring air masses trajectories
(Samara et al., 1992; Beverland et al., 1998; Avila and Alarcon, 1999;
Dayan and Lamb, 2003; Hong et al., 2011).
Significant variations in the rainwater composition were
observed in southwest England due to different transport patterns
of air masses. Among the results, the authors highlighted the in-
verse effect on the nitrate and sulfate ratio as a function of air
masses originated from highly urbanized regions (Beverland et al.,
1998). Rainfall data collected daily during nine consecutive sum-
mers (1993e2001) showed the synoptic circulation influence in the
rainwater chemical composition in Pennsylvania, United States. The
seven types of events were classified, being higher average con-
centrations associated with warm front conditions (Dayan and

1352-2310/$ e see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.atmosenv.2013.05.047
300 M.S. Vieira-Filho et al. / Atmospheric Environment 79 (2013) 299e307
Lamb, 2003). In another study, the ionic composition of precipita-
tion was evaluated through analysis and identification of synoptic
air masses trajectories in a rural area of Spain. Alkaline rainwater
(mean pH 7.2) was associated with air masses from Africa. Neutral
rainwater (average pH 5.5) was associated with local events, and
acid rain events originated from the Atlantic Ocean (average pH 4.8)
and Europe (pH 4.4) with high ionic concentrations. However, the
acid rain represented less than 10% of total rainfall (Avila and
Alarcon, 1999). In southern Korea, the higher formate and acetate
ions concentration were associated with air masses trajectories
from the continental area, and lower concentrations with oceanic
air masses (Hong et al., 2011).
The wind sector analysis and trajectory models have been used
to assess the impact of air masses in the chemical composition of
dry and wet deposition (Fleming et al., 2012). In the last years there
have been an increasing number of studies about the air pollution
evaluation through HYSPLIT (Hybrid Single Particle Lagrangian In-
tegrated Trajectory) model, specifically connecting the backward
trajectories with analyzed wet deposition concentrations. For
instance, Morselli et al. (2008) showed the presence of Sahara dust
in particulate matter samples from Bologna, Italy. Báez et al. (2007)
evaluated the transport of trace metals to the Central Mexico City
region. The effect of air masses transport in rainwater ionic
composition has also been discussed in urban centers with heavy
vehicular fleet and industries (Garcia et al., 2006; Strayer et al.,
2007).
The relations between ionic species in rainwater samples are
usually applied to identify potential sources for atmospheric
pollution. Traditionally, the sodium ion has been used to identify
marine influence by the presence of sea-salts in rainwater or in the
particulate matter soluble phase. Keene et al. (1986) applied con-
centration ratios to estimate the sea-salt fractions in atmospheric
samples, and also emphasized that the sodium ion may not always
be appropriate. Thus, the evaluation criteria must be carefully
analyzed for the region under study. Recently, Dos Santos et al.
(2012) discussed the influence of the South Atlantic Ocean on sea
salt mass concentrations in aerosol from urban atmosphere of
Buenos Aires and highlighted the episodes of high Naþ concen-
trations when onshore winds speed were higher than 4.3 m s1. On
the other hand, the potassium ion was identified as indicator of
biomass burning processes and biogenic sources (such as leaf
fragments, pollen, log, etc.), in both wet and dry deposition (Forti
et al., 2000; Pauliquevis et al., 2007).
In Brazil, the agriculture activities involving sugar cane har-
vesting period, deforestation process and pasture maintenance
fires in the Amazon are important sources of air pollution (Allen
et al., 2004; Pauliquevis et al., 2007; Honório et al., 2010). High
potassium concentrations in rainwater and particulate matter were
associated with events in some regions with these characteristics.
Also it is important to highlight that the period of fires always co-
incides with drought or low rainfall seasons, mostly in June through
November (Cavicchioli et al., 2010).
In a site such as the Metropolitan Area of São Paulo (MASP), the
local anthropogenic sources of sodium and potassium are signifi-
cant and could mask these ions origins. In this study, the influence
of mesoscale phenomena, such as the entry of sea breeze (wind
speed and direction analysis) and transport of air masses (HYSPLIT
model) of the Naþ and/or Kþ in rainwater was evaluated. To identify
events with excess of sodium and potassium species in rainwater
samples, ratios between Kþ and/or Naþ ions and anthropogenic
ions (NO 2
3 and SO4 ) were calculated. For the MASP, the environ-
mental agency of São Paulo State (CETESB) attributes more than
80% of NOx to vehicular emissions and around 60% of SOx to fixed
sources. Moreover, the precursors of Naþ and Kþ in the atmosphere
are related with land-use activities, marine air masses and other
natural processes. The assessments for potassium and sodium
compounds sources are mixed and little is known about the
anthropogenic emissions of these species in urban areas. Therefore,
this study aims to improve the discussion of possible relations from
local or external sources in atmospheric samples.
2. Methodology
2.1. Site sampling
The State of São Paulo occupies 2.9% of the Brazilian territory and
is responsible for over 31% of Brazilian Gross Domestic Product
(GDP). The state has over 41 million inhabitants and the largest
vehicle fleet in the country with over 19 million units. Besides
hosting the largest number of different sectors industries, the state
of São Paulo centralizes over 50% (3.6 million hectares of cultivated
area) of the Brazilian production of sugar cane. Many regions (cen-
tral, north and northwest) in the state suffer serious problems of air
pollution during the harvest period due to the practice of burning
sugar cane manual harvesting (CETESB, 2011). The MASP is located
in the eastern region of the state of São Paulo (23.5  S and 46.7  W),
with an average altitude of 800 m and w50 km from the shore.
From 2002 to 2005, MASP had an inhabitant/vehicle ratio around
2.8, with a total of 6 million vehicles. According to CETESB, around
90% of air pollutants emissions were due the transport sector, which
burns a complex mixture of different fuel types: gasohol (78%
gasoline þ 22% anhydrous ethanol), hydrous ethanol (95% v/v) and
diesel. In the period of study the use of ethanol ranged from 34 to
44%, gasoline 50e60% and diesel decreases from 6.5 to 5.6% of the
total fuel burned. This change could be attributed to the beginning of
the flex fuel vehicles popularization process; this engine repre-
sented 3% of 2005’s vehicular fleet (CETESB, 2005). Samples were
collected at Mackenzie University campus, in the central area of the
MASP, which is a densely urbanized region, surrounded by large
avenues with intense traffic of buses and light vehicles.
2.2. Rainwater sampling
The wet-only samples were collected with an automatic rain-
water collector G. K. Walter Eigenbrodt model (Königsmoor, Ger-
many) inside the campus of Mackenzie University (23 320 S,
46 390 W), in the central area of MASP, from July 2002 to December
2005. The high density polyethylene flasks were maintained with
deionized water until the time of use, in addition to receiving pe-
riodical maintenance for cleaning. After the end of the rainfall
event, or in the early morning (for evening events), the sampling
flask was transferred to the laboratory where the sample volume
was determined. Afterward, fractions of w50 mL were separated
for different analytical determinations (pH, conductivity and major
ions). For ionic chromatography analysis, the sample fraction was
filtered with Millex membrane (pore 0.22 mm) and frozen (18  C)
until analysis.
2.3. Reagents and solutions
All reagents were of analytical grade (>99% purity, Merck) and
were used without further purification. Ultra-pure water used to
prepare the solution was obtained from a Barnstead Nanopure
system (resistivity >18 MU cm). The stock standard solutions
(100 mmol L1) were prepared by dissolution of their salts in
deionized water and stored at 4  C in the dark. The multi-element
standard solutions at 5.0, 10, 25 or 50 mmol L1 concentrations
which were used for ion-chromatography measurements were
prepared daily by dilution in deionized water of stock standard
solutions just before use (Leal et al., 2004).
 M.S. Vieira-Filho et al. / Atmospheric Environment 79 (2013) 299e307
2.4. Instrumentation
The pH direct measurements were carried out by a potentiom-
eter Digimed model DM-20. The glass electrode and potentiometer
were calibrated with pH 7.00 and 4.00 buffers solutions. Conduc-
tivity measurements were made with a conductivimeter Digimed
Model DM-31. The measurements of conductivity and pH were
done with samples at 25  0.2  C. The analyses of major ionic
species were carried out by a chromatographic system Metrohm
761 model with conductivity detection. The detection limits for
Mg2þ and Ca2þ ions were 2.4 and 1.3 mmol L1, respectively, while
for all other evaluated species (CH3COO, HCOO, Cl, NO 
2 , NO3 ,
þ þ þ
SO2 2
4 , C2O4 , Na , K and NH4 ), the detection limits were lower
than 1.0 mmol L1. The details for these determinations were dis-
cussed previously (Leal et al., 2004).
2.5. Meteorological data
The hourly wind data at ground level were obtained from the
Center for Weather Forecasting and Climate Studies (CPTEC/INPE),
and meteorological data station from Institute of Astronomy,
Geophysics and Atmospheric Sciences, University of São Paulo (IAG/
USP) were evaluated for the events of sodium and potassium excess.
Backward trajectories were generated with 72 h from HYSPLIT
model (Draxler and Hess, 1997; Draxler and Rolph, 2012; HYSPLIT e
http://ready.arl.noaa.gov/HYSPLIT.php) obtained through the
READY (Real-time Environmental Applications and Display System)
platform from NOAA (National Oceanic and Atmospheric Admin-
istration). Meteorological data (Reanalysis) - products of NOAA/
OAR/ESRL PSD, Boulder, Colorado, USA e from the global atmo-
spheric model for the years 2002e2005 were used as input data
(http://www.esrl.noaa.gov/psd/, accessed in November, 2012).
Backward trajectories of each excess event end at 23.55 S and
46.64 W, the last 24 h of integration represent the sample collec-
tion date. The vertical velocities from the Reanalysis were used for
all potassium events. The height levels trajectories used were 500,
1500 and 3000 m above ground level (AGL) to avoid complicated
local topography.
3. Results and discussion
3.1. Ionic composition
Regarding the acidity evaluation, the total of 315 rainwater
samples can be considered neutral, characterized by a 5.3 pH
Fig. 1. a) Ionic balance in rainwater samples collected from July 2002 to December 2005 (n ¼ 315) in downtown MASP. b) Ionic balance in rainwater samples with ratios of sodium
(n ¼ 31) and potassium (n ¼ 7) exceeding 1.5 in relation with sulfate and nitrate concentrations.
301
average value, and only 9 samples out of those had pH values 4.5.
The results of the ionic balance (Allan, 2004) indicate a good dis-
tribution around the ideal model (y ¼ x), with linear coefficient of
0.98 (R2 ¼ 0.95) (Fig. 1a). In order to avoid misleading conclusions,
the ionic balance for rainwater samples with ratios Kþ/NO þ
3 and K /
2 þ  þ 2
SO4 or Na /NO3 and Na /SO4 above 1.5 was calculated separately
to ensure that these samples are representative of the total data set
(Fig. 1b).
The potassium and sodium monthly average concentration
and monthly rainfall are presented in Fig. 2. The annual climato-
logical rainfall from 1939 to 2007 was 1400 mm, with correspon-
dent rainy period from December to March with rainfall
>150 mm month1 and drought period from May to August with
rainfall <50 mm month1. Winter (dry) and summer (rainy) tem-
peratures range from moderate (average 16  C in July) to high
(average 28  C in February). The total rainwater volume studied
(3960 mm) represented 70% of the rainfall for the period in the
MASP, and was also well distributed in all months. The potassium
maximum monthly concentration was usually observed in the dry
period (shaded regions) while sodium did not show a clear pattern
with some relative maximums in both seasons.
Fig. 3 shows potassium, sodium, chloride, sulfate and nitrate
concentrations variability, with extreme, mean and median values.
Nitrate and sulfate were the dominant species, exceeding all other
ions due to the expressive anthropogenic emission processes in a
metropolitan area like São Paulo. The potassium arithmetic mean
concentration (4.89  10.5 mmol L1) was 5.8 times lower than the
nitrate (28.5  27.5 mmol L1) and 3.4 times lower than the sulfate
(16.7  17.7 mmol L1). Regarding the sodium ion, the average
concentration (17.6  30.3 mmol L1) to the set of samples was
substantially equal to sulfate (Fig. 3). However, the median value for
sodium was 1.4 times lower than sulfate (10.9 mmol L1).
Within the total set of samples evaluated, samples with sodium
and potassium concentrations higher than nitrate and sulfate
concentrations were identified. In order to characterize these
events, the adopted criteria based on ratios [Kþ]/[NO þ
3 ] and [K ]/
2 þ  þ 2
[SO4 ] or [Na ]/[NO3 ] and [Na ]/[SO4 ] > 1.5 were used to evaluate
the possible long-range transport influence to the rainwater ionic
composition in MASP. From this restriction, only 7 samples fol-
lowed this condition for potassium, <3%, and 31 for sodium, w10%.
The volume weight mean (VWM) concentrations of some rain-
water ions from MASP are compared with other sites sampling on
Table 1. While the cations concentrations presented little differ-
ences between the dry and rainy seasons, sodium (32.4 mmol L1)
and potassium (10.9 mmol L1) excess events were significantly
302 M.S. Vieira-Filho et al. / Atmospheric Environment 79 (2013) 299e307
Fig. 2. Variation of monthly concentrations of potassium, sodium in rainwater samples
in MASP. Shaded regions show the dry season. The monthly mean climatological
pluviometry from 1933 to 2002 (bars), and pluviometry from July 2002 to December
2005 (dots).
higher than their values for dry and rainy seasons, as well as for
region of sugar cane plantation (Coelho et al., 2011) and Central
Amazonia region (Pauliquevis et al., 2007). The sodium excess
events concentrations exceeded all other sites, except Taiwan and
Hong Kong (Wai et al., 2007), which represent coastal areas. On the
other hand, the potassium excess events showed VWM concen-
trations higher than any other sites. Comparing the ions variations
it is noteworthy that the Central Amazonia (Pauliquevis et al., 2007)
and Northern India (Khare et al., 2004) showed the higher
Fig. 3. Box chart distribution of the potassium, sodium, chloride, nitrate and sulfate
ions concentrations in rainwater samples in MASP from July 2002 to December 2005
(n ¼ 315). The symbol (*) represents the volume weighted mean (VWM). Horizontal
box lines: 25th, 50th and 75th percentile values; error bars, 5th and 95th percentile
values; () minimum and maximum values (in the box) arithmetic mean.
differences among dry and rainy seasons. The nitrate and sulfate
concentrations in São Paulo from the dry period were at least two
times lower than those evaluated in Mexico City (Báez et al., 2007).
3.2. Potassium excess events
From the total samples, only 7 e in six different days e could be
characterized as potassium excess events. The data of burning areas
through satellite images (Fig. 4), and the air masses backward tra-
jectories calculated by the HYSPLIT model (Fig. 5) were analyzed to
identify possible long range transport contribution for potassium
concentrations. This assumption is based on the incorporation of
potassium compounds originated from areas with high fire spots
(sugar cane plantations or/and forests) to the air mass across the
72 h transport.
One factor that contributes to the biomass burning is the sugar
cane management, being that the sugar cane fields cover about 5.3
million hectares in São Paulo State. The central-north region of the
state accounts for a substantial part of this total. In the sugar cane
management cycle, the burning occurs before the manual harvest,
from May to November (Cuadra et al., 2012). It is important to
highlight that all the excess potassium events occurred in this
period, except for January 28, 2003. January 2003 presented the
highest precipitation rate in comparison with all the events. Besides
the high rainfall for this month, few fire spots were observed over
Brazilian region due to the rainy season.
Concerning the fires, it is important to quote that in the litera-
ture, Setzer (1993) was one of the pioneers to use the term “arc of
fire” for the Midwest and North of Brazil. These regions comprise of
the States of Acre, Rondonia, Mato Grosso, Pará, Tocantins and
Maranhão, where fires are related to the forests burning, besides
the intense agricultural and pastoral activities. The drier months,
from June to October offer favorable conditions for these fires to
spread.
The backward trajectories, calculated for 72 h (Fig. 5), indicate
transport of air masses from the countryside of Brazil toward São
Paulo in at least one altitude level. It is observed that the maximum
for concentration and fire spots (28  28 km) occurred on
September 20, 2002. For this event, the lowest number of fire spots
reached 1076 for the 500 m altitude backward trajectory. Also, the
State of São Paulo presented 209 fire spots, the maximum for the
period studied. There is a common pattern among the backward
trajectories for the events presented in Fig. 5 a, b, e and f, showing a
long range transport to the MASP in different levels. For these four
specific events, potassium ion concentrations were directly pro-
portional to the fire spot numbers, suggesting contributions by
long-range transport from the “arc of fire” region, coupled with
mid-range transport from sugar cane harvest burning in the
central-north of São Paulo State.
In the July 11, 2003 event (Fig. 5c), the backward trajectories
starting in southern South America showed air parcels moving with
high speed towards São Paulo, which was related with the frontal
system arrival in São Paulo State. The same pattern was observed in
October 11, 2003 event (Fig. 5d).
For some potassium excess events, air masses from regions with
high fire spots numbers move towards MASP, characterizing an
external source and is reflected in the collected rainwater chemical
composition. Nonetheless, not all the excess events presented this
characteristic. Therefore, it would be inaccurate to link potassium
in atmospheric samples only with external sources. The local
component is always present in megacities like São Paulo, due to
the various activities related to the manufacturing, construction
and other anthropogenic processes that can release potassium
compounds to the atmosphere. For instance, the Energy Secretariat
of the São Paulo State reports the use of 5.5  107 kg yr1 firewood
 M.S. Vieira-Filho et al. / Atmospheric Environment 79 (2013) 299e307 303

Table 1
Comparison of the volume weight mean (VWM) ions concentrations in MASP and other sites in Brazil and worldwide.

Locations Period N Naþ Kþ Cl NO

3 SO2
4 Ref.

VWM (mmol L1)

São Paulo (rainy)a 07/2002e12/2005 231 10.9 3.0 8.5 19.9 11.1 a
São Paulo (dry)b 84 8.8 3.8 9.2 20.3 12.2
São Paulo (Kþ excess) 7 15.9 10.9** 5.2 6.1 5.6
São Paulo (Naþ excess) 31 32.4* 6.4 20.9 13.0 8.8
Sugar cane non-harvestc 01/2003e07/2007 264 0.9 1 1.1 8.1 1.8 b
Sugar cane harvestc 2.5 5.7 4.5 17.5 5.3
Central Amazonia (rainy) 04/1998e12/2001 87 2.4 1.1 3.3 3.7 2.0 c
Central Amazonia (dry) 5.6 1.9 7.4 7.5 5.1
Northern India (rainy) 06/1999e09/2001 57 8.4 2.4 9.9 7.7 4.6 d
Northern India (dry) 11 17.3 4.2 15.9 13.4 7.1
Mexico City (rainy season) 2001e2002 71 7.0 2.2 9.6 42.6 30.9 e
Taiwan 01/1999e12/2000 79 75 6 90 22 29 f
Hong Kong 152 37 4 43 27 37

The p-values from t-test for both set of samples are indicated in bold. For sodium (*p-value < 105) and potassium (**p-value < 0.06). References: a) this study; b) Coelho et al.,
2011; c) Pauliquevis et al., 2007; d) Khare et al., 2004; e) Báez et al., 2007; f) Wai et al., 2007.
a
VWM for the months with rainfall >100 mm: OctobereMarch.
b
VWM for the months with rainfall <100 mm.
c
Central region of São Paulo State: AprileSeptember.

Fig. 4. Spatial distribution of the number of monthly fire spots (28 km  28 km) in Brazil based on data from satellite NASA GSFC, calculated by INPE (Instituto Nacional de Pesquisas
Espaciais). The statistics for maximum, minimum and mean fire spots per square area are presented as well. Accessed in August 07, 2012: http://www.inpe.br/queimadas.
304 M.S. Vieira-Filho et al. / Atmospheric Environment 79 (2013) 299e307
Fig. 5. Backward trajectories plots (72 h) from HYSPLIT model for the days when excess potassium concentrations were registered in the rainwater samples in MASP. The above
ground levels are 500 m (red), 1500 m (blue) and 3000 m (green).
for pizzerias and steakhouses in 2003 (BESP, 2004). This total
firewood can be expressed as 7.5 ha of eucalyptus forest per month
(considering the wood density of 390 kg m3). These sources are
usually not considered in atmosphere chemical balances and
contribute to potassium concentrations.
3.3. Sodium excess events
The land use changes and urbanization could cause alteration on
atmospheric circulations, changing surface fluxes and dispersion of
pollutants (Morris et al., 2001; Freitas et al., 2007). The MASP is
located at a distance of w50 km from the coastline, and there is a
mountain range (Serra do Mar) with ridges up to 1000 m, between
this region and the sea coast, which imposes a natural barrier to the
air masses from the Atlantic Ocean. The sea breeze circulation is
considered important mesoscale circulation to MASP, which is
characterized by an intensification of southeast winds at the sur-
face at 13e14 h (local time) in 50% of year days (Oliveira and Dias,
1982; Oliveira et al., 2003). In order to observe the phenomenon of
the incoming sea breeze, data from IAG/USP weather station,
located in the path of the incoming breeze toward the MASP were
analyzed.
Regarding to the surface winds, the sea breeze in São Paulo can
be characterized by the intensification or change to quadrant Easte
South, and these changes may be preceded by quadrant NortheEast
winds (Oliveira and Dias, 1982; Oliveira et al., 2003). From the total
of our rainwater samples, only 31, were identified as sodium excess
events by calculating the ratios ([Naþ]/[NO 3 ] and [Na ]/
þ
2
[SO4 ] > 1.5). The surface wind speed and direction (Fig. 6) were
analyzed for all these days. The maximum scale for velocity is
indicated in the upper right corner, and the vector absence corre-
sponds to calm winds condition (<1 m s1). The wind intensity
distribution indicates a predominance of calm conditions (47%),
while only 10% reaches values higher than 6 m s1. Based on these
observations it is possible to confirm the light wind speed char-
acteristic of MASP as previously discussed by Rodriguez et al.
(2010). For the 31 days with sodium excess, 45% of them exhibit
the sea breeze wind direction pattern for MASP. In these 14 days,
the wind maximum intensity occurred during the mid-afternoon,
where the shaded areas (Fig. 6) indicate the largest period of the
sea breeze occurrence (Pereira Filho et al., 2005; Vemado, 2012).
Regarding the classification by surface winds, the sea breeze was
absent in 55% of the days analyzed. Therefore, the influence of the
ocean is not mandatory for high relative sodium concentrations in
 M.S. Vieira-Filho et al. / Atmospheric Environment 79 (2013) 299e307 305

Fig. 6. Hourly wind variation for the events with higher sodium concentrations in rainwater samples in MASP. The events are separated by year: a) 2002. b) 2003. c) 2004 and d)
2005. Data were from Weather Station of IAG/USP. The shaded hours represent the sea-breeze occurrence period.

MASP. In the earth’s crust the sodium has a mixing ratio of
22,700 ppm (2.27%), ranking it as the seventh most abundant
element and the fifth most abundant metal. There are unlimited
supplies of sodium chloride in ocean waters and natural brines,
besides the incorporation in basic industries for inorganic matter
manufacturing (Greenwood and Earnshaw, 1984). Thus, these re-
sults highlight the importance of local sources for sodium ion in
MASP.
Several studies showed that the ions ratios in seawater indicated
the influence of sea spray in the atmospheric samples; this
connection was deeply evaluated by Keene et al. (1986). Analyses of
aerosols composition in remote coastal regions indicated Naþ and
Mg2þ concentrations similar to seawater samples. Therefore, these
species retain oceanic properties and could be considered tracers
(Talbot et al., 1986). It is important to note that these assumptions
have been systematically used for rainwater samples (Lara et al.,
2001; Mello, 2001; Teixeira et al., 2004; Khare et al., 2004;
Zhuang et al., 2008; Huneau et al., 2009). However, using this
relation to quantify the sea salt fraction on aerosol or rainwater
implies that one of these ions, usually the sodium, is not consumed
or produced by other processes. In other words, no local sources are
considered in this evaluation. The reported ratios between chloride
and sodium in seawater from the literature vary 20% around the
value found by Wilson (1975). This variability is due to difference in
local sampling sites, since particles from the soil dust are also added
to the rainwater, changing this ratio.
Another aspect of sodium chloride in atmospheric samples,
particularly aerosol and fogs (i.e., low liquid water content), it is the
possibility of Cl loss due to the incorporation of nitric acid (HNO3),
nitrogen dioxide (NO2), sulfuric acid (H2SO4) or sulfur dioxide (SO2)
in the droplets, which volatilizes hydrochloric acid (HCl) to atmo-
sphere due to the increase in acidity (Martens et al., 1973). The loss
of Cl is dependent on the acidity, and statistically significant
(w89%) under high acidity conditions, pH < 3 (Keene et al., 1998). It
is important to note that the chloride depletion phenomenon was
reported in dry deposition of coastal regions, and part of this HCl
can reincorporate in wet deposition regardless of the mechanism.
So the loss of Cl is lower or even not observed in rainwater

Table 2
Pearson correlation among sodium, chloride and potassium ions identified in the São Paulo rainwater samples, also the chloride and sodium ratios are presented. The values
were classified by dry and rainy seasons.

N [Cl]/[Naþ] [Naþ]  [Cl] [Kþ]  [Cl] [Naþ]  [Kþ]

Rainy (OcteMar) 231 0.78 0.88** 0.27** 0.38**

Dry (ApreSet) 84 1.05 0.74* 0.42* 0.39*

*p values lower than 0.0003; **p values lower than 0.00003.

306 M.S. Vieira-Filho et al. / Atmospheric Environment 79 (2013) 299e307
Fig. 7. Ratio Cl/Naþ concentrations in function of: (a) Naþ concentration and (b) pH values in rainwater samples in MASP from July 2002 to December 2005. The shaded ratio
represents the values reported in Keene et al. (1986) for seawater samples plus a deviation of 20%. The bold dots represent the samples with ratios that exceeded 1.5 in relation with
sulfate and nitrate concentrations.
samples (Andreae et al., 1990; Wai et al., 2007). Other potential
sources of sodium or magnesium, such as soil dust, represent
important aspect to be evaluated in coastal regions (Ayears et al.,
1999).
To evaluate the presence of the sodium chloride in rainwater
samples, the ratios between these ions were calculated and dis-
cussed as follows. The Pearson correlation for rainy and dry seasons
was evaluated for potassium, sodium and chloride (Table 2). The
corresponding p-values indicated a high significant level for so-
dium and chloride correlations, and the association was also
stronger in the rainy season (0.88). This value implies that these
species could have originated from similar sources, for instance, the
sea-spray. However, the ratio between these species was closer to
the seawater value (1.17) in the dry season, which mixes the hy-
pothesis about the similarity of the sources and suggests a Cl
depleted state during the rainy season. To evaluate the Cl loss
effect, Cl/Naþ ratios as a function of sodium concentration and pH
were evaluated (Fig. 7); the shaded region represents the seawater
ratios (20% standard deviations) reported in the literature. The
samples with sodium excess (n ¼ 31), predominantly presented
chloride deficiency. As no significant pH effect on Cl/Naþ ratios
(Fig. 7b) was observed, it is not reasonable to evaluate the sodium
excess due to the HCl volatilization (chloride loss). Considering the
above discussion, it is possible to consider other sodium com-
pounds sources, rather than only the sea-spray sodium chloride for
the MASP.
4. Conclusions
In this study the wet deposition was measured continuously at a
highly urbanized site with rainfall representativeness higher than
70%. As a result, 7 samples were classified as potassium excess
samples with a VWM concentration of 10 mmol L1, and 31 for the
sodium with 32 mmol L1, while for the total samples these ions
VWM concentrations were 4 and 15 mmol L1, respectively. From
backward trajectories analyses, we observed that half out of these
potassium excess events, presented concentrations directly pro-
portional to fire spot numbers, which means that some maximum
potassium concentrations in rainwater samples could be associated
with biomass burning activities. However, it was not possible to
distinguish the sugar cane management from the long-range
transport from forest biomass burning in Brazilian central-
northern regions. For the sodium events it was observed that the
ocean influence did not play an important role in the occurrence of
the excess concentrations, considering that less than half of these
events were preceded by the sea breeze. This high concentration
implies a considerable bias in the estimate of non-sea salt fraction
for atmospheric deposition, if it is considered that the sodium only
originated from oceanic processes. From 2002 to 2005, the sodium
and potassium excess events were circumstantial and indicate that
binding these ions concentrations with only external sources could
lead to underestimation and/or equivocal findings in MASP.
Acknowledgments
This research has been supported by FAPESP (03/01194). The
authors are grateful to the NOAA for the provision of the HYSPLIT
trajectories model used in this publication, and also thanks to the
Meteorological Station (IAG/USP), CPTEC/INPE and INMET for the
other atmospheric data. We also thank to CAPES and CNPq for the
student grants provided. We thank Dr. E. Abrantes for the English
revision.
References
Allan, M.A. (Ed.), 2004. Manual for the GAW Precipitation Chemistry Programme e
Guidelines, Data Quality Objectives and Standard Operating Procedures GAW-
160. WMO GAW-World Meteorological Organization Global Atmosphere Watch.
Allen, A.G., Cardoso, A.A., da Rocha, G.O., 2004. Influence of sugar cane burning on
aerosol soluble ion composition in Southeastern Brazil. Atmospheric Environ-
ment 38, 5025e5038.
Andreae, M.O., Talbot, R.W., Berresheim, H., Beecher, K.M., 1990. Precipitation
chemistry in central Amazonia. Journal Geophysical Research 95 (D10), 16987e
16999.
Avila, A., Alarcon, M., 1999. Relationship between precipitation chemistry and
meteorological situations at a rural site in NE Spain. Atmospheric Environment
33, 1663e1677.
Ayears, G.P., Gillett, R.W., Cainey, J.M., Dick, A.L., 1999. Chloride and bromide loss
from sea-salt particles in Southern Ocean air. Journal of Atmospheric Chemistry
33, 299e319.
Báez, A., Belmont, R., García, R., Padilla, H., Torres, M.C., 2007. Chemical composition
of rainwater collected at a southwest site of Mexico City, Mexico. Atmospheric
Research 86, 61e75.
BESP, 2004. Balanço Energético do Estado de São Paulo Ano Base 2003. Secretaria de
Energia, Recursos Hídricos e Saneamento, São Paulo (Estado), p. 205. (Série
Informações Energéticas, 002).
Beverland, I.J., Crowther, J.M., Srinivas, M.S.N., Heal, M.R., 1998. The influence of
meteorology and atmospheric transport patterns on the chemical composition
of rainfall in south-east England. Atmospheric Environment 32 (6), 1039e1048.
Cavicchioli, A., Morrone, E.P., Marques, R., Fornaro, A., 2010. Impactos atmosféricos
das transformações territoriais e do crescimento do agronegócio na Amazônia
matogrossense. Confins (Online), 10/2010, URL: http://confins.revues.org/6646.
 M.S. Vieira-Filho et al. / Atmospheric Environment 79 (2013) 299e307
CETESB, 2005. Relatório de qualidade do ar no Estado de São Paulo 2004. Série
Relatórios/Secretaria do Estado do Meio Ambiente, São Paulo. ISSN 0103e4103.
http://www.cetesb.sp.gov.br/ar/qualidade-do-ar/31-publicacoes-e-relatorios.
CETESB, 2011. Relatório de qualidade do ar no Estado de São Paulo 2010. Série
Relatórios/Secretaria do Estado do Meio Ambiente, São Paulo. ISSN 0103e4103.
http://www.cetesb.sp.gov.br/ar/qualidade-do-ar/31-publicacoes-e-relatorios.
Coelho, C.H., Allen, A.G., Fornaro, A., Orlando, E.A., Grigoletto, T.L.B.,
Campos, M.L.A.M., 2011. Wet deposition of major ions in a rural area impacted
by biomass burning emissions. Atmospheric Environment 45, 5260e5265.
Cuadra, S.V., Costa, M.H., Kucharic, C.J., da Rocha, H.R., Tatsch, J.D., Inman-
Bamber, G., da Rocha, R.P., Leite, C.C., Cabral, O.M.R., 2012. A biophysical model
of sugar cane growth. GCB Bioenergy 4, 36e48.
Dayan, U., Lamb, O., 2003. Meteorological indicators of summer precipitation
chemistry in central Pennsylvania. Atmospheric Environment 37, 1045e1055.
Dos Santos, M., Dawidowski, L., Smichowski, P., Ulke, A.G., Gómez, D., 2012. Factors
controlling sea salt abundances in the urban atmosphere of coastal South
American megacity. Atmospheric Environment 59, 483e491.
Draxler, R.R., Rolph, G.D., 2012. HYSPLIT (HYbrid Single-particle Lagrangian Inte-
grated Trajectory). Model Access via NOAA ARL READY Website. NOAA Air Re-
sources Laboratory, Silver Spring, MD. http://ready.arl.noaa.gov/HYSPLIT.php.
Draxler, R.R., Hess, G.D., 1997. Description of the Hysplit_4 Modeling System. NOAA
Technical Memorandum ERL ARL-224, December, p. 24.
Fleming, Z.L., Monks, P.S., Manning, A.J., 2012. Untangling the influence of air-mass
history in interpreting observed atmospheric composition. Atmospheric
Research 104-105, 1e39. http://dx.doi.org/10.1016/j.atmosres.2011.09.009.
Forti, M.C., Melfi, A.J., Astolfo, R., Fostier, A.H., 2000. Rainfall chemistry composition
in two ecosystems in the northeastern Brazilian Amazon (Amapa State). Journal
of Geophysical Research 105 (D23), 28,895e28,905.
Freitas, E.D., Rozoff, C.M., Cotton, W.R., Silva Dias, P.L., 2007. Interactions of an urban
heat island and sea-breeze circulations during winter over the metropolitan
area of São Paulo, Brazil. Boundary-Layer Meteorology 122, 43e46.
Garcia, R., Torres-Ma, C., del, Padilha, H., Belmont, R., Azpra, E., Arcega-Cabrera, F.,
Báez, A., 2006. Measurement of chemical elements in rain from Rancho Viejo, a
rural wooded area in the State of Mexico, Mexico. Atmospheric Environment
40, 6088e6100.
Gonçalves, C., Santos, M.A., Fornaro, A., Pedrotti, J.J., 2010. Hydrogen peroxide in the
rainwater of Sao Paulo megacity: measurements and controlling factors. Journal
of Brazilian Chemical Society 21, 331e339.
Greenwood, N.N., Earnshaw, A., 1984. Chemistry of Elements, 1 ed. Pergamon Press.
Hong, S.B., Kim, W.H., Ko, H.J., Lee, S.B., Lee, D.E., Kang, C.H., 2011. Characteristics of
formate and acetate concentrations in precipitation at Jeju Island, Korea. At-
mospheric Research 101, 427e437.
Honório, B.A.D., Horbe, A.M.C., Seyler, P., 2010. Chemical composition of rainwater
in western Amazonia d Brazil. Atmospheric Research 98, 416e425.
Huneau, F., Celle-Jeanton, H., Travi, Y., Loye-Pilot, M.D., Bertrand, G., 2009. Rain-
water chemistry at a Mediterranean inland station (Avignon, France): local
contribution versus long-range supply. Atmospheric Research 91, 118e126.
Keene, W.C., Sander, R., Pszenny, A.A.P., Vogt, R., Crutzen, P.J., Galloway, J.H., 1998.
Aerosol pH in the marine boundary layer: a review and model evaluation.
Journal of Aerosol Science 29 (3), 339e356.
Keene, W.C., Pszenny, A.A.P., Galloway, J.N., Hawley, M.E., 1986. Sea-salt corrections
and interpretation of constituent ratios in marine precipitation. Journal of
Geophysical Research 91 (D6), 6647e6657.
Khare, P., Goel, A., Patel, D., Bejari, J., 2004. Chemical characterization of rainwater at a
developing urban habitat of Northern India. Atmospheric Research 69, 135e145.
Lara, L.B.L.S., Artaxo, P., Martinelli, L.A., Victoria, R.L., Camargo, P.B., Krusche, A.,
Ayers, G.P., Ferraz, E.S.B., Ballester, M.V., 2001. Chemical composition of rain-
water and anthropogenic influences in the Piracicaba River Basin, Southeast
Brazil. Atmospheric Environment 35, 4937e4945.
307
Leal, T.F.M., Fontenele, A.P.G., Pedrotti, J.J., Fornaro, A., 2004. Composição iônica
majoritária de águas de chuva no centro da cidade de São Paulo. Quimica Nova
27 (6), 855e861.
Martens, C.S., Wesolowski, J.J., Harris, R.C., Kaifer, R., 1973. Chlorine loss from Puerto
Rican and San Francisco Bay area marine aerosols. Journal of Geophysical
Research 78, 36.
Mello, W.Z., 2001. Precipitation chemistry in the coast of the Metropolitan Region of
Rio de Janeiro, Brazil. Environmental Pollution 114, 235e242.
Morris, C.J.G., Simmonds, I., Plummer, N., 2001. Quantification of the influences of
wind and cloud on the nocturnal urban heat island of a large city. Journal
Applied Meteorology 40, 169e182.
Morselli, L., Bernardi, E., Vassura, I., Passarini, F., Tesini, E., 2008. Chemical
composition of wet and dry atmospheric depositions in an urban environment:
local, regional and long-range influences. Journal Atmospheric Chemistry 59,
151e170. http://dx.doi.org/10.1007/s10874-008-9099-9.
Oliveira, A.P., Dias, P.L.S., 1982. Aspectos observacionais da brisa marítima em São
Paulo. In: Anais do II Congresso Brasileiro de Meteorologia, Pelotas, Rio Grande
do Sul, Brasil, pp. 129e161.
Oliveira, A.P., Bonstein, R.D., Soares, K., 2003. Annual and diurnal wind patterns in
the city of São Paulo. Water Air Soil Pollution Focus 3, 3e15.
Pauliquevis, T., Lara, L.L., Antunes, M.L., Artaxo, P., 2007. Aerosol and precipitation
chemistry in a remote site in Central Amazonia: the role of biogenic contri-
bution. Atmospheric Chemistry and Physics Discussions 7, 11465e11509.
Pereira Filho, A.J., Massambani, O., Hallak, R., Karam, H., 2005.
A hidrometeorological forecast system for the Metropolitan Area of Sao Paulo.
In: World Weather Research Program Symposium on Nowcasting and Very
Short Range Forecasting e Toulouse, France, Meteofrance (CDROM).
Rodriguez, C.A.M., da Rocha, R.,P., Bombardi, R., 2010. On the development of
summer thunderstorms in the city of São Paulo: mean meteorological charac-
teristics and pollution effect. Atmospheric Research 96 (2e3), 477e488.
Samara, C., Tsitouridou, R., Balafoutis, C.H., 1992. Chemical composition of rain in
Thessaloniki, Greece, in relation to meteorological conditions. Atmospheric
Environment 26B (3), 359e367.
Seinfeld, J.H., Pandis, S.N., 1998. Atmospheric Chemistry and Physics of Air Pollution,
2o ed. John Wiley, New York.
Setzer, A.W., 1993. Operational satellite monitoring of fires in Brasil. International
Forest Fire News, FAO-UN 9, 8e11. http://queimadas.cptec.inpe.br/wrqueimadas/
documentos/1993_Setzer_Operational_Monitoring_FAO.pdf.
Strayer, H., Smith, R., Mizak, C., Poor, N., 2007. Influence of air mass origin on the
wet deposition of nitrogen to Tampa Bay, Florida e an eight-year study. At-
mospheric Environment 41, 4310e4322.
Talbot, R.W., Harris, R.C., Browell, E.V., Gregory, G.I., Sebacher, D.I., Beck, S.M., 1986.
Distribution and geochemistry of aerosols on tropical North Atlantic tropo-
sphere: relationship to Saharan dust. Journal of Geophysical Research 91 (D4),
5173e5182. http://dx.doi.org/10.1029/JD091iD04p05173.
Teixeira, E.C., Migliavacca, D., Pires, M., Fachel, J., 2004. Study of chemical elements
in atmospheric precipitation in South Brazil. Atmospheric Environment 38,
1641e1656.
Vemado, F., 2012. Sea Breeze Circulation Analysis and Its Impacts over Metro-
politan Area of São Paulo Precipitation Using ARPS Model (M.S. thesis).
Institute of Astronomy, Geophysics and Atmospheric Sciences, University of
São Paulo.
Wai, K.M., Wang, S.H., Tanner, P.A., Lin, N.H., 2007. A dual study of the rainwater
chemistry within the western Pacific region. Journal Atmospheric Chemistry 57,
85e103.
Wilson, T.R.S., 1975. Salinity and the Major Elements of Sea Water. In: Chemical
Oceanography, vol. 1. Academic, Orlando, pp. 364e413.
Zhuang, G., Huang, K., Xu, C., Wang, Y., Tang, A., 2008. The chemistry of the severe
acidic precipitation in Shanghai, China. Atmospheric Research 89, 149e160.