EVALUATION OF THE ATMOSPHERIC DEPOSITION ... - Springer Link

1 downloads 0 Views 594KB Size Report
REGION IN SOUTH BRAZIL. DANIELA MIGLIAVACCA1 ... atmospheric deposition samples in the Guaıba Hydrographic Basin (GHB), in south Brazil. Samples.
EVALUATION OF THE ATMOSPHERIC DEPOSITION IN AN URBAN REGION IN SOUTH BRAZIL ´ DANIELA MIGLIAVACCA1 , ELBA CALESSO TEIXEIRA1,∗ , FLAVIO WIEGAND1 , JOSETE DANI SANCHEZ2 , JANDYRA FACHEL3 and MARIANA RIBEIRO3 1

Fundac¸a˜ o Estadual de Protec¸a˜ o Ambiental, Rua Carlos Chagas 55/802, 90030-020, Porto Alegre, RS, Brazil; 2 Fundac¸a˜ o de Ciˆencia e Tecnologia, Rua Washington Luiz, 675, 90010-460, Porto Alegre, RS, Brazil; 3 Departamento de Estat´ıstica – UFRGS, Av. Bento Gonc¸alves 9500, 91509-900, Porto Alegre, RS, Brazil (∗ author for correspondence, e-mail: [email protected], Tel.: +55 5132251588; Fax: +55 5132124151)

(Received 17 March 2004; accepted 31 May 2005)

Abstract. The purpose of the present study is to analyse the chemical composition of bulk and wet atmospheric deposition samples in the Gua´ıba Hydrographic Basin (GHB), in south Brazil. Samples of bulk and wet deposition were analysed during a 1-year’s period (January to December 2002) at three different stations, i.e., 8◦ Distrito and CEASA stations in the city of Porto Alegre, and Charqueadas sta2− + − + + 2+ 2+ tion, in Charqueadas city. Conductivity, pH, Cl− , NO− were 3 , F , SO4 , Na , K , Mg , NH4 and Ca determined. The pH presented an average value between 4.75 and 7.45. Enrichment factor was characterised based on groups of acid (pH < 5.65) and alkaline (pH > 5.65) samples. For most of the studied ions, EF in bulk deposition was higher in alkaline samples, while in wet deposition there was little difference between acid and alkaline samples. The Multivariate Analysis technique, i.e. the Canonical Correlation Analysis (CCA), determined relationships between the two different data set (chemical and meteorological), identified the source (anthropogenic or natural) of the studied variables. Keywords: atmospheric deposition, canonical correlation analysis, major ions

1. Introduction Atmospheric deposition is an important process of removing pollutants from the atmosphere and it may occur either by wet or dry deposition. Sulphur compounds and acids have already been evidenced in the 17th century; however, in the 19th century there was an increase in rain acidity, mainly due to anthropogenic sources (Cowling, 1982). Acid rain has a pH lower than 5.6, which is the value expected on the basis of equilibrium of pure water and atmospheric CO2 (Seinfeld and Pandis, 1998; de Mello, 2001). Lower pH values are mainly due to the presence of sulphuric, nitric and carboxylic acids, whose major precursors are emissions of nitrogen and sulphur oxides, volatile hydrocarbons and carboxylic acids. Such processes involve complex chemical reactions in the atmosphere (Losno et al., 1991). In south Brazil, there are only a few studies related to the chemical characterisation of atmospheric deposition (Migliavacca et al., 2004; Flues et al., 2002). Some Water, Air, and Soil Pollution (2005) 167: 91–110

C 

Springer 2005

92

D. MIGLIAVACCA ET AL.

studies were conducted at the Metropolitan Area of Porto Alegre (MAPA) in order to determine the existence and the source of acid rain (Milano et al., 1989; Luca et al., 1991; Luca and V´asquez, 2000). These studies confirmed acid rain during some sampling campaigns, with pH values smaller than 5.6 and the contribution of deposition to the contamination of urban rain drainage of that area. Although there are only few studies related to atmospheric deposition and air quality in the MAPA, they are still incipient considering the complex chemical reactions that occur in the atmosphere. The Canonical Correlation Analysis has proved to be an effective technique used to find the correlations between two sets of variables; this technique and other multivariate techniques have been used to analyse environmental data (Statheropoulos et al., 1998). The purpose of this study is to characterise the chemical composition of the bulk and wet atmospheric deposition in the Gua´ıba Hydrographic Basin (GHB), by applying a Multivariate Analysis Technique, i.e. the Correlation Analysis (CCA), in order to associate the analysed chemical variables (cations and anions) with the meteorological data from the surface. 2. Area of Study The area under study is part of the Gua´ıba Hydrographic Basin (GHB) and it is located in the NE of the state of Rio Grande do Sul, between 28◦ S–31◦ S and 50◦ W– 54◦ W (Figure 1). The area is 84,763 km2 wide, which corresponds to 30% of the total area of the state. It presents various anthropogenic sources, and geographical and climatic conditions (Table I). The Metropolitan Area of Porto Alegre (MAPA) is presently formed by 30 counties, distributed over 9616 km2 , and concentrating approximately 36% of the population of the state of Rio Grande do Sul (3.7 million inhabitants, according to the 2000 census by IBGE, the Brazilian Institute of Geography and Statistics). For the present study, we have selected the cities of Porto Alegre and Charqueadas, both located within the boundaries of the Metropolitan Area of TABLE I Location of the sampling stations of atmospheric deposition in the Gua´ıba Hydrographic Basin Sampling stations

County

Deposition samplers

Geographic coordinates

Charqueadas

Charqueadas

Bulk and Wet

CEASA

Porto Alegre

Bulk and Wet

8◦ Distrito

Porto Alegre

Bulk and Wet

29◦ 59 S 51◦ 38 W 29◦ 59 S 51◦ 10 W 30◦ 03 S 51◦ 10 W

EVALUATION OF THE ATMOSPHERIC DEPOSITION IN SOUTH BRAZIL

Figure 1. Location of the sampling stations.

93

94

D. MIGLIAVACCA ET AL.

Porto Alegre, an area characterised by the existence of fixed and moving sources, amongst which an oil refinery (Refinaria Alberto Pasqualini – REFAP), a steel plant (Sider´urgica Riograndense and Gerdau S.A. A¸cos Finos Piratini), the III Petrochemical Industrial Complex, two coal-fired power plants (Termel´etrica de Charqueadas – TERMOCHAR and Usina Termel´etrica de S˜ao Jerˆonimo – UTSJ), and a considerable number of all sorts of vehicles. The most common emissions found in Porto Alegre city come from automobiles and there is an estimate that 20% of the total fleet of 3.1 million vehicles in the state are within this area, although there are several industrial complexes around the city which also contribute to emission of several pollutants (FEPAM, 2003). Charqueadas city is affected mainly by emissions from the TERMOCHAR coalfired power plant (72 MW of installed capacity), from A¸cos Finos Piratini steel plant (AFP), as well as by the UTSJ coal-fired power plant, however, vehicular emissions from local traffic should not be disregarded. There is also reason to assume an influence of atmospheric emissions from the III Petrochemical Industrial Complex. Due to its location, the GHB climate is influenced by cold air masses coming in from the South Pole. The seasons are well defined, with rains well distributed throughout the year. Winter is the rainiest season. The historical average (1960– 1990) for relative humidity is 75–85%, whilst the yearly average of accumulated rainfall is 1300–1400 mm year−1 (CPTEC/INPE, 2003). The winds prevail mainly from SE, and secondly from NE (EMBRAPA, 2003). According to K¨oppen’s international system, the GHB climate can be classified as humid subtropical (Cfa), with well distributed rains during the whole year (with more than 60 mm rain at any given month) and an average temperature above 22.0 ◦ C during the hottest month (EMBRAPA, 2003). The El Ni˜no and La Ni˜na phenomena influence the GHB as follows: during El Ni˜no there is a positive deviation from the historical rainfall and temperature averages, while during La Ni˜na, the deviation is negative.

3. Methodology 3.1. S AMPLING Three different locations were chosen as sampling stations: 8◦ Distrito and CEASA, in Porto Alegre, and Charqueadas, in the city of same name (Table I). The first two stations are located in urban area, with heavy traffic. Hospital S˜ao Lucas lies SE from the 8◦ Distrito station and is the main source of particulate matter, due to the burning of its hospital waste. The CEASA station is located near CEASA (Centrais de Abastecimento do Rio Grande do Sul, a big complex of vegetables and fruit supply deposits), which, in its turn, is located near the BR-116 and BR-290 highways, both with heavy traffic. In the NE quadrant of this station, it is found the Refinaria Alberto Pasqualini oil refinery (REFAP), as well as the Adubos Trevo

EVALUATION OF THE ATMOSPHERIC DEPOSITION IN SOUTH BRAZIL

95

port terminal, located in the Ca´ı River. Adubos Trevo is a fertiliser manufacturer and its terminal handles the unloading and storage of solid and liquid bulk cargo. Charqueadas station is located approximately 60 km from Porto Alegre and is served by the BR-290 and the RS-401 highways. This sampling station is located NW from the AFP steel plant and SE from TERMOCHAR coal-fired power plant. The UTSJ coal-fired power plant lies approximately 10 km SW, near the city of S˜ao Jerˆonimo, and it has a 20 MW installed capacity. The III Petrochemical Industrial Complex lies 26 km SW in a straight line between the cities of Montenegro and Triunfo. The sampling went from January to December 2002 and deposition samples were collected after each rain event (maximum 12 h after rainfall) with bulk and wet deposition samplers (Table I). The bulk sampler consists of a 21.5 cm Ø polyethylene funnel coupled to a 5-l collecting flask of same material, fixed 2 m above the ground and free from obstacles. The funnel is covered with a nylon net to avoid contamination (leaves, insects, etc.). The wet sampler, adapted from Fornaro et al. (1993), consists of a metallic protection box and a 5-l polyethylene collecting flask coupled to a lidded acrylic funnel. The lid opens only during rain events, closing as soon as the rain stops. It is operated either electrically or by a 12-Volt battery. The collecting and storage flasks were washed, first with an Extran Neutral solution (Merck), and afterwards several times with deionised water Type I (MilliQ-Millipore), having been stored in this water for at least 24 h. After that, water conductivity was measured and it did not exceed 2 µ S cm−1 (ASTM 5012, 1996). 3.2. CHEMICAL

ANALYSIS

One hundred seventy seven samples of atmospheric deposition were analysed: 107 of bulk, and 70 of wet deposition. After the samples had been collected, they were sent to the laboratory, where their volume was measured. pH and conductivity of non-filtered aliquots were determined with recently gauged Digimed equipment (DM20 and DM31, precision ±0.01, respectively). For determining the major ions, one aliquot of the atmospheric deposition samples was filtered through 0.22 µm pore membrane filter (Millipore), preserved with 3 µL of chloroform and stored at 4 ◦ C until future chemical analysis. The major ions were determined by Ion Chromatography (Dionex DX 500, with electric conductivity detector). Columns AS4A-SC and CS12A (Dionex) were used 2− − + 2+ 2+ for the analysis of anions (Cl− , NO− 3 , SO4 and F ) and cations (Na , Ca , Mg , −1 K+ and NH+ for the Ion 4 ), respectively. Detection limits were 0.01–0.05 µeq l Chromatography (Barrionuevo et al., 2004). In order to certify the quality of the analyses done by Ion Chromatograph for the present study, rainwater standard reference materials (CRM 408 and CRM 409) were also analysed (Community Bureau of Reference). The reference material

96

D. MIGLIAVACCA ET AL.

consists of artificial rainwater with relatively low electrolyte levels stored in sealed quartz flasks of approximately 100 ml. The reproducibility of the data obtained in six replicates was satisfactory and showed a coefficient of variation of 2%. The accuracy of anions and cations was largely satisfactory, being higher than 95% for all studied ions. The quality control of the analyses of atmospheric deposition was evaluated by ionic balance and by comparing the calculated and the measured conductivities. The analysis of linear regression between the sum ofanions and cations showed  a correlation of 0.79, for cations (101 ±73) and for anions (64 ± 47), indicating that organic anions (formiate and acetate), in spite of not having been analysed, are probably present in the deposition of that area, as seen in Figure 2a. As for measured and calculated conductivities, the analysis of linear regression showed a correlation of 0.89, indicating that the majority of the ions had been analysed (Figure 2b). 3.3. METEOROLOGICAL

DATA

The meteorological data used in this study were: rainfall (mm) – R; air temperature (◦ C) – T; atmospheric pressure (hPa) – P; relative humidity (%) – RH; wind speed (m s−1 ) – V, and the frequency of winds from NE, SE, SW, and NW. The wind direction was originally obtained considering 16 wind sectors at 22.5◦ intervals, starting clockwise from N (Andrade, 1994; Poissant et al., 1996; Hermann and Hanel, 1997). The meteorological data were supplied by the Porto Alegre Flight Protection Detachment of the Aeronautical Command (Ministry of Defence), having been collected by the meteorological surface station of the Salgado Filho International Airport from January to December 2002. 3.4. ENRICHMENT

FACTOR

The Enrichment Factor (EF) was calculated according to the method suggested by various authors, as described in Keene et al. (1986): X C sample

EF =  X 

C seawater

where X is the concentration of the ion of interest and C is the concentration of the reference ion. In this study, we used Na as reference element. 3.5. C ANONICAL

CORRELATION ANALYSIS

Canonical Correlation Analysis seeks to identify and quantify the associations between two sets of variables. This analysis focuses on the correlation between a linear combination of the variables in the one set and a linear combination of the

EVALUATION OF THE ATMOSPHERIC DEPOSITION IN SOUTH BRAZIL

97

Figure 2. Linear regression of the (a) cations and anions sum and (b) measured and calculated conductivity.

variables in another set. The pairs of linear combinations are called the canonical variables, and their correlations are called canonical correlations (Johnson and Wichern, 1998). The basic principle of the Canonical Correlation is to find linear combinations in each set of variables in such a way that the correlation between both sets is maximised. In the Canonical Correlation there is no distinction between the independent and the dependent variable, there are only two sets of variables for which one tries to find the maximum correlation between both.

98

D. MIGLIAVACCA ET AL.

TABLE II Average, minimum and maximum values, volume weighted average (VWA) for pH, conductivity (µS cm−1 ) and major ion concentrations (µeq L−1 ) for atmospheric deposition samples of the GHB (December 2001 to December 2002) Bulk deposition (n = 107)

Wet deposition (n = 70)

Variables

Average

Range

VWA

Average

Range

VWA

pH Conductivity F− Cl− NO− 3 SO2− 4 Na+ NH+ 4 K+ Mg2+ Ca2+

6.13 13.2 6.91 18.5 5.22 30.2 19.3 47.2 8.44 13.1 33.1

4.91–7.45 3.2–49.0 0.11–41 3.17–114 0.54–49.4 6.42–112 3.35–101 2.28–225 11.14–79.8 1.68–56.6 2.67–153

5.35 28.1 17.6 37.3 12.9 72.1 44.7 113 17.2 16.8 76.8

5.71 8.37 4.96 9.18 2.74 15.9 10.9 30.5 3.15 4.60 9.83

4.75–6.94 2.7–22.5 0.43–39.6 2.18–44.6 0.56–12.3 4.67–44.8 2.11–38.7 2.87–121 0.82–11.0 1.21–18.3 1.79–76.8

5.50 9.07 6.78 10.7 2.81 32.8 16.1 40.7 4.10 6.02 13.1

n = number of sample.

4. Results and Discussion 4.1. CHEMICAL

COMPOSITION OF THE ATMOSPHERIC DEPOSITION

Table II shows the results for average, maximum and minimum values and volume weighted average (VWA) for pH, conductivity, major ions and H+ (obtained from pH measurements) concentrations for the 177 samples of atmospheric deposition. 2− 2+ NH+ were the ions with the most significant concentrations 4 , SO4 and Ca for the bulk and the wet samplers, except for Ca2+ in the wet deposition. This might be explained by the smaller accumulation of particulate in suspension in the atmosphere, within the collecting system of the wet sampler, resulting in smaller concentrations of species from the soil, such as Ca2+ (Al-Momani et al., 1995). The presence of SO2− 4 in the results obtained can be related to the coal-fired power plants, and its highest concentrations have been obtained in Charqueadas, a station under the direct influence of the coal-fired power plants (TERMOCHAR and the UTSJ) and under a less significant influence of the A¸cos Finos Piratini steel plant. The presence of NH+ 4 in the atmosphere might be explained by the volatilisation of animal manure, domestic sewage, as well as from industrial processes, such as the use or the manufacture of fertilisers and from emissions of fossil fuel combustion (Berner and Berner, 1996; Ugucione et al., 2002). Figure 3 shows the pH histogram for samples of bulk and wet deposition. pH values for atmospheric deposition in the

EVALUATION OF THE ATMOSPHERIC DEPOSITION IN SOUTH BRAZIL

99

Figure 3. Histogram of pH values for bulk and wet deposition samples.

area ranged from 4.75–7.45, showing an average of 5.97 ± 0.54, with a coefficient of variation of 9%. pH values ranged from 4.75–5.6 in 28% of analysed events of atmospheric deposition (32 wet and 17 bulk), indicating acid deposition in the studied area. Approximately 76% of the studied events of bulk and wet deposition had a pH between 5.6 and 7.0, indicating alkaline deposition. These values might indicate that alkaline species, such as NH3 and carbonates, might neutralise the rainwater of the area under study.

100

D. MIGLIAVACCA ET AL.

Figure 4. Deposition rate of (a) bulk and (b) wet deposition collected at the Gua´ıba Hydrographic Basin.

The deposition rates of ionic species (µeq m−2 ) analysed in the atmospheric deposition of the Gua´ıba Hydrographic Basin were calculated by multiplying the ion concentration (µeq l−1 ) by the volume collected during each event and then dividing by the funnel area of each sampler. Figure 4 presents the calculated wet and bulk deposition rates of ion as percentages. NH+ 4 showed a maximum percentage of wet and bulk deposition, 32 and 27%, respectively. Ions Ca2+ and SO2− 4 also presented a dominant deposition rate (16%) and an average deposition of 2.95 and 2.84 µeq m−2 in bulk deposition and 1.14 and 1.83 µeq m−2 in wet deposition, respectively. 2+ + − Bulk deposition rates for Cl− , Na+ , NO− 3 , Mg , K , and F were 1.69; 2.10; −2 0.54; 1.20; 0.82; and 0.67 µeq m , respectively. For wet deposition, however, rates were lower: 0.97; 1.49; 0.33; 0.53; 0.36; and 0.60 µeq m−2 for Cl− , Na+ , 2+ + − NO− 3 , Mg , K , and F , respectively. From these data, one can observe that there was a small difference between bulk and wet deposition for F− , probably due to the proximity of the port terminal of the Adubos Trevo, a fertiliser manufacturer, and of the TERMOCHAR and the UTSJ coal-fired power plants, which contribute to F− emissions in the area of study.

EVALUATION OF THE ATMOSPHERIC DEPOSITION IN SOUTH BRAZIL

101

Migliavacca et al. (2004) suggested a HF contribution of approximately 13% probably due to the coal fired power plant, and reported that H+ and F− ions are usually present in preciptation. To determine the existence of possible common sources for the ions found in the atmospheric deposition of the area, the Pearson’s correlation matrix was applied. Tables III and IV show the correlation coefficients for bulk and wet deposition, respectively. The highest correlations were observed between Na+ × Cl− (0.92), TABLE III Pearson’s correlation matrix for samples of bulk deposition from the GHB H+ H+ F− Cl− NO− 3 SO2− 4 Na+ NH+ 4 K+ Mg2+ Ca2+ Rainfall Alkalinity

F−

Cl−

NO− 3

SO2− 4

Na+

NH+ 4

K+

Mg2+ Ca2+

Rainfall

0.30 −0.12 0.12 0.03 0.15 0.05 −0.11 0.57 0.24 0.20 −0.19 0.02 0.92 −0.02 0.12 −0.04 0.28 0.02 0.15 0.31 −0.03 −0.17 0.12 0.32 −0.02 0.13 0.27 0.55 −0.29 0.44 0.57 0.08 0.71 0.52 0.13 0.34 −0.28 0.51 0.29 0.09 0.78 0.25 0.14 0.15 0.87 0.19 −0.28 −0.33 −0.22 −0.29 −0.34 −0.27 −0.22 −0.37 −0.29 −0.33 0.04 0.35 −0.08 0.27 0.35 0.45 0.45 0.42 0.40 −0.16

Values in bold p > 0.01, n = 107. TABLE IV Pearson’s correlation matrix for samples of wet deposition from the GHB H+

Na+

NH+ 4

K+

Mg2+

Ca2+

F−

Cl−

NO− 3

SO2− 4

H+ Na+ −0.04 NH+ 0.13 −0.04 4 K+ 0.13 0.15 0.85 Mg2+ −0.14 0.67 0.06 0.18 −0.11 0.28 0.20 0.30 0.81 Ca2+ − F 0.48 0.07 0.62 0.52 0.17 0.30 Cl− 0.06 0.85 −0.11 0.05 0.59 0.16 0.09 NO− 0.16 −0.04 0.95 0.83 0.01 0.16 0.54 −0.12 3 SO2− 0.30 0.15 0.70 0.70 0.38 0.49 0.46 0.10 0.66 4 Rainfall 0.06 −0.37 −0.32 −0.34 −0.46 −0.38 −0.18 −0.23 −0.28 −0.40 Alkalinity −0.14 0.08 −0.12 −0.04 0.01 −0.05 0.01 0.11 −0.07 −0.14 Values in bold p > 0.01, n = 70.

102

D. MIGLIAVACCA ET AL.

2− 2+ Mg2+ × Ca2+ (0.87), Ca2+ × SO2− 4 (0.78), and Mg × SO4 (0.71) for bulk deposition. These correlations probably indicate that the source of these ions is the same or that the ionic species might have the same route during transport by air masses and might originate from different sources. Na and Cl in bulk and wet deposition suggest the presence of sea salts, results which were also reported by other authors (Flues et al., 2002; Alastuey et al., 1999; Akkoyunlu and Tayan¸c, 2003). + The correlation between SO2− 4 and NH4 in atmospheric deposition from the GHB under analysis might be explained by the process of neutralisation of sulphate + by the NH3 . Sulphate, NO− 3 , and NH4 presented a higher correlation in samples of wet deposition (r > 0.6), suggesting an insignificant influence of these ions on the process of dry deposition. These ions are related to anthropogenic activities in the area, such as SO2 and NOx emission from coal burning and the manufacture and use of fertilisers. The correlation between Ca2+ and Mg2+ with SO2− 4 in bulk deposition suggests that the particulate matter might originate from the soil and that these might sometimes neutralise the bulk deposition in relation to the wet deposition, as already mentioned before. The rainfall volume showed an inverse correlation with most of the chemical species analysed in this study (Table III) and this might be explained by the dilution process of the chemical species present in the atmosphere, which might return more efficiently by the scavenging process of the droplets (Akkoyunlu and Tayan¸c, 2003; Bravo et al., 2000; Beverland and Crowther, 1992).

4.2. ENRICHMENT

FACTOR

(EF)

The Enrichment Factor (EF) for the studied ions was calculated according to Equation (1), which allows the evaluation of marine contribution in samples of bulk and wet deposition collected between January and December 2002 in the GHB. Table V shows average of pH values, concentrations (µeq l−1 ) and EF for the studied ions. EF was divided into two separate groups: acid samples (pH < 5.65) and alkaline samples (pH > 5.65). Bulk deposition characterised the majority of the studied ions with higher concentrations for samples considered alkaline. Ca2+ shows a high concentration in alkaline deposition, approximately twice than in acid deposition. Ca2+ might originate from the soil, which was confirmed by the correlation of this element with Mg2+ in bulk deposition, while the correlation of Ca2+ with SO2− 4 in bulk deposition is believed to originate both from coal-fired power plants and other terrestrial sources (Akkoyunlu and Tayan¸c, 2003). As for wet deposition, the EF for the studied ions in samples characterised as acid and alkaline did not show any difference, except for SO2− 4 . This ion had

EVALUATION OF THE ATMOSPHERIC DEPOSITION IN SOUTH BRAZIL

103

TABLE V Average pH values, average ion concentration (µeq l−1 ) and average enrichment factors (EF) for samples of bulk and wet deposition from the GHB Acid deposition

Bulk deposition pH SO2− 4 Cl− K+ Ca2+ Mg2+ Wet deposition pH SO2− 4 Cl− K+ Ca2+ Mg2+

Alkaline deposition

Average

EF

Average

EF

5.45 24.6 16.3 4.44 16.5 7.77

– 22.1 0.98 25.4 41.2 2.83

6.31 31.6 18.7 9.32 37.1 14.3

– 22.9 0.82 30.2 73.4 4.34

24.6 0.95 23.3 30.8 2.32

6.10 14.4 9.52 3.25 10.6 5.31

17.5 0.68 22.6 32.6 2.62

5.32 17.2 9.12 3.10 9.29 3.68

an average EF of 24 (0.14–116) in samples of wet deposition with an average acid pH (5.32). The enrichment of this ion might be explained by the influence of fossil fuel combustion processes, especially of coal, and by industrial activities. The EF of SO2− 4 calculated herein was similar to that found by Akkoyunlu and Tayan¸c (2003) in Istanbul, Turkey. According to these authors, one of the sources of SO2− 4 is the massive contribution of industrial and residential areas, thus confirming our anthropogenic source. Studies conducted in South Brazil in areas near coalfired power plants showed significant sulphate concentrations in deposition samples (Flues et al., 2002; Migliavacca et al., 2004). EF for K+ was nearly constant, with a maximum value of 157 for the alkaline sample of bulk deposition. This suggests the soil as main source of K+ , whose EF values (which in acid and alkaline deposition are 24 and 26, respectively) are comparable to those found by other authors in Istanbul, Turkey (acid and alkaline deposition 26 and 17, respectively (Akkoyunlu and Tayan¸c, 2003). The low enrichment of Cl− and Mg2+ in samples of atmospheric deposition of the area under study might indicate an association of these ions with sea aerosols. Other authors have reported the sea as source of Mg and Cl ions (Akkoyunlu and Tayan¸c, 2003; Zunchel et al., 2003; Migliavacca et al., 2004). The EF for Cl was higher than 1 in 28% (51 samples) of the analysed samples of atmospheric deposition, while the highest value reported was 2.17.

104

D. MIGLIAVACCA ET AL.

4.3. CANONICAL

CORRELATION ANALYSIS

(CCA)

The Canonical Correlation Analysis (CCA) was applied to the ions in atmospheric deposition (bulk and wet) and to the surface meteorological data from January to December 2002. This statistical method investigates the possible interrelations between two sets of data. Tables VI and VII show the results of the canonical correlation of the two sets of variables, chemical and meteorological, for samples of bulk and wet deposition, respectively. The results of the CCA are presented as Canonical Functions (CF), which are the linear combination of the canonical sets of variables (Statheropoulos et al., 1998). The canonical correlation between both sets of data was higher than 0.7 (R = 0.75 for bulk deposition and R = 0.84 for wet deposition), revealing a high correlation between them. The total redundancy for samples of bulk deposition showed that 26% of the variance on of the chemical variables is explained by the meteorological canonical variables and 32% of the variance of the meteorological variables is expressed by the chemical canonical variables (Table VI). The total redundancy for samples of wet deposition showed that 32% of the variation of the chemical variables is explained by the meteorological canonical variables and that 40% of the meteorological variables is expressed by the chemical canonical variables (Table VII). TABLE VI Results of the canonical correlation analysis for bulk deposition samples from the GHB Canonical correlation (R) = 0.75 Variables

Chemical

1 2 3 4 5 6 7 8 9 10 11 12 Extracted variance Total redundancy

H+ Conductivity (Cond) Alkalinity (Alc) F− Cl− NO− 3 SO2− 4 Na+ NH+ 4 K+ Mg2+ Ca2+ 81% 26%

Square-chi (χ 2 ) = 253.56 Significance ( p) = 0.000 Meteorological Rainfall (R) NE SE SW NW Wind speed (V) Atmospheric pressure (P) Temperature (T) Relative humidity (RH)

100% 32%

EVALUATION OF THE ATMOSPHERIC DEPOSITION IN SOUTH BRAZIL

105

TABLE VII Results of the canonical correlation analysis for wet deposition samples from the GHB Canonical correlation (R) = 0.84 Variables

Chemical

1 2 3 4 5 6 7 8 9 10 11 12 Extracted variance Total redundancy

H+ Conductivity (Cond) Alkalinity (Alc) F− Cl− NO− 3 SO2− 4 Na+ NH+ 4 K+ Mg2+ Ca2+ 82% 32%

Square-chi (χ 2 ) = 221.46 Significance ( p) = 0.000 Meteorological Rainfall (R) NE SE SW NW Wind speed (V) Atmospheric pressure (P) Temperature (T) Relative humidity (RH)

100% 40%

The two first canonical functions (CF1 and CF2) were selected for the interpretation of both sets of data in bulk and wet deposition. In bulk deposition, the total variation was 41 and 44%, for the chemical and the meteorological variables, respectively. In wet deposition, the variation was 32 and 40% for the chemical and meteorological variables, respectively. The selection and interpretation of both canonical functions showed a significant Lambda Prime test, a satisfactory interrelation of chemical and meteorological data and a high canonical correlation. This methodology was suggested by Hair et al. (1998). Figures 5 and 6 suggest the distribution of canonical loads between CF1 and CF2 for both sets of data (chemical and meteorological) for bulk and wet deposition, respectively. For the interpretation of the figures, this study considers as valid correlations higher than 0.4 for CF1 and CF2. From the analysis of bulk deposition (Figure 5) it can seen that in the GHB CF1 is explained by the correlation between the chemical variables, from natural (Na+ , Cl− , Mg2+ , and Ca2+ ) and anthropogenic (NO− 3 ) sources, with canonical loads ranging from 0.73 to 0.42. CF1, related to the meteorological data, shows a positive correlation with wind speed and temperature, and a negative correlation with rainfall, relative humidity and atmospheric pressure. This correlation is meteorologically explained by the fact that regions with higher temperatures are associated with lower pressures and with an increase in wind speed. The negative

106

D. MIGLIAVACCA ET AL.

Figure 5. Scattering of canonical loads between CF1 and CF2, for samples of bulk deposition from the GHB, from January to December 2002.

Figure 6. Scattering of canonical loads between CF1 and CF2, for samples of wet deposition from the GHB, from January to December 2002.

EVALUATION OF THE ATMOSPHERIC DEPOSITION IN SOUTH BRAZIL

107

correlation between the chemical variables and rainfall and relative humidity might be related to periods of less rain. This suggests that the less rainfall, the higher the concentration of ions of natural source (with higher canonical loads) in bulk deposition of the GHB, since this type of sampler (bulk) is also responsible for the sampling of dry deposition in periods without rain. This dry deposition is largely associated with soil particles and sea salt aerosols, carried by sea breeze (coming from the Atlantic Ocean) to the area of study (Wiegand, 2000). CF2 presented a chemical canonical variable originating from anthropogenic − + sources (F− and SO2− 4 ) and natural sources (Cl and Na ) related to the meteorological canonical variable represented by atmospheric pressure and wind from NE and SE, in contrast to wind speed and wind direction from NW. The correlation between the winds from NE and SE and the canonical variable − − (F , SO2− 4 , Cl ) might be explained by anthropogenic contributions, such as from the III Petrochemical Industrial Complex, the A¸cos Finos Piratini steel plant, the Klabin (formerly Riocel) paper industry and from the Metropolitan Area of Porto Alegre (MAPA), which influenced the Charqueadas station more significantly. This station is located west from these contribution areas, where more significant F− and SO24 concentrations have been found than at the other stations. Despite its high value, the inverse correlation with the NW quadrant is not very clear. It might be explained by the diversity of anthropogenic activities in the area during the sampling events of deposition. The results obtained herein indicate that approximately 30% of the Cl− found at the three sampling stations does not have sea aerosols as their source. This data suggests that the source of this ion is probably the paper industry and the burning of wastes, especially from hospitals in the area. The presence of F− in atmospheric deposition is related to industrial activities, such as a fertiliser manufacturer, coal-fired power plants, where the emission of this chemical compound is typical (Zuncket et al., 2003; Assis et al., 2003). In relation to bulk deposition, wet deposition (Figure 6) did not show a similar correlation between the chemical and the meteorological data. CF1 grouped SO2− 4 , Cl− , Na+ , and Mg2+ , and these ions couldn’t be identified as originating exclusively from anthropogenic sources. Akkoyunlu and Tayan¸c (2003) suggest the dilution of Mg2+ in sea water and its later concentration in the soil. However, some authors (Migliavacca et al., 2004; Zunchel et al., 2003) reported that Mg2+ might originate from sea aerosols, as well as Na+ and Cl− . These ions showed an inverse correlation with rainfall, indicating, however, a positive correlation with wind from NE. This can be explained by the influence of mesoscale atmospheric circulation in the Metropolitan Area of Porto Alegre (MAPA), such as sea breeze and lagoon breeze (Lagoa dos Patos), described in detail by Wiegand (2000). When the winds prevail from SE, the Metropolitan Area of Porto Alegre receives a stronger influence from the lagoon, thus showing lower

108

D. MIGLIAVACCA ET AL.

concentrations of Cl− , Na+ , and Mg2+ . Conversely, when the wind prevails from NE, the MAPA is more influenced by the sea breeze and shows a high and positive correlation for these ions. As for SO2− 4 , described above, the highest concentrations were observed at Charqueadas, a station heavily influenced by anthropogenic activities located to E/NE (coal-fired power plant and steel plant). CF2, however, presented a significant correlation of Ca2+ and Mg2+ with temperature, NW and wind speed (canonical load above 0.85); and an inverse correlation with atmospheric pressure and relative humidity. It might be suggested that this correlation is related to intense meteorological phenomena, such as fronts and squall line, with moderate to strong winds from NW, an increase in temperature and a drop in atmospheric pressure. These events are associated with heavy rain, preceded by moderate to strong winds, all of which might be responsible for re-suspension of particles from the soil. Thus, it can be said that less intense and longer rain events are associated with weaker winds, higher atmospheric pressure and higher relative humidity, which in their turn are associated with lower Ca2+ and Mg2+ concentrations, since scavenging process of the atmosphere is more efficient during these events.

5. Conclusions The pH of wet and bulk deposition ranged from 4.75–7.45, with 28% of the samples showing values between 4.0 and 5.6, indicating acid precipitation in the GHB. 2− 2+ Higher concentrations of NH+ in the bulk and wet samples 4 , SO4 and Ca were found in those from the GHB. In bulk deposition, EF was higher in alkaline samples for most of the studied ions (Ca2+ , K+ , and Mg2+ ). In wet deposition, there was almost no difference in EF for the studied ions, except for SO4 . CCA applied to the studied data (bulk and wet deposition) in the Gua´ıba Hydrographic Basin showed a relationship amongst the chemical and meteorological variables, with high canonical correlation between the chemical and meteorological sets of data. The canonical correlation between both sets of data was higher than 0.7 (R = 0.75 for bulk deposition and R = 0.84 for wet deposition), revealing a high correlation between them. The distribution of canonical loads between CF1 and CF2 for bulk and wet deposition identified the source (anthropogenic or natural) of the chemical variables associated with the meteorological conditions. Sulphate with higher EF in wet deposition of acid samples, and Na+ and Ca2+ , with higher EF in bulk deposition of alkaline samples, were confirmed in the canonical correlation analysis by ions associated with the group of variables of anthropogenic and natural sources, respectively.

EVALUATION OF THE ATMOSPHERIC DEPOSITION IN SOUTH BRAZIL

109

Acknowledgements We are grateful to CNPq and FAPERGS for their financial support. We are grateful to the Sampling Department of FEPAM and to the Laboratory of Chemistry of FEPAM for their technical support, as well as to scholarship-holders IC Jos´e Manuel Marcolino and Camila de Souza Schuch, both for their technical assistance. References Akkoyunlu, B. O. and Tayan¸c, M.: 2003, ‘Analyses of wet and bulk deposition in four different regions of Istanbul, Turkey’, Atmospheric Environment 37, 3571–3579. Alastuey, A., Querol, X., Aura Chaves, C. R., Carratala, A. and Lopez-Soler, A.: 1999, ‘Bulk deposition in rural area located around a large coal-fired power station, northeast Spain’, Environment Pollution 106, 359–367. Allegrini, I. and De Santis, F.: 1989, ‘Measurement of atmospheric pollutants relevant to dry acid deposition’, Analytical Chemistry 21, 237–255. Al-Momani, I. F., Ataman, O. Y., Anwari, A. M., Tuncel, S., K¨ose, C. and Tuncel, G.: 1995, ‘Chemical composition of deposition near an industrial area at Izmir, Turkey’, Atmospheric Environment 29, 1131–1143. Andrade, M. F., Orsini, C. and Maenhaut, W.: 1994, ‘Relation between aerosol sources and meteorological parameters for inhalable atmospheric particles in S˜ao Paulo city, Brazil’, Atmospheric Environment 28(14), 2307–2315. Assis, C. M., Silveira, I. L. and Anizelli, R. C. M.: 2003, ‘Automonitoramento da qualidade do ar em Ouro Preto/MG’, in Anais do III Congresso Interamericano da Qualidade do Ar, Canoas, RS, Brasil. ASTM: 1996, Standard guide for preparation of material used for the collection and preservation of atmospheric wet deposition: D 5012. West Conshohocken, PA, vol. 11.03, pp. 387–391. Barrionuevo, S., Nascimento, R. S. H., Machado, A. C. M. and Pires, M.: 2004, ‘Caracteriza¸ca˜ o qu´ımica da precipita¸ca˜ o atmosf´erica por cromatografia iˆonica. Coor. Elba Calesso Teixeira, Mar¸cal Jos´e Rodrigues Pires’, in Estudos ambientais em Candiota: carv˜ao e seus impactos. FINEP/PADCT/CIAMB/FABERGs/FEPAM, Porto Alegre. 193–200. Berner, E. K. and Berner, R. A.: 1996, Global Environment: Water, Air, and Soil Geochemistry Cycles, Prentice-Hall, New Jersey, 376 pp. Beverland, I. J. and Crowther, J. M.: 1992, ‘On the interpretation of event and sub-event rainfall chemistry’, Atmospheric Environment 75, 163–174. Bravo, H. A., Saavedra, M. I. R., S´anchez, P. A., Torres, R. J. and Granada, L. M. M.: 2000, ‘Chemical composition of deposition in a Mexican Maya region’, Atmospheric Environment 34, 1197–1204. EMBRAPA: 2003, ‘Empresa Brasileira de Pesquisa Agropecu´aria’, http:// www. cnpt. embrapa.br/ agromet/cli pf8.html. Cowling, E. B.: 1982, ‘Acid precipitation in historical perspective’, Environmental Science and Technology 16(2), 110A–123A. CPTEC/INPE: 2003, ‘Centro de Previs˜ao do Tempo e Estudos Clim´aticos/Instituto Nacional de Pesquisas Espaciais’, http://www. cptec.inpe.br/products/climanalise. html. de Mello, W. Z.: 2001, ‘Deposition chemistry in the coast of the metropolitan region of Rio de Janeiro, Brazil’, Environmental Pollution 114, 235–242. FEPAM: 2003, ‘Funda¸ca˜ o Estadual de Prote¸ca˜ o Ambiental Henrique Luiz Roessler’, Descri¸ca˜ o do Estado do Rio Grande do Sul. Cap´ıtulo 1.2.2 Relevo, Superf´ıcie e Popula¸ca˜ o. CD-ROM. Porto Alegre.

110

D. MIGLIAVACCA ET AL.

Flues, M., Hamma, P., Lemes, M. J. L, Dantas, E. S. K. and Fornaro, A.: 2002, ‘Evaluation of the rainwater acidity of a rural region due to a coal-fired power plant in Brazil’, Atmospheric Environment 36, 2397–2404. Fornaro, A., Isolani, P. C. and Gutz, I. G. R.: 1993, ‘Low-cost automated rain collector and its application to the study of wet only deposition in S˜ao Paulo (Brazil)’, Atmospheric Environment 27B(3), 307–312. Johnson, R. A. and Wichern, D. W.: 1998, Applied Multivariate Statistical Analysis. Prentice-Hall, New Jersey, 650 pp. Herrmann, P. and Hanel, G.: 1997, ‘Wintertime optical proprieties of atmospheric particles and weather’, Atmospheric Environment 31(24), 4053–4062. Keene, W. C., Pszenny, A. A. P., Galloway, J. N. and Hawley, M. E.: 1986, ‘Sea-salt corrections and interpretation of constituent rations in marine deposition,’ Journal of Geophysical Research 91, 6647–6658. Luca, S. J., Milano, L. B. and Ide, C. N.: 1991, ‘Rain and urban storm water quality’, Water Science Technology 23, 133–140. Luca, S. J. de and V´asquez, S. G.: 2000, ‘Qualidade do ar e das chuvas na Regi˜ao Metropolitana de Porto Alegre’, in Zurita, M. L. L. and Tolfo, A. M. (eds.), Qualidade do Ar em Porto Alegre, Secretaria Municipal do Meio Ambiente, Porto Alegre, pp. 54–60. Losno, R., Bergametti, G., Carlier, P. and Mouvier, G.: 1991, ‘Major ions in marine rainwater with attention to sources of alkaline and acidic species’, Atmospheric Environment 25A, 763–770. Milano, L. B. M., Luca, S. J., Rosauro, N. M. L. and Castro, C. M. B. de.: 1989, ‘An´alise da qualidade da a´ gua de chuva na Regi˜ao Metropolitana de Porto Alegre e fatores meteorol´ogicos associados’, Revista Brasileira de Engenharia, Caderno de Recursos H´ıdricos 7(2), 39–67. Migliavacca, D., Teixeira, E. C., Pires, M. and Fachel, J.: 2004, ‘Study of chemical elements in atmospheric deposition in South Brazil’, Atmospheric Environment 38(11), 1641–1656. Poissant, L., Bottenheim, W. J. and Roussel, P.: 1996, ‘Multivariate analysis of a 1992 sontos data subset’, Atmospheric Environment 30(12), 2133–2144. Seinfeld, J. H. and Pandis, S. N.: 1998, Atmospheric chemistry and physics: From Air Pollution to Climate Change, J. Wiley & Sons, New York, 1309 pp. Statheropoulos, M., Vassiliadis, N. and Pappa, A.: 1998, ‘Principal component and canonical correlation analysis for examining air pollution and meteorological data’, Atmospheric Environment 32, 1087–1095. Tuncer, B., Bayar, B., Yesilyurt, C. and Tuncel, G.: 2001, ‘Ionic composition of deposition at the Central Anatolia (Turkey)’, Atmospheric Environment 35, 5989–6002. Ugucione, C., Felix, E. P., Rocha, G. O. and Cardoso, A. A.: 2002, ‘Daytime and nighttime removal processes of atmospheric NO2 e NH3 in Araraquara’s region – SP’, Eclet. Quim. 27, 1–9. Wiegand, F.: 2000, ‘Circula¸co˜ es Cl´assicas e N˜ao-Cl´assicas na Regi˜ao Metropolitana de Porto Alegre/RS’, Disserta¸ca˜ o de Mestrado, Centro Estadual de Pesquisa em Sensoriamento Remoto e Meteorologia, Universidade Federal do Rio Grande do Sul, Porto Alegre, Brasil. Zunckel, M., Saizar, C. and Zarauz, J.: 2003, ‘Rainwater composition in northeast Uruguay’, Atmospheric Environment 37, 1601–1611.