Atmospheric Research 80 (2006) 67 – 85 www.elsevier.com/locate/atmos

Rainwater chemical composition at two sites in Central Mexico A.P. Ba´ez *, R.D. Belmont, R.M. Garcı´a, M.C.B. Torres, H.G. Padilla Laboratorio de Quı´mica Atmosfe´rica, Centro de Ciencias de la Atmo´sfera, Universidad Nacional Auto´noma de Me´xico, Circuito Exterior, Ciudad Universitaria, Me´xico D.F. 04510, Mexico Received 28 February 2004; received in revised form 30 March 2005; accepted 24 June 2005

Abstract Chemical analyses were performed on rainwater samples collected at the National Autonomous University of Mexico (UNAM) in Mexico City and at a wooded site, Rancho Viejo (RV) in the State of Mexico, for the periods 1994–2000 and 1994–1999, respectively. At UNAM, rainwater was collected for the entire rainy season period each year, while at RV, technical considerations limited collection to weekends only. The results showed large variations in rainwater chemical composition in most years, mainly because of the variability of meteorological conditions and also because of changes in source emissions. Sulfates and NH4+ showed higher annual volume-weighted mean concentrations (VWMC) in both sites. At UNAM, the maximum annual VWMC for SO42 occurred in March and the minimum in July and August. Lower concentrations of almost all ions were found at RV; however, the H+ concentration was higher at this site. The pH in Mexico City, calculated from the annual VWMC of H+, was 4.95, which is a little higher than pH values reported in some other countries. Despite the fact that sulfate and NO3 concentrations were lower at RV, the pH was lower. Air-mass back trajectories were calculated for individual concentrations of SO42, H+, NH4+, Ca2+, and Mg2+, observed at each sampling site for weekend data. At RV, sulfate concentrations were higher when air-mass back trajectories indicated a wind flow from Mexico City and Toluca at 1000 MAGL (meters above ground level) and 3000 MAGL. The hydrogen ion exhibited the same behavior. Calcium and Mg2+ concentrations were also higher when the wind blew from urban areas at 1000 and 3000 MAGL. At UNAM, H+ concentration was lower and Ca2+ and Mg2+ were higher when wind blew from the northern sector of the city at 1000 and 3000 MAGL. In UNAM, the NO3/SO42 and NH4+/SO42 ratios were 0.5 and 1.09 in 1994 and 0.86 and 1.64 in 2000, respectively, indicating a decrease in SO2 emissions resulting from the change of fuel oil to gas fuel. The SO42/Ca2+ ratio was significantly lower at the UNAM site (1.82) compared to RV (5.36), and the SO42/H+ ratio was significantly higher at the UNAM site (6.77) compared to RV (2.01). The Spearman’s rho correlation between ionic concentrations indicated a positive correlation in most cases ( p b 0.05) for data from UNAM and RV. The multiple regression correlation analysis to predict * Corresponding author. Tel.: +52 55 5622 4071; fax: +52 55 5622 4050. E-mail address: [email protected] (A.P. Ba´ez). 0169-8095/$ - see front matter D 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.atmosres.2005.06.008

68

A.P. Ba´ez et al. / Atmospheric Research 80 (2006) 67–85

H+ concentration in Mexico City showed that NO3, NH4+, SO42, and Ca2+ contributed 23.2%, 20.9%, 8.0%, and 6.1%, respectively, to the H+ prediction, while Cl plus Na+ plus K+ only contributed 2.2%, and Mg2+ did not contribute. Sea-salt contribution to rainwater chemical composition was negligible with any wind direction at both levels. Excess sulfate (non-sea-salt sulfate) represented 98.7% of the total sulfate in rainwater collected during weekends at RV and 98.6% for weekend and annual rain samples at UNAM. D 2005 Elsevier B.V. All rights reserved. Keywords: Rainwater chemistry; Annual variations; Acidity in forests; Central Mexico

1. Introduction Although the chemical composition of wet precipitation has been studied in many countries for more than 30 years, such investigations remain important because anthropogenic emissions of some gases and aerosols in the atmosphere are continuously increasing. Air pollutant emissions are increasing rapidly in Asia, and the opposite is the case in Europe and North America (Higashino et al., 1997). Lee et al. (2000) reached the same conclusion, stating that the fast-growing economy in East Asia resulted in increased emissions of SO2 and NOx . According to Galloway (1995), future emissions of these pollutants in Asia will be equal to or even greater than the combined emissions of Europe and North America by 2020. Emissions from fossil fuel combustion have produced an environmental acidification of rain (Higashino et al., 1997; Lee et al., 2000; Terada et al., 2002). As a consequence, there has been damage to terrestrial, aquatic, and biological systems. Regarding ammonia, its concentration has increased substantially over the past years (Walker et al., 2000). The geographical extent to which precipitation chemistry is affected by industrial emissions has produced transboundary pollution with the associated ecological and political problems (Lee, 1993; Tuncer et al., 2001). Herut et al. (2000) performed chemical analyses of rain samples collected during five winter seasons and studied the variation of rain chemistry in relation to natural and anthropogenic sources and transport of the constituents. Civerolo and Rao (2001) reported a space–time analysis of precipitation-weighted SO42 concentration data across the USA. Several studies of temporal trends in precipitation chemistry in different regions have been published. Dillon et al. (1988) reported a 10-year trend for sulfate, nitrate, and hydrogen deposition in Central Ontario. Fay et al. (1989) published observed and modeled trends of sulfate and nitrate in precipitation in eastern North America. Puxbaum et al. (1998) concluded that sulfate and hydrogen concentrations have decreased with well-defined significant trends in central Austria in a 10-year trend study of the precipitation chemistry. Other studies of trends for nitrate and ammonium concentrations have been conducted (Erisman et al., 1989; Lynch et al., 1995; Avila, 1996). In the present study, the chemical composition of precipitation and temporal variability were determined in a highly polluted urban area and in a rural area from 1994 to 2000. The aim was to examine the long-term temporal variability of rainwater chemical composition in Mexico City and Rancho Viejo and its relationship to governmental policies of anthropogenic gas emissions. 2. Materials and methods 2.1. Sampling sites One sampling site was chosen in Mexico City (Fig. 1). The rain collector was located on the roof of the Atmospheric Sciences Center building at the campus of the National Autonomous

A.P. Ba´ez et al. / Atmospheric Research 80 (2006) 67–85

Fig. 1. Sampling site locations; M.C.M.Z. stands for Mexico City Metropolitan Zone. 69

70

A.P. Ba´ez et al. / Atmospheric Research 80 (2006) 67–85

University of Mexico (UNAM), at an elevation of 2250 m above sea level (masl). The mean annual rainfall is 860 mm. Rancho Viejo (RV) is located in a mountainous wooded region about 80 km southwest of Mexico City (Fig. 1). It is a small place with scattered houses mainly used on weekends and holidays. It is situated at 2700 masl surrounded by mountains covered with pine forests. This site was included for comparison of the results with those obtained at UNAM. It was also included because it lies beyond Mexico City limits with no nearby air-polluting sources but is also downwind of Mexico City, especially during the rainy season. Therefore, it is only responsive to regional emission sources and consequently is ideal for evaluating the impact that Mexico City has on rainwater chemical composition in forest ecosystems. 2.2. Sample collection Daily wet-only rainwater samples were collected at the UNAM site from 1994 to 2000 with an automatic wet/dry precipitation collector (Andersen Metal Works, Inc.) and transferred into standard high-density polyethylene buckets (29.5 cm diameter). The total rainfall amount collected was 4363 mm. In RV, unfortunately, it was only possible to collect rain samples on weekends because of technical limitations. The primary consideration was lack of personnel able to remain continuously at this isolated place during the entire rainy season. Rain samples were collected from 1994 to 1999 also with an automatic wet/dry precipitation collector (Andersen Metal Works, Inc.). The total rainfall amount collected was 1205 mm. After collection, rainwater from the bucket was poured out into a high-density polyethylene bottle, sealed with its respective cap, and refrigerated until chemical analyses were performed. Buckets were cleaned thoroughly with deionized water just after rain sample collection. 2.3. Statistical analysis The Kruskal–Wallis ANOVA Ranks Test was used to compare sulfate and calcium concentrations among the different wind sectors, obtained from air-mass back trajectory analysis for 1000 and 3000 m above ground level (MAGL). If this test detected a statistically significant difference, the Mann–Whitney Rank Sum Test was applied to identify the pair of wind sectors producing the significant difference observed in the Kruskal–Wallis Test. 2.4. Chemical analysis The samples were analyzed for pH, SO42, NO3, Cl, alkalinity (HCO3), Ca2+, Mg2+, Na+, K , and NH4+. Rain samples were filtered through Millipore 0.45-Am membrane filters leached with deionized water before chemical analyses. Rainwater pH was measured within 24 h after collection of samples by using an Orion 960 autochemistry system. Calibration was performed by means of buffer solutions at pH 4 and 7. Sulfates, NO3, and Cl were determined by nonsuppressed ion chromatography, and ammonium was determined by suppressed chromatography with a Perkin Elmer chromatograph, using an isocratic LC pump 250 and a conductivity detector. Sodium, K+, Ca2+, and Mg2+ were analyzed by flame atomic absorption spectrometry with a GBC 932AA instrument. Conductance was determined with a YSI 3200 conductivity instrument. Alkalinity was determined using the Gran’s titration method with an Orion 960 autochemistry system, and HCO3 concentrations were computed using the equation described for Stumm and Morgan (1981). +

A.P. Ba´ez et al. / Atmospheric Research 80 (2006) 67–85

71

The detection limits (DLs), in Aeq l 1, were 4.58, 2.74, 1.13, and 2.33 for SO42, NO3, Cl, and NH4+, respectively. DLs for Na+, K+, Ca2+, and Mg2+ were 0.074, 0.16, 0.5, and 0.13, respectively. 2.5. Data quality Precision of methods was determined by analysis of standards from 10 replicate measurements at different concentrations. The results showed that the precision for SO42, NH4+, NO3, Ca2+, and Mg2+ was b 5% RSD (relative standard deviations). The quality of analysis of each sample was checked for ion balance and specific conductance calculations, defined by Peden et al. (1986) as: Ion Percent Difference = [(Cations  Anions) / Cations + Anions)]  100 and Conductance Percent Difference = [(Calculated  Measured Conductance) / (Measured Conductance)]  100, respectively. 3. Results and discussion 3.1. Annual and monthly variations Annual and monthly volume-weighted mean concentrations (VWMC) and standard deviations of VWMC of the analyzed ions for UNAM are shown in Figs. 2 and 3, respectively. Fig. 4 shows the weekend VWMC of the main analyzed ions for UNAM and RV. The rainy season extends from about May to mid-October, depending on synoptic conditions. Rains are scarce or even absent from November to April in central Mexico. On some occasions, the rainy season starts in late June, as happened in 1998 during the strong 1997–1998 El Nin˜o event. Consequently, there are large and systematic variations in rainwater chemical composition in most of the years. These large variations during the course of a single year, solely the result of variability in meteorological conditions, render a trend analysis of rainwater chemical composition meaningless. The large variations produced by meteorological conditions would mask the change in rainwater chemical composition that could be produced throughout a long period. In general, monthly ionic concentrations at the UNAM site were higher at the beginning of the rainy season in May. H+ and Cl were exceptions, decreasing significantly as the rainy 2+ season advanced. Sulfates and NH4+ showed a higher annual VWMC, followed by NO3 , Ca ,  + 2+ + 2  + Cl , Na , Mg , and K . The maximum annual VWMC for SO4 , NO3 , and NH4 occurred in 1996. The maximum monthly VWMC of SO42 occurred in March, and the minimum concentration was observed in July and August. Maximum VWMC of NO3 occurred in April and May, and minimum concentrations occurred in June and July. Regarding weekend VWMC, the concentration of SO42 was slightly higher than that of NH4+. + 2+  + 2+ Sulfate, NO3 concentrations were lower in RV, as was , Ca , Cl , NH4 , Na , and Mg + expected. However, the H concentration was higher in RV than at the UNAM site. Potassium concentrations were also higher than at the UNAM site. These differences were significant at p b 0.05, according to the Mann–Whitney test (Table 1). The pH in the UNAM site, calculated from annual VWMC of H+, was 4.95 (ranging from 4.71 to 5.43); these are slightly acidic values but a little less so than those reported for southern China, the Korean peninsula, Japan, Central Eastern Europe, and Albany, New York (Table 2). It is important to note that despite the fact that the VWMC of sulfate and nitrate ions in rainwater was lower at RV than at the UNAM site, the VWMC of H+ was higher at RV (Fig. 4)

72

A.P. Ba´ez et al. / Atmospheric Research 80 (2006) 67–85

120 SO4

NO3

eq L-1

80

40

eq l-1

0 30 Cl

H

K

Mg

20

10

0 12

eq L-1

Na 8

4

0 150

eq L-1

Ca

NH4

100 50 0

1000

mm

Rain amount

500

0 1994

1995

1996

1997 Samping year

1998

1999

2000

Fig. 2. Annual volume-weighted mean concentrations (VWMC) and standard deviations of the VWMC (bars on lines) of analyzed ions in wet precipitation collected at the Autonomous University of Mexico (UNAM).

A.P. Ba´ez et al. / Atmospheric Research 80 (2006) 67–85

73

160 SO4

NO3

eq L-1

120 80 40 0 25 Cl

H

K

Mg

eq l-1

20 15 10 5 0 30 Na

eq L -1

20

10

0

180

eq L -1

Ca

NH4

120

60

0 Mar

Apr

May

Jun

Jul

Aug

Sep

Oct

Nov

Month

Fig. 3. Monthly volume-weighted mean concentrations (VWMC) and standard deviations of the VWMC (bars on lines) of analyzed ions in wet precipitation collected at the Autonomous University of Mexico (UNAM).

because the extensive pine forests and higher rain levels prevent alkaline soil particles from being blown by the wind. It is difficult to relate rainwater chemical composition to a complex phenomenon like the 1997–1998 El Nin˜o event. Even more, the way El Nin˜o affects the rain amount depends on the geographical situation of a specific region (Avila, 2002, cited by Gay et al., 2004). This outcome was also observed in the precipitation data from UNAM and RV; the rain amount increased 38% at the UNAM site from 1997 to 1998, whereas in RV, the rain amount in 1998 was only 7%

74

A.P. Ba´ez et al. / Atmospheric Research 80 (2006) 67–85 120

20

80

(Cl-)

-1

-1

(SO42-)

10 40 0

0 1994

1995

1996

1997

University

1998

1994

Rancho Viejo

60

1996

1997

University

1998

1999

Rancho Viejo

10

(NO3-)

40

1995

(Na+)

-1

-1

1999

5 20 0

0 1994

1995

1996

1997

University

1998

1999

1994

Rancho Viejo

10

1996

1997

University

1998

-1

1999

Rancho Viejo

60

(K+)

-1

1995

(Ca2+)

40

5

20 0

0

1994

1995

1996

1997

University

1998

1999

1994

Rancho Viejo

8

1996

University 120

(Mg2+) -1

6 -1

1995

1997

1998

1999

Rancho Viejo (NH4+)

80

4 40

2 0

0 1994

1995

1996

University

1997

1998

1994 1995 1996 1997 1998 1999

1999

Rancho Viejo

80

University 500

(H+)

(Rain amount) mm

-1

40 0

Rancho Viejo

250 0

1994

1995

1996

University

1997

1998

1999

Rancho Viejo

1994 1995 1996 1997 1998 1999 University

Rancho Viejo

Fig. 4. Annual volume-weighted mean concentrations (VWMC) of analyzed ions in wet precipitation collected at the Autonomous University of Mexico (UNAM) and Rancho Viejo (RV) on weekends.

higher than in 1997. The increase in precipitation from 1997 to 1998 was enhanced by the fact that the summer of 1997 was characterized by a negative precipitation anomaly (Magan˜a and Va´zquez, 2003). There does not seem to be a direct relationship between rain amount and rainwater chemical composition. On one hand, the increase in precipitation in 1998 was

A.P. Ba´ez et al. / Atmospheric Research 80 (2006) 67–85

75

Table 1 Comparison of ion concentrations in Aeq l 1 between Mexico City and Rancho Viejo using the Mann–Whitney test for weekend data from 1994 to 1999 T statistica

Median SO2 4 Cl NO 3 Na+ K+ Ca2+ Mg2+ NH+4 H+

Mexico City (N = 83)

Rancho Viejo (N = 82)

76.667 9.296 42.857 4.522 2.046 34.600 3.689 95.000 1.148

44.144 4.198 20.633 1.022 3.033 7.743 1.885 38.106 14.986

5227y 5002y 4701y 4304y 7703y 4122y 5095y 4012y 8356y

N is the number of samples. a The T statistic is the sum of the ranks in the smaller sample group. This value is compared to the population of all possible rankings to determine the possibility of T occurring. y There is a statistically significant difference at p b 0.05 (two-tailed test).

accompanied by a decrease in ionic species concentration in rainwater at the UNAM site. The increase in the amount of precipitation in 1998 produced a lesser content of soil particles containing alkaline metals in rainwater, therefore, the hydrogen ion concentration increased. On the other hand, despite below-normal rains in the summer of 1997, the concentration of ionic species in rainwater was not noticeably higher than in any other year throughout the whole studied period because of a concentration effect. In contrast, the outcome was different at RV, where despite higher levels of rain in 1998, the concentration of alkaline metals did not decrease that year, as it did at the UNAM site. This difference may possibly be a result of the perennial forests that prevent soil particles from being easily blown off by the wind, regardless of whether it is a rainy year or not. At this site, rainwater chemistry appeared to depend on wind direction rather than on El Nin˜o, based on the analysis of air-mass back trajectories at 1000 and 3000 MAGL (meters above ground level), which are discussed in the next section. 3.2. Air-mass back trajectories The VWMC in relation to wind directions obtained by air-mass back trajectory analysis are presented in Tables 3 and 4. The individual concentrations of sulfate (as a chemical tracer of anthropogenic pollution) and calcium (as a chemical tracer of crustal particles) were associated with the corresponding air-mass back trajectories calculated by the NOAA HYSPLIT model (Hybrid Single-Particle Lagrangian Integrated Trajectory Model) (Draxler and Rolph, 2003). Table 2 Average pH in Mexico City and other locations Site

pH

References

Albany, New York Central Eastern Europe Southern China Japan Korean peninsula Mexico City

4.2 4.4 to 4.5 4.7 4.8 4.7 4.95 (4.71 to 5.43)

Khwaja and Husain (1990) Hjellbrekke et al. (1995) Wang and Wang (1996) Hara (1998) Lee et al. (2000) This study

76

A.P. Ba´ez et al. / Atmospheric Research 80 (2006) 67–85

Table 3 Volume-weighted mean concentrations in Aeq l 1 for three wind direction sectors (sector 1, 0–908; sector 2, 91–1358; and sector 3, 136–3608) at the Rancho Viejo for weekend data from 1997 to 1999 Ion

N SO2 4 Cl NO 3 Na+ K+ Ca2+ Mg2+ NH+4 H+

Air-mass trajectory sector at 1000 MAGLa

Air-mass trajectory sector at 3000 MAGL

Sector 1

Sector 2

Sector 3

Sector 1

Sector 2

Sector 3

30 54.09 6.98 25.48 1.57 4.97 10.01 2.76 47.74 23.17

3 68.88 7.21 11.94 1.00 2.40 3.69 0.45 26.81 36.64

17 15.85 3.60 11.95 0.88 3.71 4.05 1.56 20.65 4.16

32 51.80 7.13 23.87 1.18 4.24 8.79 2.40 41.28 23.60

3 19.60 2.49 9.21 1.37 1.94 4.91 1.08 27.22 5.46

15 20.48 2.86 11.41 1.28 5.72 5.47 1.99 29.31 3.14

N is the number of samples. a Meters above ground level.

Hydrogen ion, NH4+, and Mg2+ were also associated with back trajectories as a complementary analysis. Air-mass back trajectories were calculated for 1000 and 3000 MAGL, and NOAA trajectories were calculated from 1997. The level 1000 MAGL was considered because storm cloud bases frequently lie around 1000 MAGL. The level 3000 MAGL is about 1200 m higher than the highest mountain summit that lies between Mexico City and Toluca and RV; at the same time, it is close to the height of the 500 millibars (mb) isobaric surface (one of the mandatory levels in meteorological analysis). The calculated air-mass back trajectories were grouped into three wind direction sectors: sector 1 from 18 to 908; sector 2 from 918 to 1358; and sector 3 from 1368 to 3608. The criteria to select these sectors in the study of rainwater chemical composition at RV were based on the geographical direction of large urban and industrial areas, in relation to this site, and on the direction of prevailing winds during the rainy season. Therefore, the large urban and industrial areas are located in sector 1, and the prevailing winds also blow from this sector, which corresponds to the direction of trade winds. Sector 2 corresponds to the wind direction frequently observed at the rear sector of easterly waves. Also, although there are no important urban areas in sector 2, it is plausible that the complex local wind patterns and wind Table 4 Volume-weighted mean concentrations of (Aeq l 1) for four wind direction sectors (sector 1, 1–908; sector 2, 91–1808; sector 3, 181–2708; and sector 4, 271–3608) at the University of Mexico for 1997–2000 data Ion

N SO2 4 Cl NO 3 Na+ K+ Ca2+ Mg2+ NH+4 H+ a

Air-mass trajectory sector at 1000 MAGLa

Air-mass trajectory sector at 3000 MAGL

Sector 1

Sector 2

Sector 3

Sector 4

Sector 1

Sector 2

Sector 3

Sector 4

95 62.3 11.0 40.8 7.4 3-0 44.4 5.1 93.9 7.3

44 76.2 10.4 55.4 5.3 2.4 33.6 3.8 106 17.5

10 55.5 10.5 27.8 2.5 2.9 15.2 1.7 74.5 9.1

10 67.5 6.9 56.5 5.4 2.4 25.5 2.7 88.8 19.2

74 56.4 10.3 34.6 7.5 2.9 42.7 4.7 90.4 3.9

59 73.9 10.8 55.5 5.2 2.3 34.1 3.8 104 16.8

12 60.6 7.4 44.9 4.6 3.2 30.4 3.5 96.1 7.9

14 75.6 12.0 45.3 4.8 3.7 27.5 3.9 79.1 21.1

Meters above ground level.

A.P. Ba´ez et al. / Atmospheric Research 80 (2006) 67–85

77

shear in mountainous areas produce a horizontal interchange of air between sectors 1 and 2. Sector 3 is the largest sector because there are only wooded areas up to several hundreds of kilometers around RV. Therefore, if this sector had been subdivided into smaller sectors, it would have resulted in more air-mass back trajectories with similar air-mass origins from the point-ofview of rainwater chemical composition. Regarding the UNAM site, because it is surrounded by intense anthropogenic emission sources, the air-mass back trajectory sectors were subdivided into four equally sized sectors: 18 to 908, 918 to 1808, 1818 to 2708, and 2718 to 3608. Fig. 5 shows some examples of air-mass back trajectories observed during the rainy season.

(a)

(b)

(c)

(d)

(e)

(f)

Fig. 5. Some air-mass back trajectories observed during the rainy season for different years. Backward trajectories ending at: (a) 00 UTC 04 Aug 97, (b) 04 UTC 12 Oct 97, (c) 22 UTC 20 Jun 98, (d) 23 UTC 29 May 99, (e) 21 UTC 06 Jul 99, and (f) 06 UTC 19 Sep 99.

78

A.P. Ba´ez et al. / Atmospheric Research 80 (2006) 67–85

At RV, volume-weighted mean sulfate concentrations (Table 3) were higher when air-mass back trajectories at 1000 MAGL came from 18 to 908 (54.1 Aeq l 1) and from 918 to 1358 (68.9 Aeq l 1) in relation to trajectories between 1368 and 3608 (15.8 Aeq l 1). The higher VWMC of sulfate observed when wind direction came from these sectors is an evidence of the interchange of air mentioned above. This implication seems emphasized by the fact that hydrogen concentrations were also higher when wind was blowing from sectors 1 and 2 (23.2 and 36.6 Aeq l 1), respectively, and lower when air was blowing from sector 3 (4.16 Aeq l 1). The results for sulfate and hydrogen confirm that sulfuric acid was transported from these sectors and rained out over the mountainous area where RV is located. A similar result was obtained for ammonium, although the difference between its concentrations for sector 2 and 3 was much less (47.7, 26.8, and 20.6 Aeq l 1 for sectors 1, 2, and 3, respectively). At 3000 MAGL, the sulfate concentration was higher only for sector 1 (51.8 Aeq l 1) versus 19.6 and 20.5 Aeq l 1 for sectors 2 and 3, respectively. The hydrogen concentration was also higher for sector 1 (23.6, 5.46, and 3.14 Aeq l 1 for sectors 1, 2, and 3, respectively). The previous results observed at RV also agreed with the geographic position of RV in relation to Mexico City and Toluca. These results strongly suggest that air-mass back trajectories from 08 to 1358 at 1000 MAGL were associated with lower pH and higher sulfate concentrations, whereas these parameters were associated only with air-mass back trajectories at 3000 MAGL from 08 to 908. This finding agreed or matched with the orographic characteristics of the studied region. The horizontal dispersion of air pollutants like sulfuric acid, transported from urban and industrial areas to downwind wooded areas where RV is, covered broader areas at low levels than at high levels simply because air currents are forced to flow around and over mountains, especially at low levels, and because light aerosols such as sulfuric acid do not settle down. Calcium concentrations were higher only for sector 1 at both levels: 10 Aeq l 1, versus 3.7 and 4 Aeq l 1 for sectors 2 and 3, respectively, at 1000 MAGL; and 8.8 Aeq l 1 versus 4.9 and 5.5 Aeq l 1 for sectors 2 and 3, respectively, at 3000 MAGL. Magnesium behaved similarly to calcium; its concentrations were 2.8, 0.45, and 1.6 Aeq l 1 for sectors 1, 2, and 3, respectively, at 1000 MAGL; and 2.4, 1.1, and 2 Aeq l 1 for sectors 1, 2, and 3, respectively, at 3000 MAGL. These results also agreed with the actual distribution of the most important sources of calcium and magnesium, which lie precisely northeast of RV. The higher Ca2+ and Mg2+ concentrations associated only with air-mass back trajectories from 08 to 908 at both levels are consistent with the fact that heavier dust particles settle down during the winding airflow around mountains at low levels. Obviously, these particles also settle down at higher levels; therefore, the maximum concentrations of Ca2+ and Mg2+ were also associated with air-mass back trajectories from 08 to 908. At the UNAM site, the concentrations of sulfate among the four sectors (Table 4) presented a lesser variation than at RV at 1000 and 3000 MAGL; they were 62.3, 76.2, 55.5, and 67.5 Aeq l 1 for sectors 1, 2, 3, and 4, respectively, at 1000 MAGL; and 56.4, 73.9, 60.6, and 75.6 Aeq l 1 for the sectors, respectively, at 3000 MAGL. This finding is consistent with the fact that the UNAM site is surrounded by numerous pollution sources. Ammonium ion concentrations had a pattern similar to sulfate (93.9, 106, 74.5, and 88.8 Aeq l 1 at 1000 MAGL, and 90.4, 104, 96.1 and 79.1 Aeq l 1 at 3000 MAGL, for sectors 1, 2, 3, and 4, respectively). The concentrations of hydrogen ion were lower for sector 1 at both levels: 7.3, 17.5, 9.1, and 19.2 Aeq l 1 for sectors 1, 2, 3, and 4, respectively, at 1000 MAGL; and 3.9, 16.8, 7.9, and 21.1 Aeq l 1 at 3000 MAGL. The lower concentrations of hydrogen ion when air-mass back trajectory indicated a flow from sector 1 could have resulted from neutralization reactions mainly between sulfuric acid and dust particles. The main sources of these components are situated in

A.P. Ba´ez et al. / Atmospheric Research 80 (2006) 67–85

79

Table 5 Ion ratios (calculated from volume-weighted mean concentrations in Aeq l 1) for four wind direction sectors (sector 1, 1– 908; sector 2, 91–1808; sector 3, 181–2708; and sector 4, 271–3608) at the National Autonomous University of Mexico for the period 1997–2000 Ion ratio 2 NO 3 /SO4 NH4+/SO2 4 2+ SO2 4 /Ca + SO2 4 /H 2+ 2+

Ca /Mg Cl/Na+

Sea salt ratioa

Air-mass trajectory sector at 1000 MAGLb

Air-mass trajectory sector at 3000 MAGLb

Sector 1

Sector 2

Sector 3

Sector 4

Sector 1

Sector 2

Sector 3

Sector 4

– – 2.44 – 0.19 1.16

0.7 1.5 1.4 8.5 8.7 1.5

0.7 1.4 2.3 4.4 8.8 2.0

0.5 1.3 3.6 1 9.1 4.2

0.8 1.3 2.6 3.5 9.3 1.3

0.6 1.6 1.3 14.6 9.1 1.4

0.8 1.4 2.2 4.4 8.9 2.1

0.7 1.6 2.0 7.7 8.6 1.6

0.6 1.0 2.8 3.6 7.1 2.5

N is the number of samples. a Calculated using Aeq. b Meters above ground level.

sector 1, as shown by the higher concentration of calcium and magnesium and by the lower concentrations of hydrogen when air-mass back trajectories at 1000 and 3000 MAGL showed that the wind blew from this sector (Table 4). 3.3. Sea salt contribution, ion ratios, and correlations Excess sulfate (non-sea-salt sulfate) represented 98.7% of the total sulfate in rainwater collected at RV and 98.6% for weekend and annual rain samples at the UNAM site. The quantification of sea-salt contribution was based on chloride. The SO42/Ca2+ ratio in rainwater was closer to the ratio in sea salt (2.44) at the UNAM site compared to RV for all the sectors (Tables 5 and 6). However, this finding does not suggest that Mexico City is much closer to the seashore, but simply that calcium concentration was much higher at the UNAM site than at RV. The Ca2+/Mg2+ ratio was much higher than in sea salt (0.195) at both sampling sites for all the sectors because of the continental character of these sites. The Ca2+/Mg2+ ratio in the crust is higher than unity, where the magnesium content is lower than that of calcium; however, it is lower than unity in sea salt, where the magnesium content is higher than that of calcium. This ratio was higher at the UNAM site than at RV because calcium concentration was much higher at the UNAM site than at RV, as was the case with the SO42/Ca2+ ratio. The Cl/Na+ ratio was higher than in sea salt (1.16) at both sampling sites because there are other sources of chloride in

Table 6 Ion ratios (calculated from volume-weighted mean concentrations in Aeq l 1) for three wind direction sectors (sector 1, 0–908, sector 2, 91–1358, and sector 3, 136–3608) for weekend data from 1997 to 1999 in the Rancho Viejo Ion ratio NO3/SO2 4 NH4+/SO2 4 2+ SO2 4 /Ca + SO2 4 /H 2+ 2+ Ca /Mg Cl/Na+ a

Sea salt ratioa

Air-mass trajectory at 1000 MAGLa

Air-mass trajectory at 3000 MAGL

Sector 1

Sector 2

Sector 3

Sector 1

Sector 2

Sector 3

– – 2.44 – 0.19 1.16

0.47 0.88 5.30 2.33 3.70 4.44

0.17 0.39 18.68 1.88 8.24 7.20

0.75 1.30 3.91 3.81 2.59 4.08

0.46 0.80 5.90 2.19 3.66 5.20

0.47 1.39 3.99 3.59 4.55 3.01

0.56 1.43 3.74 6.52 2.74 2.23

Meters above ground level.

80

A.P. Ba´ez et al. / Atmospheric Research 80 (2006) 67–85

continental regions. The main reason this ratio was noticeably higher at RV than at the UNAM site is that the concentration of sodium was higher at the UNAM site, where the content of dust particles was also higher. All the sectors showed clear continental characteristics for SO42/Ca2+, Ca2+/Mg2+, and Cl/Na+ ratios corresponding to the obvious continental characteristics of the sampling sites, where the main components in rainwater come from sources other than sea salt. For instance, Mexico City is situated at more than 300 km from the closest shoreline; even more, high mountains surround it. Also, the Sierra Madre Oriental with mountains above 4000 masl lies close to the Gulf of Mexico coast. The heavy rains that occur in these mountains (several thousands millimeters per year) are another factor that scavenges sea-salt aerosols even further. Therefore, the concentration of sea-salt aerosols is depleted rapidly, as tropical maritime air masses invade central Mexico during the rainy season. These environmental conditions make the quantification of sea-salt contribution highly uncertain. The Spearman’s rho correlation was applied to identify relationships between annual ionic concentrations at UNAM and in both sites on weekend data (Tables 7 and 8). It must be noted that a significant statistical association between variables should be interpreted with care. For example, sulfate and nitrate concentrations commonly show a significant correlation; however, this correlation is related to common emissions sources and transport processes rather than to chemical reactions between precursor species (Lee, 1993). The Spearman’s rho correlation analysis between ionic concentrations indicates a positive correlation in most cases at the significance level of 0.05. At the UNAM site, considering annual VWMC of SO42, NO3, and NH4+, the NO3/SO42 and NH4+/SO42 ratios were 0.5 and 1.09 in 1994 and 0.86 and 1.64 in 2000, respectively (Fig. 6). According to Fig. 2, the increase of NO3/SO42 and NH4+/SO42 ratios resulted mainly from a decrease in SO42 concentration (34% from 1994 to 2000). The SO42 and NO3 correlation can also be a consequence of their common source origin. For weekend data, the NO3/SO42 increased from 0.58 to 0.85 and from 0.41 to 0.71 at the UNAM site and at RV, respectively. The correlation between NH4+ and SO42 indicates that ammonium sulfate is a common constituent in the atmosphere in central Mexico, consistent with the fact that ammonium and sulfate were present in high concentrations in rainwater. The NH4+/SO42 ratio increased from 0.98 to 1.68 and from 0.87 to 1.38 at the UNAM site and at RV, respectively. A decrease for sulfates has also been observed in other countries (Fay et al., 1989; Puxbaum et al., 1998). In Mexico City, the decrease of SO42 can be explained by the SO2 reduction in industrial emissions and by the Table 7 Spearman’s rho correlation between ion concentrations in wet precipitation collected at Mexico City from 1994 to 2000 (N = 303a) Cl SO4 Cl NO 3 Na+ K+ Ca2+ Mg2+ NH+4 H+ a y

NO 3 y

0.524

Na+ y

0.663 0.449y

K+ y

0.463 0.621y 0.482y

Ca2+ y

0.546 0.569y 0.510y 0.874y

Number of samples. Correlation is significant at p V 0.05 (two-tailed).

Mg2+ y

0.628 0.521y 0.558y 0.840y 0.787y

0.623 0.554y 0.565y 0.861y 0.843y 0.955y

HCO 3

H+

NH+4 y

y

0.657 0.512y 0.711y 0.621y 0.619y 0.696y 0.660y

y

0.158 0.117y 0.136y 0.484y 0.367y 0.414y 0.360y 0.269y

0.0754 0.177y 0.0242 0.551y 0.442y 0.505y 0.450y 0.385y 0.885y

A.P. Ba´ez et al. / Atmospheric Research 80 (2006) 67–85

81

Table 8 Spearman’s rho correlation among ion concentrations in wet precipitation collected in Mexico City and Rancho Viejo on weekends from 1994 to 1999 Cl

NO 3

Mexico City (N = 83)§ SO2 0.557* 0.737* 4 Cl 0.451* NO 3 Na+ K+ Ca2+ Mg2+ NH+4 H+ HCO 3

Na+

K+

Ca2+

Mg2+

NH+4

H+

HCO 3

Rain amount

0.518* 0.724* 0.522*

0.586* 0.658* 0.580* 0.893*

0.735* 0.589* 0.624* 0.821* 0.794*

0.712* 0.618* 0.626* 0.824* 0.828* 0.952*

0.737* 0.562* 0.737* 0.701* 0.717* 0.796* 0.757*

0.224* 0.0987 0.175 0.419* 0.288* 0.267* 0.225* 0.225*

0.076 0.177 0.010 0.508* 0.415* 0.394* 0.372* 0.377* 0.909*

0.486* 0.368* 0.466* 0.328* 0.440* 0.401* 0.422* 0.492* 0.057 0.089

0.621** 0.564** 0.785** 0.638** 0.399**

0.537** 0.490** 0.632** 0.675** 0.650** 0.859**

0.673** 0.729** 0.496** 0.504 0.736** 0.471** 0.588** 0.0495 0.441** 0.116 0.745** 0.225** 0.695** 0.107 0.173

0.289** 0.148 0.179 0.0079 0.172 0.0182 0.0114 0.170 0.625**

0.078 0.088 0.143 0.170 0.192 0.135 0.098 0.075 0.199 0.198

Rancho Viejo (N = 82) a SO2 0.673** 0.765** 0.427** 0.264** 4 Cl 0.551** 0.555** 0.211 NO 0.477** 0.261** 3 0.451** Na+ K+ Ca2+ Mg2+ NH+4 H+ HCO 3

a Number of samples. * Correlation is significant at p V 0.05 (two-tailed). ** Correlation is significant at p b 0.05 (two-tailed).

reduction of sulfur content in fuel oil. Besides other SO2 control strategies applied in 1986, 1989, 1991, and 1993, industrial diesel oil with a sulfur content of 1% replaced fuel with a sulfur content of 2% in 1997. The result of the measurements of ambient SO2 by the government monitoring network, operating in the metropolitan area of Mexico City, indicated that the concentration of SO2 in the last 6 years has been below the air quality standard (0.13 ppm per hour and 0.03 ppm annually). The increase of NH4+/SO42 ratio from 1994 to 2000 (Fig. 6) is a consequence of the decrease in rainwater sulfate concentration. In contrast, ammonium ion did not present significant variations throughout the 7-year sampling period. In Mexico City Valley, there are numerous sources of ammonium that have not yet been reliably inventoried. However, some ammonium sources are agricultural, including manure handling (around Mexico City there are still several agricultural areas); cleansing-product manufacturing and domestic use; and hair-coloring products. The SO42/Ca2+ ratio was significantly lower at the UNAM site (1.8 for overall data and 1.82 for weekend data) than at RV (5.36 for weekend data), primarily because calcium concentration at RV was much less as a result of a much lower concentration of crustal particles. The SO42/H+ ratio was significantly lower at the UNAM site (6.34 for overall data and 6.77 for weekend data) than at RV (2.01 for weekend data) because the sulfate concentration was lower at RV, but most of all because hydrogen concentration is higher at RV. This concentration is not higher as a result of greater air pollution but as a result of fewer acid-neutralizing soil

82

A.P. Ba´ez et al. / Atmospheric Research 80 (2006) 67–85 1.0

NO3/SO4 ratio

(a)

0.5

R = 0.848, p = 0.016 0.0 2

NH4/SO4 ratio

(b)

1

R = 0.850, p = 0.015 0 12

(c)

Ca/Mg ratio

9

6

3 R = - 0.0823, p = 0.861 0 1994

1995

1996

1997 Years

1998

1999

2000

Fig. 6. Trends of ionic concentration ratios for the period 1994–2000 at the Autonomous University of Mexico (UNAM): 2+ 2 + 2 a) NO3/SO2 4 , b) NH4 /SO4 , and c) Ca /Mg . The line represents the trend or regression line.

particles at this site. Hydrogen ion correlated positively with SO42 and NO3, indicating the presence of nitric and sulfuric acids as H+ donor species. Although there was a good correlation between H+ and SO42 at the UNAM site, it is meaningful that this correlation was much higher at RV (Table 3). The correlation between H+ and NO3 was not significant at the UNAM site but significant at RV. These results indicate a greater relationship between H+ and SO42 and NO3 ions at RV because neutralization reactions between soil particles and acids in rainwater are less important at this location than at the UNAM site. The high rainwater acidity observed at RV demonstrates the large impact of Mexico City emissions on air and rainwater quality in downwind wooded areas, with the possible acidification of the forest soil. Deposition of the

A.P. Ba´ez et al. / Atmospheric Research 80 (2006) 67–85

83

Table 9 Multiple linear regression analysis using H+ concentration as the dependent variable in Mexico City (N = 303a) Model Constant NO 3 Ca2+ NH+4 a b c y

Coefficient 8.133 0.536 0.195 0.140

Standard error 2.121 0.034 0.023 0.026

t-value y

3.835 15.842y 8.504y 5.360y

VIFb

F-ratio ANOVAc

R

0.000 2.025 1.767 2.237

90.810y

0.690

Number of samples. Variance inflation factor. Analysis of variance for the variables included in the model. Significant at p b 0.05.

different ions was not calculated for RV because rainwater samplings were made only on weekends. In contrast, hydrogen ion correlated negatively with Ca2+, Mg2+, Na+, K+, and NH4+, as a result of neutralizing reactions between hydrogen ion and alkaline particles and ammonia. Chloride correlated negatively to H+ at a significance level of 0.05. The correlation between the alkaline ions Ca2+, Mg2+, Na+, and K+, and the poorly defined trend of the Ca2+/Mg2+ ratio from 1994 to 2000 for overall data at the UNAM site and weekend data at the UNAM and RV sites, suggest a common crustal origin of these metals. The Cl/Na+ ratio was significantly lower at the UNAM site (1.71 for overall data and 1.8 for weekend data) than at RV (4.84 for weekend data) because the sodium concentration was much lower as a result of a lesser content of crustal particles at RV. A multiple regression correlation analysis (MRC) was performed to predict H+ concentration, and the hierarchical MRC analysis used by Cohen and Cohen (1975) was applied to select the variables included in the final model (Table 9). The MRC analysis showed that NO3, NH4+, SO42, and Ca2+ contributed 23.2%, 20.9%, 8.0%, and 6.1%, respectively, to the H+ prediction, while Cl plus Na+ plus K+ only contributed 2.2%, and Mg2+ did not contribute. The fact that nitrate contributed 23.2% and sulfate contributed only 8.0% is possibly the result of the sulfur-dioxide control emissions and the slight increase of nitrate concentration in rainwater. 4. Conclusions Because of the large variations in rainwater chemical composition observed in most of the years at both sampling sites, it was difficult to evaluate changes in rainwater chemical composition during a long period based solely on changes in source emission characteristics. Also, the variability of meteorological conditions during a single year produces large variations in rainwater composition capable of masking changes resulting from pollution sources characteristics. The VWMC of sulfate decreased during the study period as a consequence of the reduction of SO2 in industrial emissions and the replacement of fuel oil by natural gas in thermoelectric power plants. The ionic ratios of nitrate and ammonium to sulfate concentration in rainwater at the UNAM increased from 1994 to 2000 because of a decrease in sulfate concentration. The concentrations of all ions, except H+, in rainwater collected during weekends were higher at the UNAM site than in the wooded area (RV) (downwind of Mexico City), as was expected, mainly because of the highly polluted atmosphere of Mexico City. The higher hydrogen ion concentration in RV was a consequence of the lower concentration of wind-blown dust at this

84

A.P. Ba´ez et al. / Atmospheric Research 80 (2006) 67–85

place; this decreased dust concentration resulted from the higher humidity and vegetation cover at this location and in the surrounding mountains, and not from a higher concentration of sulfate in rainwater. The presence of acid rain at RV indicates a possible harmful impact of Mexico City and other upwind anthropogenic sources on forested downwind regions. The analysis of the chemical composition of rainwater in relation to the analysis of air-mass back trajectories seemed to perform reasonably well for the two levels. However, it is not easy to know the level at which this analysis performed best; thus, a multilevel air-mass back trajectory analysis is mandatory in mountainous regions. More-extensive studies are necessary to identify the best level for this kind of analysis for the region studied in this work. Also, despite the fact that the air-mass back trajectory analysis seemed to perform reasonably well, it must be stressed that air-mass back trajectory analyses cannot be completely reliable and must be interpreted with caution in regions with very complex terrain and wind patterns, such as those in this study. A costly network of numerous meteorological stations and atmospheric soundings must be carefully situated for study of the actual transport mechanism of air pollutants. Acknowledgments The authors gratefully acknowledge the NOAA Air Resources Laboratory (ARL) for providing the HYSPLIT transport and dispersion model and/or READY website (http:// www.arl.noaa.gov/ready.html) used in this publication. The authors are indebted to Delibes Flores Roman, Rosa Marı´a Ayala, and Carlos Contreras for their support in computation and also grateful to Jorge Escalante, Wilfrido Gutie´rrez, and Alfredo Rodrı´guez for the maintenance of the sampling equipment and to Calixto Cuevas for his assistance in the laboratory. References Avila, A., 1996. Time trends in the precipitation chemistry at a mountain site in northeastern Spain for the period 1983– 1994. Atmos. Environ. 30, 1363 – 1373. Civerolo, K., Rao, S.T., 2001. Space–time analysis of precipitation-weighted sulfate concentrations over the eastern US. Atmos. Environ. 35, 5657 – 5661. Cohen, J., Cohen, P., 1975. Applied Multiple Regression Correlation Analysis for the Behavioral Sciences. Lawrence Erlbaum Associates, Hillsdale, NJ. Dillon, P.J., Lusis, M., Reid, R., Yap, D., 1988. Ten-year trends in sulfate, nitrate and hydrogen deposition in Central Ontario. Atmos. Environ. 22, 901 – 905. Draxler, R.R., Rolph, G.D., 2003. HYSPLIT (HYbrid Single-Particle Lagrangian Integrated Trajectory) Model access via NOAA ARL READY Website (http://www.arl.noaa.gov/ready/hysplit4.html). NOAA Air Resources Laboratory, Silver Spring, MD. Erisman, J.W., Frank, A.A., De Leeuw, M., Van Aalst, R.M., 1989. Deposition of the most acidifying components in Netherlands during the period 1980–1986. Atmos. Environ. 23, 1051 – 1066. Fay, J.A., Golomb, D., Zemba, S.G., 1989. Observed and modeled trend of sulfate and nitrate in Eastern North America. Atmos. Environ. 23, 1863 – 1866. Gay, C., Herna´ndez, M., Jime´nez, J., Lezama, J., Magan˜a, V., Morales, T., Orozco, S., 2004. Evaluation of climatic forecasts of rainfall for the Tlaxcala State (Me´xico): 1998–2002. Atmosfera 17, 127 – 150. Galloway, J.N., 1995. Acid deposition: perspectives in time and space. Water Air Soil Pollut. 85, 15 – 24. Hara, H., 1998. Acid deposition chemistry in Asia, Europe, and North America. Prog. Nucl. Energy 32, 331 – 338. Herut, B., Starinsky, A., Katz, A., Rosenfeld, D., 2000. Relationship between the acidity and chemical composition of rainwater and climatological conditions along a transition zone between large deserts and Mediterranean climate, Israel. Atmos. Environ. 34, 1281 – 1292.

A.P. Ba´ez et al. / Atmospheric Research 80 (2006) 67–85

85

Higashino, H., Tonooka, Y., Yanagisawa, Y., Ikeda, Y., 1997. Emission inventory of SO2 and NOx in East Asia with grid data system. In: Murano, K. (Ed.), Proceedings of the International Workshop on Unification of Monitoring Protocol of Acid Deposition and Standardization of Emission Inventory. Tsukuba, Japan, pp. 124 – 144. Hjellbrekke, A.G., Lo¨vblad, G., Sjo¨berg, K., Schaug, J., Skjelmoen, J.E., 1995. Data report 1993: Part 1. Annual summaries. EMEP/CCC-Report 7/95, Norwegian Institute for Air Research, pp. 4–18. Khwaja, H.A., Husain, L., 1990. Chemical characterization of acid precipitation in Albany, New York. Atmos. Environ. 24A, 1869 – 1882. Lee, S.D., 1993. Spatial variability of urban precipitation chemistry and deposition: statistical associations between constituents and potential removal processes of precursor species. Atmos. Environ. 27B, 321 – 337. Lee, B.K., Hong, S.H., Lee, D.S., 2000. Chemical composition of precipitation and wet deposition of major ions on the Korean peninsula. Atmos. Environ. 34, 563 – 575. Lynch, J.A., Grimm, J.W., Bowersox, V.C., 1995. Trends in precipitation chemistry in the United States: a national perspective, 1980–1992. Atmos. Environ. 29, 1231 – 1246. Magan˜a, V., Va´zquez, J., Pe´rez, J.L., Pe´rez, J., 2003. Impact of El Nin˜o on precipitation in Mexico. Geofı´s. Int. 42, 313 – 330. Peden, M.E., Bachman, S.R., Brennan, C.J., Demir, B., James, K.O., Kaiser, B.W., Lockard, J.M, Rothert, J.E., Sauer, J., Skowron, L.M., Slater, M.J., 1986. Development of standard methods for the collection and analysis of precipitation. Contract CR810780-01. Illinois State Water Survey, Analytical Chemistry Unit, Champaign, IL. Puxbaum, H., Simeonov, V., Kalina, F.M., 1998. Ten years trends (1984–1993) in the precipitation chemistry in Central Austria. Atmos. Environ. 32, 193 – 202. Stumm, W, Morgan, J.J., 1981. Aquatic Chemistry: An Introduction Emphasizing Chemical Equilibria in Natural Waters, second edition. A Wiley-Interscience Publication, John Wiley and Sons, New York, Chichester, Brisbane, Toronto, Singapore, pp 226–228. Terada, H., Ueda, H., Wang, Z., 2002. Trend of acid rain and neutralization by yellow sand in east Asia—a numerical study. Atmos. Environ. 36, 503 – 509. Tuncer, B., Bayar, B., Yesilyurt, C., Tuncel, G., 2001. Ionic composition of precipitation at the Central Anatolia (Turkey). Atmos. Environ. 35, 5989 – 6002. Walker, J.T., Aneja, V.P., Dickey, D.A., 2000. Atmospheric transport and wet deposition of ammonium in North Carolina. Atmos. Environ. 34, 3407 – 3418. Wang, W., Wang, T., 1996. On acid rain formation in China. Atmos. Environ. 30, 4091 – 4093.