- Email: [email protected]
The composition of bulk precipitation on a coastal island with agriculture compared to an urban region
Atmospheric Environment Vol. 24A, No. 12, pp. 3021 3031, 1990
00046981/90 $3.00+0.00 Pergamon Press plc
Printed in Great Britain.
THE COMPOSITION OF BULK PRECIPITATION ON A COASTAL ISLAND WITH AGRICULTURE COMPARED TO AN URBAN REGION E. P. WEIJERS a n d H. F. VUGTS Department of Meteorology, Institute of Earth Sciences, Free University, De Boelelaan 1085, 1081 HV Amsterdam, The Netherlands (First received 3 June 1989 and in final form 13 June 1990)
Abstract Results of chemical analyses of monthly bulk samples from Schiermonnikoog, one of the islands in the northern part of The Netherlands, are interpreted. The continuous record covers a period of more than 15 years. A comparison (10 years) is made with Ouderkerk, a village near Amsterdam. Non-sea salt contributions, relations between ion species, long-time trends, annual cycles and meteorological influence are discussed. The study reveals enhanced levels of ammonium in the Schiermonnikoog samples with respect to Ouderkerk. Also, concentrations of sulfate and nitrate were higher. The high concentrations of ammonium are ascribed to dry-deposited NH 3 caused by cattle breeding, the only economical activity on the island. A significant positive trend reflects its intensifyingnature. Annual cycles and statistical computations indicate prior combination of parts of ammonium and excess sulfate as ammonium sulfate. The nitrate content appears to be strongly related to ammonium (r= 0.79). In the Ouderkerk dataset this correspondence is much weaker (0.37), whereas its pH values are systematically lower. It is therefore believed that on Schiermonnikoog concentrations of nitrate are increased by nitrification of ammonium in the collector. Annual cycles of sodium, magnesium and chloride, and to a lesser extent potassium, are very similar (maximum concentrations in November, December and January, and a relative maximum in April). The other annual patterns peak in the first half of the year: maximum concentrations are found in February (ammonium, excess sulfate), June (nitrate), January (potassium) and in April (excess calcium). A combination of frequently occurring offshore winds and low precipitation amounts will account for this behavior. Key word index: Bulk precipitation, NH3, agriculture, annual cycle; The Netherlands.
1. INTRODUCTION
The role of the acidifying oxides and their reaction products has been intensely studied for several decades. Part of this research effort is devoted to monitoring of chemical composition of precipitation preferably in areas that are not dominated by polluting h u m a n activities. The chemical rain collector on Schiermonnikoog, one of the small islands in the northern part of The Netherlands (Fig. 1), is situated far from major industrial sources and densely populated areas but near agricultural land. At the time the station was raised (1971) it was anticipated that the degree of pollution measured would be lower than in industrialized areas on the mainland. This appears to be untrue and in the light of recent studies we try to elucidate the reasons. The primary goal of this paper is to document those characteristics of the chemical composition that are determined by positional, economical and meteorological circumstances on the island. To this purpose we evaluated results obtained from chemical analyses on monthly bulk samples. The period of observation extended over more than 15 years (May 1972-December 1987). As the collector is placed at the site of a climatological weather station we were also able to study possible effects of some
meteorological parameters. Further we undertook a comparative study on bulk samples collected in the region southeast of Amsterdam (Ouderkerk, Fig. 1) for the years 1973-1982 inclusive.
2. METHODS The island of Schiermonnikoog (53.47°N, 6.19°E) is 16 km long, 4 km wide and oriented in a west-east direction. It consists of sand beaches, dunes, meadows and salt marshes. The distance between Schiermonnikoog and the mainland is 8-10 km. The island is thinly inhabitated. About 800 people live in a village on the west side of the island. No industry exists and cars are rarely used. Seven farms (cattlebreeding) are all situated in the southern part of the island. In 1971 the Department of Meteorology of the Free University in Amsterdam established a climatological weather station on the island located 2 km east of the village at the northern edge of the farming area. The nearest farm buildings are some 200 m southwest of the research site. The Schiermonnikoog station operated continuously with the result that a 16-year long series of measurements is now at our disposal. This includes the records of meteorological parameters as wind
3021
3022
E.P. WEIJERSand H. F. VUGTS $chiermonnikob~.
Z
,,3"
w'w"°,/ Amster
52"
==
m
Ouderkerk . ~ T h e HacJue • Rotterdam
/" "/i" ,~,.. ~
// t //v 4*
, 50kin ,
r ~t...:
i
Fig. 1. Location of sites.
speed and direction, air temperature and relative humidity. Rainfall amount is collected with a Dutch standard rain sampler (rim of the collector is 40 cm above the surface, area 200cm 2, daily measuring routine), From 1972 onwards a rainwater collector is available for the collection of monthly bulk samples (the rim of the collector is 1.50 m above the surface, dry deposition takes place in the open funnel). For the period May 1972-January 1983, our institute also gathered precipitation chemistry data with a collector placed in Ouderkerk, 4 km southeast of Amsterdam. The location of Ouderkerk may be considered as rural though urban influences of Amsterdam are to be expected. There are no polluting activities in the immediate surroundings. Collector and analyzing procedures are identical with the Schiermonnikoog set-up. After each month the collecting bottle was replaced and the funnel and conducting parts rinsed with distilled, deionized water. Precautions were taken against macroscopic contaminants (plastic grid over the funnel rim), and against birds (sharp funnel rim). The polyethylene bottle was placed in the darkness in the inside of the collector. During transport the sample was subject to similar conditions. The samples were subsequently analyzed to determine their pH, from which the concentrations of H + were derived, together with their ionic concentrations of Ca 2+, K +, Mg 2+, Na +, CI-, NH2, SO~- and N O r . Their PO~- content of the sample was determined to detect possible contamination by birds (Asman et al., 1982). All chemical analyses were performed in the laboratory of the Free University of Amsterdam. The quality of the analyzing procedures was controlled by interlaboratory comparisons. The non-sea salt contributions (indicated with *) have been calculated for every month by using the reported sodium concentrations as a tracer. It is assumed that
all sodium originates from sea salt and that no fractionation of the sea salt aerosols occurs. If the excess contribution is only a minor part of the concentration found in the monthly sample, it may be inaccurate due to accumulating analysis errors. For the major ions, analysis accuracy was tested by calculating the charge balance (CB). The concentrations of cations and anions should balance, and the water must be electrically neutral. The charge balance is expressed as a percentage of the sum of cations and anions: CB (%) =
sum cations + sum anions sum cations
-
sum anions
x 100,
where cations and anions are in #eq # - 1. When data of a particular month were obviously contaminated or one of the analyses was missing, the month was removed. We calculated the CB on the cleaned-up dataset. For Schiermonnikoog its mean was - 0 . 3 % (standard deviation: 6.4%). Monthly samples of which the CB deviated more than twice the standard deviation were left out from this study. With concern to Schiermonnikoog these procedures ultimately resulted in a database that contained the data of 152 months for each chemical constituent (deleted: 19% of total). For Ouderkerk 112 samples were available (14% deleted). The reliability of the data has been examined further in several ways including the dividing of precipitation samples and making separate measurements, and using reference samples without prior knowledge of the analysts. These tests showed that analytical precision is approximately 2.5% (pH, sulfate and chloride), 4% (sodium, magnesium and calcium), 5% (ammonium), 10% (nitrate), and 15% (potassium). To obtain some idea on the experimental accuracy it is tempting to compare with chemistry precipitation data from adjacent stations. However, due to differences in construction, location, sampling and analytical procedures results should be interpreted with caution. The comparison is made with a collector located in Ouderkerk maintained by the Provincial Department of Environmental Control in NorthHolland (PW). Daily samples were collected here and ultimately combined to a monthly sample. Precipitation weighted averages of 18 (in the case of sodium and ammonium), and 26 (calcium, sulfate, nitrate) months, respectively, have been compared with results obtained with our collector (FU) in Ouderkerk. The test shows that differences between the PW and the FU measurements divided by the mean were approximately: - 5 % (calcium), - 2 5 % (chloride), + 4 % (ammonium), + 5% (nitrate), and .- 7% (sulfate). So representativity seems to be of satisfactory level. The systematic negative difference found for the sea salt related components may well be explained by a smaller catchment of dry deposition by the PW collector due to the daily cleaning routine.
Composition of bulk precipitation on a coastal island
3023
From Fig. 2 we also deduce that acids and bases of marine origin merely neutralize each other. The difference between the sum of cations (H ÷ excluded) and the 3.1. General considerations For Schiermonnikoog weighted concentrations sum of anions is larger at Ouderkerk. A stronger acidic (XwA,/~mol ~-1) and standard deviationst (SDwA) of tendency with respect to Schiermonnikoog is conthe XWA are summarized in Table 1 (the period firmed by the frequency distribution o f p H values over monitored is May 1972-December 1987). To deter- 112 months (Fig. 3). The use of the pH parameter mine the importance of sources other than the sea, without further notice may be misleading: the lowest non-sea salt contributions to the calcium, magnesium, pH values are found at Ouderkerk but pollution of potassium, chloride and sulfate content were also samples is larger on the island of Schiermonnikoog. Independent studies have revealed that bulk preestimated. The weighted averaging procedure normalcipitation from Schiermonnikoog contained more izes the concentration by the monthly rainfall amount chemical components than expected at a station in this as measured by the Dutch standard rain collector. We compared the XWA for each ion with those remote area (no heavy industry is present) of The calculated for Ouderkerk. We therefore tested the null Netherlands. (1) For the period 1978-1982 the island of Schierhypothesis that there is no difference in the XWA values between both sites. Using a 1 per cent level we monnikoog participated in a precipitation sampling find that the null hypothesis is rejected for CI-, Na +, network of the Dutch Water Supply Institute (van de Mg 2 +, and NH~, and at the 5 per cent level for NO 3 Meent, 1984). The collector was located 2 km westand SO~-. At Ouderkerk only CI-* (p<0.01) is wards from our experimental site. In general, sampling higher. The XWA of Ca 2 ÷* and SO 2-* are of the same stations of this network were located away from urban and industrial source areas. Precipitation weighted magnitude. The calculated non-sea salt contributions are concentrations of 2-weekly samples were calculated. largest for sulfate (76%) and calcium (72%). For From the 27 stations, Schiermonnikoog showed the potassium 27% is found and in the case of chloride highest levels of ammonium and nitrate, whereas and magnesium less than 7% is expected to be from chloride, sodium, potassium, calcium, magnesium and another source. At Ouderkerk non-sea salt parts sulfate were always ranked among the 'top three' of become more important: 91% (sulfate), 90% (calcium), concentration levels. (2) From July 1972 onwards a rain collector was 53% (potassium), 35% (magnesium) and 25% (chlorlocated on the experimental site of the Schiermonniide). At both stations the correction for sea salt koog station maintained by the Royal Netherlands sometimes led to negative monthly concentrations Meteorological Institute. After making a comparison for magnesium, potassium and chloride, mainly durwith the W M O station Witteveen (see Fig. 1), Ridder ing the winter period. With the help of bar diagrams (Fig. 2) we can (1974, pers. comm.) concluded that a stronger envisualize the comparison with Ouderkerk (monitoring richment of the collected rainwater took place on the period for both stations is May 1972-January 1983) island of Schiermonnikoog. Only excess sulfate was and interpret the chemical compositions of wet and approximately of the same level. dry deposition in terms of source type (marine or nonmarine) and charge (cations and anions). Concentra- Table 1. Precipitation weighted averages (Xw^) and stantions are in #eq f - 1 . The figure shows that in the dard deviations (SDwA) of monthly concentrations samples of Schiermonnikoog nearly twice as many (#mol d- 1) of the dissolved substances in bulk precipitation constituents are found, which is almost entirely due to for Schiermonnikoog during May 1972-December 1987. * indicates non-sea salt contributions. Number of months of the sea salt components (CI-, Na ÷, Mg 2 +). While on measurements is 152 Schiermonnikoog the total contribution from the sea (65%) is larger than the other input sources, at XWA SDwA~ SD/X (%) Ouderkerk it accounts for 42% (by molar concentra4.46 0.04 1.0 pH tion). Ca 2+ 27.5 1.2 4.2 Mg2 + 43.0 3.3 7.6 i The standard deviations of volume-weighted averages K + 10.6 0.7 6.3 (SDwA) are calculated according to Na + 354.0 28.2 8.0 149.1 7.6 5.1 I N // N X~l 1/2 NH,~ 435.2 31.2 7.2 90.7 4.4 4.9 72.3 3.9 5.4 SDwA = ~ j , NO3 19.7 0.9 4.7 C a 2+* i Mg 2+* 2.8 0.6 21.7 K +* 2.9 0.4 15.3 where Pi is the precipitation amount corresponding to the ith CI - * 22.1 4.9 22.2 sample, xl is the chemical constituent of interest and N is the SO42- * 69.3 3.2 4.7 number of samples (Galloway et al., 1985). 1 RESULTS
L ( lp, (N-I)
E.P. WEIJERSand H. F. VUGTS
3024
F--~hydronium ~ammonium 500-
potassium ~]
400-
•
magnesium
nitrate
~ ' ~ ] t a l c iu m
[////
sulfate
sodium
FFTT] c h l o r i d e
c
300-
o
._> 0"
200-
o u
\\\
100-
\ \\
~x \\ •
irrrr
,\\
non-marine
marine
marine
Schier monnikoog
non - m a r i n e Ouderkerk
Fig. 2. Marine and non-marine concentrations (in/~eq f - ~) found on the island of Schiermonnikoog and at Ouderkerk. The marine cationic contribution (left bar, from bottom to top) contains sodium, and the sea salt fractions of calcium, magnesium and potassium, whereas the marine anionic part (right bar, from bottom to top) is given by the sea salt parts of chloride and sulfate. The non-marine contributions are built up by the non-sea salt fractions of the above-mentioned components (sodium excluded), ammonium and hydronium (placed on top of the left bar), and nitrate (right bar). 26 24\
22 20
\
\
\
\
\
\
\4
\
\
\
\
\
\A
\A
18 16 14. .o
E
12
// // // .// ,// //
c
10 8 6 4
\ 2 0
I
1
I
I
3.2
3.4"
3.6
3.8
4-
4".2
Ouderkerk
,// ,// // // // // ,/// ,¢f// 1
l
I
I
I
I
I
I
I
4.4
4.8
4..8
5
5.2
5.4
5.6
5.8
6
\/
/
~
Schiermonnikoog
Fig. 3. Frequency distribution of pH values for Schiermonnikoog and Ouderkerk. O u r comparison between the datasets gathered at Schiermonnikoog and Ouderkerk confirms these findings; it rules out the possibility of serious errors in sampling procedures. So despite the fact that this island was selected initially as having the potential to provide data representative of clean conditions there is evidence that this is not the case. It was suspected
that the intensive farming would be the m a m cause as no other polluting activity is found on the island.
3.2. Correlations between the ion species Correlation analysis was done for two purposes. First, to detect possible c o m m o n sources or inorganic
Composition of bulk precipitation on a coastal island compounds and second, to determine which components are responsible for the acidity content. Calculating skewness and kurtosis revealed lognormal distributions for every concentration set. Before calculating Pearson product-moment correlations all parameters were therefore log-transformed in order to attain the required normal distribution. Coefficients and significance levels are shown in Table 2 for Schiermonnikoog and in Table 3 for Ouderkerk. For the non-sea salt fractions of magnesium, potassium and chloride it was not possible to use a logtransformation because of the occurrence of negative values. On the island of Schiermonnikoog the largest correlation coefficients (>0.92) are found for chloride, sodium and magnesium confirming that these components descend almost entirely from the sea salt aerosol. At Ouderkerk the influence of other sources decreases statistical significances. After subtracting the sea salt fractions from calcium and sulfate, relationships with chloride, sodium and magnesium have become lower as expected. However, it is remarkable that on the island of Schiermonnikoog significant correlations (>0.37) still remain between excess sulfate and sea salt components. This seems to be a local effect since at Ouderkerk coefficients are much smaller ( < 0.25). At both sites significant correlations (>0.41) between the non-sea salt fractions of calcium and sulfate are found (the highest at Ouderkerk). This may indicate the presence of non-acidic sulfates. In general, relationships of ammonium and nitrate with the sea salt originating components are weak. On Schiermonnikoog relatively large coefficients are found for ammonium and sulfate (0.65), improving when sulfate is corrected for sea salt (0.69), ammonium and nitrate (0.79), and excess sulfate and nitrate (0.55). At Ouderkerk the correlation coefficient for ammonium and excess sulfate is about the same but relations of ammonium and excess sulfate with nitrate are much weaker (0.37, respectively 0.23) here. Apparently the strong correspondence of ammonium and sulfate with nitrate is locally determined. In the Schiermonnikoog samples weak or insignificant relationships are detected between the presence of H ÷ ions, and the content of ammonium, excess sulfate and nitrate. Statistical evidence (r=0.79) exists for a relation between NH~ and the sum of 2(SO 2-*) and NO3. This is merely a restatement that the ion balance holds. However, adding H ÷ to NH~ and recalculating the correlation coefficient between both sums improves the statistical strength (r=0.87), thereby suggesting some hidden complementary relationship which concerns H ÷. To find evidence for this assumption and also to uncover possibly hidden and spurious relationships we used partial correlation techniques. Partial correlation coefficients measure the relationship between two variables while controlling for the possible effects of other variables. These are controlled by removing the
3025
linear relationship with the other variables before calculating the correlation coefficients between the two variables of interest. For example, we find that the correspondence between excess sulfate and nitrate disappears (0.55 ~ 0.03) when the relationship with ammonium is removed, indicating that the apparent correlation between nitrate and sulfate is due to the latter's positive correlation with ammonium. The result is also more in line with the Ouderkerk findings. It appears that correlation coefficients for hydrogen ions with excess sulfate and nitrate, grow (0.15 ~ 0.30, resp. 0.22~0.48) when the effect of ammonium is controlled for. When one of the strong acids is held constant, ammonium correlates more negatively with hydrogen ions ( - 0.09 ~ 0.25, resp. ~ - 0.40). After removing the possible effects of all the remaining major ions we obtain correlation coefficients as given in Table 4. Statistical strength has been increased considerably compared to Tables 2 and 3. Coefficients are of the correct sign as they should if excess sulfate, nitrate and ammonium were originally present as SO 2, NOx and NH 3. At Ouderkerk the strongest correlation with H ÷ is found for nitrate. Summarizing, the following mechanisms in the Schiermonnikoog samples are suggested by the correlation analysis: (1) the sea salt-related components form an interconnected group; (2) an interaction exists between the ammonium and the sulfate ions (also found at Ouderkerk); (3) an interaction exists between the ammonium and the nitrate ions (not apparent in the Ouderkerk data set); (4) an interaction between sea salt and sulfate exists that remains after correcting sulfate for the sea salt contribution (not measured at Ouderkerk); (5) most of the variation in the hydrogen ion concentrations is due to variations in nitrate, excess sulfate and ammonium. 3.3. Annual cycles and meteorological influences Precipitation weighted monthly concentrations were calculated in order to obtain annual cycles at both sites. The normalized concentrations (expressed as a percentage of the Xw^ in Table 1) are shown in Fig. 4. The ammonium content in the bulk samples is high in the first half of the year. The month with the lowest concentration is October. On Schiermonnikoog a pronounced maximum is found in February. At Ouderkerk the variability is larger which may be due to the smaller record length. Higher excess sulfate levels occur during the first 4 months of the year while the lowest values are found in July, August and September. Its pattern is very similar to ammonium but contrasts with that of nitrate: maximum concentrations in the early summer period and minimum values in December. No annual trend was evident in the hydronium concentrations but at both stations
3026
E.P. WEIJERSand H. F. VUOTS
Table 2. Correlation coeffÉcients (significance levels) between ion species in bulk samples collected on the island of Schiermonnikoog. Number of months of measurements is 152
H Ca Mg K Na NH4 CI SO4 NO 3 Ca* SO*
H 1.000 (0.000) 0.073 (0.382) -0.059 (0.484) 0.000 (0.998) -0.042 (0.616) -0.092 (0.269) -0.039 (0.643) 0.125 (0.132) 0.218 (0.008) 0.090 (0.286) 0.145 (0.082)
Ca 0.073 (0.382) 1.000 (0.000) 0.466 (0.000) 0.493 (0.000) 0.412 (0.000) 0.309 (0.000) 0.447 (0.000) 0.609 (0.000) 0.390 (0.000) 0.920 (0.000) 0.569 (o.ooo)
Mg -0.057 (0.484) 0.466 (0.000) 1.000 (0.000) 0.637 (0.000) 0.941 (0.000) 0.292 (0.000) 0.927 (0.000) 0.647 (0.000) 0.224 (0.006) 0.152 (0.063) 0.437 (o.ooo)
K 0.000 (0.998) 0.493 (0.000) 0.637 (0.000) 1.000 (0.000) 0.593 (0.000) 0.341 (0.000) 0.618 (0.000) 0.548 (0.000) 0.313 (0.000) 0.286 (0.000) 0.422 (o.ooo)
Na -0.042 (0.616) 0.412 (0.000) 0.941 (0.000) 0.593 (0.000) 1.000 (0.000) 0.250 (0.002) 0.974 (0.000) 0.608 (0.000) 0.172 (0.034) 0.074 (0.365) 0.374 (o.ooo)
NH, -0.092 (0.268) 0.309 (0.000) 0.292 (0.000) 0.341 (0.000) 0.250 (0.002) 1.000 (0.000) 0.269 (0.000) 0.647 (0.000) 0.790 (0.000) 0.264 (0.001) 0.685 (o.ooo)
CI -0.039 (0.643) 0.447 (0.000) 0.927 (0.000) 0.618 (0.000) 0.974 (0.000) 0.269 (0.001) 1.000 (0.000) 0.625 (0.000) 0.175 (0.031) 0.115 (0.161) 0.398 (o.ooo)
SO4 0.125 (0.132) 0.609 (0.000) 0.647 (0.000) 0.548 (0.000) 0.608 (0.000) 0.647 (0.000) 0.625 (0.000) 1.000 (0.000) 0.510 (0.000) 0.437 (0.000) 0.956 (o.ooo)
NOa 0.218 (0.008) 0.390 (0.000) 0.224 (0.006) 0.313 (0.000) 0.172 (0.034) 0.790 (0.000) 0.175 (0.031) 0.510 (0.000) 1.000 (0.000) 0.382 (0.000) 0.554 (o.ooo)
Ca* 0.089 (0.286) 0.920 (0.000) 0.152 (0.063) 0.286 (0.000) 0.074 (0.365) 0.264 (0.001) 0.115 (0.161) 0.437 (0.000) 0.382 (0.000) 1.000 (0.000) 0.498 (o.ooo)
SO~ 0.145 (0.082) 0.569 (0.000) 0.437) (0.000) 0.422 (0.000) 0.374 (0.000) 0.685 (0.000) 0.398 (0.000) 0.956 (0.000) 0.554 (0.000) 0.498 (0.000) 1.000 (o.ooo)
Table 3. Correlation coefficients (significance levels) between ion species in bulk samples collected at Ouderkerk. Number of months of measurements is 112
H Ca Mg K Na NHa CI SO4 NO s Ca* SO*
H Ca 1.000 -0.071 (0.000) (0.463) -0.071 1.000 (0.463) (0.000) 0.148 0.278 (0.138) (0.004) -0.116 0.329 (0.239) (0.001) 0.116 -0.060 (0.236) (0.532) -0.115 0.416 (0.237) (0.000) 0.161 0.213 (0.096) (0.025) 0.113 0.406 (0.242) (0.000) 0.480 0.405 (0.000) (0.000) -0.076 0.984 (0.433) (0.000) 0.093 0.412 (0.338) (0.000)
Mg 0.148 (0.138) 0.278 (0.004) 1.000 (0.000) 0.380 (0.001) 0.428 (0.000) 0.182 (0.062) 0.609 (0.000) 0.356 (0.000) 0.217 (0.026) 0.181 (0.065) 0.253 (0.009)
K -0.116 (0.239) 0.329 (0.001) 0.380 (0.001) 1.000 (0.000) 0.360 (0.000) 0.322 (0.001) 0.381 (0.000) 0.273 (0.004) 0.397 (0.000) 0.243 (0.012) 0.195 (0.044)
Na 0.116 (0.236) -0.060 (0.532) 0.428 (0.000) 0.360 (0.001) 1.000 (0.000) 0.002 (0.982) 0.816 (0.000) 0.142 (0.137) -0.017 (0.857) -0.198 (0.038) -0.018 (0.854)
maxima occurred in June coinciding with the time that nitrate shows its maximum. The a n n u a l pattern of chloride is very similar to those of sodium and magnesium. Cycles show peak values during the winter period (November, December and January) with a relative maximum in April.
NH,, -0.115 (0.237) 0.416 (0.000) 0.182 (0.062) 0.322 (0.001) 0.002 (0.982) 1.000 (0.000) 0.012 (0.901) 0.727 (0.000) 0.368 (0.000) 0.418 (0.000) 0.738 (0.000)
CI 0.161 (0.096) 0.213 (0.025) 0.609 (0.000) 0.381 (0.000) 0.816 (0.000) 0.012 (0.901) 1.000 (0.000) 0.241 (0.011) 0.023 (0.814) 0.067 (0.488) 0.078 (0.416)
SO4 0.113 (0.242) 0.406 (0.000) 0.356 (0.000) 0.273 (0.004) 0.142 (0.137) 0.727 (0,000) 0.241 (0.011) 1.000 (0.000) 0.228 (0.016) 0.378 (0.000) 0.980 (0.000)
NO 3
0.480 (0.000) 0.405 (0.000) 0.217 (0.026) 0.397 (0.000) -0.017 (0.857) 0.368 (0.000) 0.023 (0.814) 0.228 (0.016) 1.000 (0.000) 0.395 (0.000) 0.231 (0.014)
Ca* -0.076 (0.433) 0.984 (0.000) 0.181 (0.065) 0.243 (0.012) -0.198 (0.038) 0.418 (0.000) 0.067 (0.488) 0.378 (0.000) 0.395 (0.000) 1.000 (0.000) 0.415 (0.000)
SO* 0.093 (0.338) 0.412 (0.000) 0.253 (0.009) 0.195 (0.044) -0.018 (0.854) 0.738 (0.000) 0.078 (0.416) 0.980 (0.000) 0.231 (0.014) 0.415 (0.000) 1.000 (0.000)
The pattern of potassium, though being slightly different, still shows a typical sea salt characteristic (maximum in January). F o r calcium, the maximum value found in April coincides with the characteristic extreme observed in the sea salt patterns. However, in winter concentrations remain low.
F
M
A
M
J
J
A
S
0
N
D
F
M
A
M J
J
A
S
0
N
D
J
A
S
0
J L~6~6
J
N
D
D
,~
o
'2°
20~
J
J
F
F
M A
M A
M
M
J
J
J
J
A
A
S
S
0
0
D
2Ot
N D
120-1
1401
16oq
~8ot
F
M
A
M
J
J
A
S
0
N
D F
M
J
A
S
O N
60-I
6O.
J
~q
80
80
M A
M J
J
A
S
0
N
D
S c h i e r monnikoo(3
Schiermonnikoog
Fig. 4. P r e c i p i t a t i o n a v e r a g e d m o n t h l y c o n c e n t r a t i o n s e x p r e s s e d a s p e r c e n t a g e s o f X w ^ .
Ouderkerk
204 F
204
20" 201
60-1 40-1
40t
e01
too-1
40.
J
140t 12o-1
6O-1
I
i00
I00
,41 12
140 120
160-1
100t
N itrote
2O-1
40-1
2o-1
4oq
120
o =
J40
160
160
J
Excess coLclum
20t
180
40q
2O
~
60t
40-
~
60-1
6o-1
j
J
F
M
F M
E x c e s s suLfote
N
60-1 40t
80-1
M J
M
$2O I00 ~q
80-
A
M A
3;M~Mj
F
40t
604
801
I00t
i201
160
~00-1 80t
J
160 140
Ammonium
100-
120"1
o
140t
~80t
Potassium
140-
180-
J
4O Z0t
40~
6O
2ol
SOq
6O~
oo
I00
~ 120-
W(
J
120
120~
,B1°
160 140
140-
ChLoride
160-
180-
A
J
J
J
J
A
A
Ouderkerk
M
A M
S
S
0
0
D
N 0
N
E
=__
8==
0
O
a
=..
o"
o
¢3 o
3028
E.P. WEIJERSand H. F. VUGTS
Table 4. Partial correlations for hydronium with the other ion species at the stations Schiermonnikoog and Ouderkerk
NH 4 SO* NO 3 Ca* Mg K Na Cl
Schiermonnikoog
Ouderkerk
H -0.573 0.422 0.540 - 0.211 -0.168 -0.013 - 0.046 0.076
H -0.444 0.434 0.757 - 0.377 -0.059 -0.463 - 0.063 0.246
Though the XwA of the non-sea salt fractions are relatively small (Table 1), annual periodicity was apparent with higher values occurring in the summer period. The reason for the negative concentrations occasionally found in winter is unknown. Systematically overestimating the presence of sodium due to a non-sea salt source or an error in the measurement method affects the estimated non-sea salt contribution in a negative direction. However, this does not explain the observed annual variability. Annual periodicities with peak values in the first half of the year seem characteristic for most components in the European area (e.g. Moldan et a1.,1987). In general, our results reasonably agree with other reports, e.g. after analyzing EACN network data Soderlund and Granat (1982) found a maximum for ammonium in the early spring (March and April) and a minimum in August and September which is confirmed by a study from Buishand et al. (1988) for the Dutch area. The spring peak in sulfate concentrations is in agreement with the annual cycle (maximum in February-May, minimum in July-October) found by Rohde and Granat (1984). Skeffington (1984) found a late spring peak for nitrate (May-June, averaged over 4 years). In England Brimblecombe and Pitman (1980) found winter maxima and summer minima for chloride. To study the influence of monthly precipitation amount on concentration, correlation coefficients between the log-transforms of these variables were calculated. For every component a negative correlation was found. Being significant at the 1% level we found the following order: Ca 2 + * < SO 2- * < N O 3 < N H 2 . For excess calcium this is probably a consequence of the prevailing terrigenous nature of this element. The strength of its source will be strongly influenced by a seasonal pattern in the appearance of dry periods. The lowest rainfall amounts are measured in April which obviously affects the presence of sea salt material in the sample. Nevertheless the lowest correlation with precipitation amount is detected for these constituents. Levels of these components are enhanced when monthly rainfall amount rises ( > 70 mm) and when averaged monthly wind speed during rainfall increases ( > 6 m s - t ) . This combination often takes place in
winter when seawater temperature is higher than air and a tendency for fairly strong vertical mixing with higher wind speed just above the water surface exists. For excess sulfate and ammonium hardly any effect of the wind speed could be detected. Two counterproductive mechanisms are expected to be at work here. Ammonia emitted by a nearby source will remain longer in the surrounding air when wind speeds are low. On the other hand, a larger upward flux of ammonia will occur with increasing wind speeds. The third meteorological variable of interest is the direction of the wind. Monthly frequency distributions reveal that (south) westerly winds dominate most of the year. Wind roses of March and April do not show any dominancy whereas in February and May continental air masses cross the island relatively often. Contaminating influences from the mainland combined with the low precipitation amounts observed during these months may account for the higher concentrations. 3.4. Lon#-term behavior Figure 5 illustrates annual precipitation weighted concentrations on both stations for the most acidifying components and sodium, as representative oftbe sea salt species. Influence of rainfall amount is evident in the year 1976 which was extremely dry (530mm, normal 850 mm): for most components (except hydronium) the maximum annual concentration was measured. In 1980 the highest H + values were found at both stations, which coincides with relatively low ammonium levels. The annual variation of ammonium is by far the largest. On Schiermonnikoog weaker but corresponding variations in the ion concentrations of nitrate and sulfate can be seen. In the case of excess sulfate some similarity with the Ouderkerk pattern is deduced. Apparently effects from the mainland are also measured on the island. When 1976 and 1977, two drought years, are excluded, the levels of sodium (and the other sea salt) species barely changed over the monitoring period. A linear regression model has been used for a trend analysis on the monthly concentration data of the five components. Regression coefficients (RC) and significance levels (SL) are given in Table 5 for both Schiermonnikoog and Ouderkerk (period I: May 1972-January 1983) and for Schierrnonnikoog (period II: January 1983-December 1987). As linear regression is sensitive to the occurrence of outliers, the calculation was repeated on the logarithm of concentration values to see if drastic changes in significances (SLIog) took place. Significant levels (i.e. SL as well as SLIog are less than 1%) indicate increasing hydronium and nitrate concentrations in the Ouderkerk data record. The RC of ammonium and excess sulfate do not differ significantly from zero. The sea salt species show small
Composition of bulk precipitation on a coastal island 800
3029
sodium
600 400
•
_
• •
•
•
•
•
•
•
•
•
•
200-[]
[]
[]
[]
[]
[]
[]
[]
[]
0-ammonium
300 -•
200 -•
100
-
-
•
•
•
[] •
• []
• []
[]
[]
•
[]
[]
[]
• []
•
•
II
•
..
m
n
•
[]
0-E o E
nitrate
200 -•
100 f n ~
•
• []
[]
•
[]
0 excess sulfate 200 I IO0 i
)
~
•
•
•
[] m
[]
[] mm []
•
i
m
m
m
0 200 1
hydronium
. 73
74 •
75
76
77
78
Schiermonnikoog
79
80
811 []
82
83
84
85
86
87
Ouderke~
Fig. 5. Annualweighted averages of H +, NH~, NO~, SO~- *, and Na ÷ for the island of Schiermonnikoog and Ouderkerk. increases that are significant at the 10% level in the case of concentrations and at a level of less than 5% when the logarithmic value is taken. Apparently some effect of outliers has been measured here. Contrary to Ouderkerk, statistical evidence exists for relatively large increasing positive trends in the ammonium and sulfate concentrations on Schiermonnikoog. This might reflect the influence of the strongly intensifying cattle breeding culture during the observation period.
4. DISCUSSION AND C O N C L U S I O N S
The role of ammonia in atmospheric chemistry for the European area has recently drawn more attention (van Breemen et al., 1982; Moiler and Schieferdecker, 1985; Asman et al., 1987; ApSimon et al., 1987; Buijsman et al., 1987). The latter reference contains an enumeration of sources of atmospheric NH 3.
Animal waste and fertilizers are held responsible for 90% or more of the anthropogenic NH 3 emission. Some information on the emission density in The Netherlands is given in the inventory of Erisman (1989). Farmers on Schiermonnikoog use artificial fertilizers but under the influence of low pH values of the sandy soils the loss of ammonia from these fertilizers is expected to be relatively small. Adding animal manure (often sprayed on the land in a very dilute form) and artificial fertilizer to the agricultural grounds on Schiermonnikoog is also usage. The manure is known to have high pH values which favors the loss of NH 3. Close to sources, high concentrations ofNH 3 in air at ground level will occur because the plumes of these sources have not been mixed over the whole mixing layer yet (Asman and Janssen, 1987). This will lead to a relatively high dry deposition in the bulk collectors which are always open, and will cause the high NH,~ levels in our samples (not to be misinterpreted as a high contribu-
3030
E.P. WEIJERSand H. F. VUGTS
Table 5. Regression coefficients(RC) for concentrations (in /zmolmonth-t) and significance levels for both concentrations (SL) and the log-transformed concentrations (SLIos) of some major ions. The H ÷ concentration is calculated from the pH value. Periods of study: May 1972-January 1983 (I), and January 1983-December 1987 (II) Period I
RC
Ouderkerk SL
SLjog
H÷ NH2 SO2- * NO~Na ÷
0.545 0.316 0.066 0.374 0.476
-0.000 0.175 0.632 0.006 0.092
0.000 0.518 0.632 0.000 0.021
Period I
RC
Schiermonnikoog SL
SZlog
H÷ NH2 SO,2- * NO~Na ÷
0.372 1.001 0.482 0.617 0.810
0.005 0.013 0.016 0.001 0.307
0.041 0.032 0.004 0.000 0.216
Period II H* NH2 SO42-* NO~ Na ÷
RC 0.072 3.534 0.627 0.971 1.487
Schiermonnikoog SL SLIo8 0.745 0.000 0.076 0.010 0.415
0.602 0.000 0.078 0.030 0.252
tion of the local source to the NH~ concentration in precipitation). The simultaneous occurrence of certain economical and meteorological circumstances in the late winter months may explain the maximum concentrations of ammonium measured in the samples during this period. (1) Dumping of the surplus of stable manure on agricultural land often happens in this period when storage capacity is not large enough. The lack of storage volume has become more acute as numbers of livestock grew. Transport to the mainland for storage or use is not done for economical reasons. (2) Often the manure cannot be ploughed under on frozen and unworkable soils in winter. This will result in a net flux upwards of NH 3. The release of animal manure on snow-covered ground has recently been forbidden (winter 1989) by Dutch legislation to prevent the pollution of waterways by runoff as soon as thaw occurs. (3) Due to temperature differences between seawater and island soil occurring in the late winter period, relatively warm air caps the colder air above the island which results in low mixing heights and increasing concentrations just above the island. The emission of ammonia seems to influence the presence of sulfate and nitrate in the bulk samples. Correlation coefficients and annual patterns suggest that most of the ammonium was present as ammonium sulfate, which is probably formed by the inter-
action of ammonia with sulfur dioxide on the wetted funnel. A natural SO 2 source might be the inter-tidal fiats between the island and the mainland (about 1000 km2). Traditionally it has been assumed that the biological decomposition of sulfur-containingorganic matter results in emissions of hydrogen sulfide (H2S) but after the discovery of dimethylsulfide ((CH3)252) produced by algae in the surface seawater it is believed that this compound is the major natural component of sulfur emission in the atmosphere (Jorgensen, 1982). Some evidence for its presence might be derived from the observed correspondence between sea salts and excess sulfate which was not measured at the Ouderkerk station. A possible effect of such a local source will mainly be measured as dry deposition and not lead to higher SO,2- levels in collected rainwater as by the time the gases are oxidized, it will be far away from the island. In view of the difference observed in the annual cycles of ammonium and nitrate the presence of ammonium nitrate seems less obvious. The annual cycles reasonably agree with other reports. One ascribes the maximum values to larger levels of oxidants and to higher temperatures during spring and summer contributing to a faster oxidation of nitrate to acids. The additional nitrate amount may be due to a volatilization of nitrogen matter from the manure. A second cause may be some artefact in the sampling routine, i.e. a temperature-related nitrification of ammonium by bacteria which becomes more apparent with a field-residence time of a month. The reason why this effect is not observed in the Ouderkerk samples is probably the more acidic regime of the samples and the absence of nearby ammonia sources. This also explains that although the abundantly present ammonia is a base it does not cause a significant increase in the pH in the Schiermonnikoog samples with respect of Ouderkerk as the oxidation of ammonium to nitrate may have contributed some acid through: NH~ + 2 0 2 ~ N O 3 + H 2 0 + 2 H +.
Acknowledgements--We gratefully acknowledge Dr Ed Buijsman's critical comment on the first draft of this paper. Our co-worker, Dr Fred Cannemeijer, offered valuable comments regarding statistical evaluation. Many thanks to Tiny Baer and the laboratory staff for performing the analyses over such a long time. We are also grateful to the Provinciale Waterstaat van Noord Holland, Dienst voor de Milieuhygiene for providing us with their chemical precipitation data gathered at Ouderkerk. REFERENCES
ApSimon H. M., Kruse M. and Bell J. N. B. (1987) Ammonia emissions and their role in acid deposition. Atmospheric Environment 21, 1929-1946. Asman W. A. H. and Janssen A. J. (1987) A long-range transport model for ammonia and ammonium for Europe. Atmospheric Environment 21, 2099-2199. Asman W. A. H., Ridder T. B., Reijnders H. F. R. and Slanina J. (1982) Influence and prevention of bird-droppings in
Composition of bulk precipitation on a coastal island precipitation chemistry experiments. Wat. Air Soil Pollut. 17, 415-420. Breemen N. van, Burrough P. A., Velthorst E. J., van Dobben H. F., de Wit T., Ridder T. B. and Reijnders H. F. R. (1982) Soil acidification from atmospheric ammonium sulphate in forest canopy throughfall. Nature 229, 548-550. Brimblecombe P. and Pitman J. (1980) Long-term deposit at Rothamsted, southern England. Tellus 32, 261-267. Buijsman E., Maas H. F. M. and Asman W. A. H. (1987) Anthropogenic NH 3 emissions in Europe. Atmospheric Environment 21, 1009-1022. Buishand T. A., Kempen G. T., Reijnders H. F. R., Frantzen A. J. and van den Eshof A. J. (1988) Trend and seasonal variation of precipitation chemistry data in The Netherlands. Atmospheric Environment 22, 349-358. Erisman J.-W. (1989) Ammonia emissions in The Netherlands in 1987 and 1988, Report 228471006. National Institute of Public Health and Environmental Protection, Bilthoven, The Netherlands. Galloway J. N., Likens G. E. and Hawley M. (1985) Acid precipitation: natural versus anthropogenic components. Science 226, 829. Jorgensen B. B. (1982) The production and fate of reduced C, N, and S gases. In Atmospheric Chemistry (edited by E. Goldberg), pp. 225. Dahlem Konferenzen, SpringerVerlag, Berlin.
AE(A)
21:12-I
3031
van de Meent D., van Oosterwijk J. and Aldenberg T. (1984) Rijksinstituut voor de drinkwater voorziening, RIDVEWlN Meetnet regenwater 1978-1982. Samenvatting en statistische bewerking van de meetresultaten, ECOWAD8401 RIVM (in Dutch). Moldan B., Vesely M. and Bartonova A. (1987) Chemical composition of atmospheric precipitation in Czechoslovakia, 1976-1984--1. Monthly samples. Atmospheric Environment 21, 2383-2395. Moiler D. and Schieferdecker H. (1985) A relationship between agricultural NH 3 emissions and the atmospheric SO 2 content over industrial areas. Atmospheric Environment 19, 48-55. Rohde H. and Granat L. (1984) An evaluation of sulfate in European precipitation 1955-1982. Atmospheric Environment 18, 2627-2639. Sketiington R. A. (1984)The chemistry of bulk precipitation at a site in southeast England--lI. Relationships between ions and comparison with other sites. Atmospheric Environment 18, 1695-1704. Soderlund R. and Granat L. (1982) Ammonium (NH2) in precipitation--a presentation of data from the European Air Chemistry Network. Report CM-59. Department of Meteorology, University of Stockholm/International Meteorological Institute in Stockholm.
00046981/90 $3.00+0.00 Pergamon Press plc
Printed in Great Britain.
THE COMPOSITION OF BULK PRECIPITATION ON A COASTAL ISLAND WITH AGRICULTURE COMPARED TO AN URBAN REGION E. P. WEIJERS a n d H. F. VUGTS Department of Meteorology, Institute of Earth Sciences, Free University, De Boelelaan 1085, 1081 HV Amsterdam, The Netherlands (First received 3 June 1989 and in final form 13 June 1990)
Abstract Results of chemical analyses of monthly bulk samples from Schiermonnikoog, one of the islands in the northern part of The Netherlands, are interpreted. The continuous record covers a period of more than 15 years. A comparison (10 years) is made with Ouderkerk, a village near Amsterdam. Non-sea salt contributions, relations between ion species, long-time trends, annual cycles and meteorological influence are discussed. The study reveals enhanced levels of ammonium in the Schiermonnikoog samples with respect to Ouderkerk. Also, concentrations of sulfate and nitrate were higher. The high concentrations of ammonium are ascribed to dry-deposited NH 3 caused by cattle breeding, the only economical activity on the island. A significant positive trend reflects its intensifyingnature. Annual cycles and statistical computations indicate prior combination of parts of ammonium and excess sulfate as ammonium sulfate. The nitrate content appears to be strongly related to ammonium (r= 0.79). In the Ouderkerk dataset this correspondence is much weaker (0.37), whereas its pH values are systematically lower. It is therefore believed that on Schiermonnikoog concentrations of nitrate are increased by nitrification of ammonium in the collector. Annual cycles of sodium, magnesium and chloride, and to a lesser extent potassium, are very similar (maximum concentrations in November, December and January, and a relative maximum in April). The other annual patterns peak in the first half of the year: maximum concentrations are found in February (ammonium, excess sulfate), June (nitrate), January (potassium) and in April (excess calcium). A combination of frequently occurring offshore winds and low precipitation amounts will account for this behavior. Key word index: Bulk precipitation, NH3, agriculture, annual cycle; The Netherlands.
1. INTRODUCTION
The role of the acidifying oxides and their reaction products has been intensely studied for several decades. Part of this research effort is devoted to monitoring of chemical composition of precipitation preferably in areas that are not dominated by polluting h u m a n activities. The chemical rain collector on Schiermonnikoog, one of the small islands in the northern part of The Netherlands (Fig. 1), is situated far from major industrial sources and densely populated areas but near agricultural land. At the time the station was raised (1971) it was anticipated that the degree of pollution measured would be lower than in industrialized areas on the mainland. This appears to be untrue and in the light of recent studies we try to elucidate the reasons. The primary goal of this paper is to document those characteristics of the chemical composition that are determined by positional, economical and meteorological circumstances on the island. To this purpose we evaluated results obtained from chemical analyses on monthly bulk samples. The period of observation extended over more than 15 years (May 1972-December 1987). As the collector is placed at the site of a climatological weather station we were also able to study possible effects of some
meteorological parameters. Further we undertook a comparative study on bulk samples collected in the region southeast of Amsterdam (Ouderkerk, Fig. 1) for the years 1973-1982 inclusive.
2. METHODS The island of Schiermonnikoog (53.47°N, 6.19°E) is 16 km long, 4 km wide and oriented in a west-east direction. It consists of sand beaches, dunes, meadows and salt marshes. The distance between Schiermonnikoog and the mainland is 8-10 km. The island is thinly inhabitated. About 800 people live in a village on the west side of the island. No industry exists and cars are rarely used. Seven farms (cattlebreeding) are all situated in the southern part of the island. In 1971 the Department of Meteorology of the Free University in Amsterdam established a climatological weather station on the island located 2 km east of the village at the northern edge of the farming area. The nearest farm buildings are some 200 m southwest of the research site. The Schiermonnikoog station operated continuously with the result that a 16-year long series of measurements is now at our disposal. This includes the records of meteorological parameters as wind
3021
3022
E.P. WEIJERSand H. F. VUGTS $chiermonnikob~.
Z
,,3"
w'w"°,/ Amster
52"
==
m
Ouderkerk . ~ T h e HacJue • Rotterdam
/" "/i" ,~,.. ~
// t //v 4*
, 50kin ,
r ~t...:
i
Fig. 1. Location of sites.
speed and direction, air temperature and relative humidity. Rainfall amount is collected with a Dutch standard rain sampler (rim of the collector is 40 cm above the surface, area 200cm 2, daily measuring routine), From 1972 onwards a rainwater collector is available for the collection of monthly bulk samples (the rim of the collector is 1.50 m above the surface, dry deposition takes place in the open funnel). For the period May 1972-January 1983, our institute also gathered precipitation chemistry data with a collector placed in Ouderkerk, 4 km southeast of Amsterdam. The location of Ouderkerk may be considered as rural though urban influences of Amsterdam are to be expected. There are no polluting activities in the immediate surroundings. Collector and analyzing procedures are identical with the Schiermonnikoog set-up. After each month the collecting bottle was replaced and the funnel and conducting parts rinsed with distilled, deionized water. Precautions were taken against macroscopic contaminants (plastic grid over the funnel rim), and against birds (sharp funnel rim). The polyethylene bottle was placed in the darkness in the inside of the collector. During transport the sample was subject to similar conditions. The samples were subsequently analyzed to determine their pH, from which the concentrations of H + were derived, together with their ionic concentrations of Ca 2+, K +, Mg 2+, Na +, CI-, NH2, SO~- and N O r . Their PO~- content of the sample was determined to detect possible contamination by birds (Asman et al., 1982). All chemical analyses were performed in the laboratory of the Free University of Amsterdam. The quality of the analyzing procedures was controlled by interlaboratory comparisons. The non-sea salt contributions (indicated with *) have been calculated for every month by using the reported sodium concentrations as a tracer. It is assumed that
all sodium originates from sea salt and that no fractionation of the sea salt aerosols occurs. If the excess contribution is only a minor part of the concentration found in the monthly sample, it may be inaccurate due to accumulating analysis errors. For the major ions, analysis accuracy was tested by calculating the charge balance (CB). The concentrations of cations and anions should balance, and the water must be electrically neutral. The charge balance is expressed as a percentage of the sum of cations and anions: CB (%) =
sum cations + sum anions sum cations
-
sum anions
x 100,
where cations and anions are in #eq # - 1. When data of a particular month were obviously contaminated or one of the analyses was missing, the month was removed. We calculated the CB on the cleaned-up dataset. For Schiermonnikoog its mean was - 0 . 3 % (standard deviation: 6.4%). Monthly samples of which the CB deviated more than twice the standard deviation were left out from this study. With concern to Schiermonnikoog these procedures ultimately resulted in a database that contained the data of 152 months for each chemical constituent (deleted: 19% of total). For Ouderkerk 112 samples were available (14% deleted). The reliability of the data has been examined further in several ways including the dividing of precipitation samples and making separate measurements, and using reference samples without prior knowledge of the analysts. These tests showed that analytical precision is approximately 2.5% (pH, sulfate and chloride), 4% (sodium, magnesium and calcium), 5% (ammonium), 10% (nitrate), and 15% (potassium). To obtain some idea on the experimental accuracy it is tempting to compare with chemistry precipitation data from adjacent stations. However, due to differences in construction, location, sampling and analytical procedures results should be interpreted with caution. The comparison is made with a collector located in Ouderkerk maintained by the Provincial Department of Environmental Control in NorthHolland (PW). Daily samples were collected here and ultimately combined to a monthly sample. Precipitation weighted averages of 18 (in the case of sodium and ammonium), and 26 (calcium, sulfate, nitrate) months, respectively, have been compared with results obtained with our collector (FU) in Ouderkerk. The test shows that differences between the PW and the FU measurements divided by the mean were approximately: - 5 % (calcium), - 2 5 % (chloride), + 4 % (ammonium), + 5% (nitrate), and .- 7% (sulfate). So representativity seems to be of satisfactory level. The systematic negative difference found for the sea salt related components may well be explained by a smaller catchment of dry deposition by the PW collector due to the daily cleaning routine.
Composition of bulk precipitation on a coastal island
3023
From Fig. 2 we also deduce that acids and bases of marine origin merely neutralize each other. The difference between the sum of cations (H ÷ excluded) and the 3.1. General considerations For Schiermonnikoog weighted concentrations sum of anions is larger at Ouderkerk. A stronger acidic (XwA,/~mol ~-1) and standard deviationst (SDwA) of tendency with respect to Schiermonnikoog is conthe XWA are summarized in Table 1 (the period firmed by the frequency distribution o f p H values over monitored is May 1972-December 1987). To deter- 112 months (Fig. 3). The use of the pH parameter mine the importance of sources other than the sea, without further notice may be misleading: the lowest non-sea salt contributions to the calcium, magnesium, pH values are found at Ouderkerk but pollution of potassium, chloride and sulfate content were also samples is larger on the island of Schiermonnikoog. Independent studies have revealed that bulk preestimated. The weighted averaging procedure normalcipitation from Schiermonnikoog contained more izes the concentration by the monthly rainfall amount chemical components than expected at a station in this as measured by the Dutch standard rain collector. We compared the XWA for each ion with those remote area (no heavy industry is present) of The calculated for Ouderkerk. We therefore tested the null Netherlands. (1) For the period 1978-1982 the island of Schierhypothesis that there is no difference in the XWA values between both sites. Using a 1 per cent level we monnikoog participated in a precipitation sampling find that the null hypothesis is rejected for CI-, Na +, network of the Dutch Water Supply Institute (van de Mg 2 +, and NH~, and at the 5 per cent level for NO 3 Meent, 1984). The collector was located 2 km westand SO~-. At Ouderkerk only CI-* (p<0.01) is wards from our experimental site. In general, sampling higher. The XWA of Ca 2 ÷* and SO 2-* are of the same stations of this network were located away from urban and industrial source areas. Precipitation weighted magnitude. The calculated non-sea salt contributions are concentrations of 2-weekly samples were calculated. largest for sulfate (76%) and calcium (72%). For From the 27 stations, Schiermonnikoog showed the potassium 27% is found and in the case of chloride highest levels of ammonium and nitrate, whereas and magnesium less than 7% is expected to be from chloride, sodium, potassium, calcium, magnesium and another source. At Ouderkerk non-sea salt parts sulfate were always ranked among the 'top three' of become more important: 91% (sulfate), 90% (calcium), concentration levels. (2) From July 1972 onwards a rain collector was 53% (potassium), 35% (magnesium) and 25% (chlorlocated on the experimental site of the Schiermonniide). At both stations the correction for sea salt koog station maintained by the Royal Netherlands sometimes led to negative monthly concentrations Meteorological Institute. After making a comparison for magnesium, potassium and chloride, mainly durwith the W M O station Witteveen (see Fig. 1), Ridder ing the winter period. With the help of bar diagrams (Fig. 2) we can (1974, pers. comm.) concluded that a stronger envisualize the comparison with Ouderkerk (monitoring richment of the collected rainwater took place on the period for both stations is May 1972-January 1983) island of Schiermonnikoog. Only excess sulfate was and interpret the chemical compositions of wet and approximately of the same level. dry deposition in terms of source type (marine or nonmarine) and charge (cations and anions). Concentra- Table 1. Precipitation weighted averages (Xw^) and stantions are in #eq f - 1 . The figure shows that in the dard deviations (SDwA) of monthly concentrations samples of Schiermonnikoog nearly twice as many (#mol d- 1) of the dissolved substances in bulk precipitation constituents are found, which is almost entirely due to for Schiermonnikoog during May 1972-December 1987. * indicates non-sea salt contributions. Number of months of the sea salt components (CI-, Na ÷, Mg 2 +). While on measurements is 152 Schiermonnikoog the total contribution from the sea (65%) is larger than the other input sources, at XWA SDwA~ SD/X (%) Ouderkerk it accounts for 42% (by molar concentra4.46 0.04 1.0 pH tion). Ca 2+ 27.5 1.2 4.2 Mg2 + 43.0 3.3 7.6 i The standard deviations of volume-weighted averages K + 10.6 0.7 6.3 (SDwA) are calculated according to Na + 354.0 28.2 8.0 149.1 7.6 5.1 I N // N X~l 1/2 NH,~ 435.2 31.2 7.2 90.7 4.4 4.9 72.3 3.9 5.4 SDwA = ~ j , NO3 19.7 0.9 4.7 C a 2+* i Mg 2+* 2.8 0.6 21.7 K +* 2.9 0.4 15.3 where Pi is the precipitation amount corresponding to the ith CI - * 22.1 4.9 22.2 sample, xl is the chemical constituent of interest and N is the SO42- * 69.3 3.2 4.7 number of samples (Galloway et al., 1985). 1 RESULTS
L ( lp, (N-I)
E.P. WEIJERSand H. F. VUGTS
3024
F--~hydronium ~ammonium 500-
potassium ~]
400-
•
magnesium
nitrate
~ ' ~ ] t a l c iu m
[////
sulfate
sodium
FFTT] c h l o r i d e
c
300-
o
._> 0"
200-
o u
\\\
100-
\ \\
~x \\ •
irrrr
,\\
non-marine
marine
marine
Schier monnikoog
non - m a r i n e Ouderkerk
Fig. 2. Marine and non-marine concentrations (in/~eq f - ~) found on the island of Schiermonnikoog and at Ouderkerk. The marine cationic contribution (left bar, from bottom to top) contains sodium, and the sea salt fractions of calcium, magnesium and potassium, whereas the marine anionic part (right bar, from bottom to top) is given by the sea salt parts of chloride and sulfate. The non-marine contributions are built up by the non-sea salt fractions of the above-mentioned components (sodium excluded), ammonium and hydronium (placed on top of the left bar), and nitrate (right bar). 26 24\
22 20
\
\
\
\
\
\
\4
\
\
\
\
\
\A
\A
18 16 14. .o
E
12
// // // .// ,// //
c
10 8 6 4
\ 2 0
I
1
I
I
3.2
3.4"
3.6
3.8
4-
4".2
Ouderkerk
,// ,// // // // // ,/// ,¢f// 1
l
I
I
I
I
I
I
I
4.4
4.8
4..8
5
5.2
5.4
5.6
5.8
6
\/
/
~
Schiermonnikoog
Fig. 3. Frequency distribution of pH values for Schiermonnikoog and Ouderkerk. O u r comparison between the datasets gathered at Schiermonnikoog and Ouderkerk confirms these findings; it rules out the possibility of serious errors in sampling procedures. So despite the fact that this island was selected initially as having the potential to provide data representative of clean conditions there is evidence that this is not the case. It was suspected
that the intensive farming would be the m a m cause as no other polluting activity is found on the island.
3.2. Correlations between the ion species Correlation analysis was done for two purposes. First, to detect possible c o m m o n sources or inorganic
Composition of bulk precipitation on a coastal island compounds and second, to determine which components are responsible for the acidity content. Calculating skewness and kurtosis revealed lognormal distributions for every concentration set. Before calculating Pearson product-moment correlations all parameters were therefore log-transformed in order to attain the required normal distribution. Coefficients and significance levels are shown in Table 2 for Schiermonnikoog and in Table 3 for Ouderkerk. For the non-sea salt fractions of magnesium, potassium and chloride it was not possible to use a logtransformation because of the occurrence of negative values. On the island of Schiermonnikoog the largest correlation coefficients (>0.92) are found for chloride, sodium and magnesium confirming that these components descend almost entirely from the sea salt aerosol. At Ouderkerk the influence of other sources decreases statistical significances. After subtracting the sea salt fractions from calcium and sulfate, relationships with chloride, sodium and magnesium have become lower as expected. However, it is remarkable that on the island of Schiermonnikoog significant correlations (>0.37) still remain between excess sulfate and sea salt components. This seems to be a local effect since at Ouderkerk coefficients are much smaller ( < 0.25). At both sites significant correlations (>0.41) between the non-sea salt fractions of calcium and sulfate are found (the highest at Ouderkerk). This may indicate the presence of non-acidic sulfates. In general, relationships of ammonium and nitrate with the sea salt originating components are weak. On Schiermonnikoog relatively large coefficients are found for ammonium and sulfate (0.65), improving when sulfate is corrected for sea salt (0.69), ammonium and nitrate (0.79), and excess sulfate and nitrate (0.55). At Ouderkerk the correlation coefficient for ammonium and excess sulfate is about the same but relations of ammonium and excess sulfate with nitrate are much weaker (0.37, respectively 0.23) here. Apparently the strong correspondence of ammonium and sulfate with nitrate is locally determined. In the Schiermonnikoog samples weak or insignificant relationships are detected between the presence of H ÷ ions, and the content of ammonium, excess sulfate and nitrate. Statistical evidence (r=0.79) exists for a relation between NH~ and the sum of 2(SO 2-*) and NO3. This is merely a restatement that the ion balance holds. However, adding H ÷ to NH~ and recalculating the correlation coefficient between both sums improves the statistical strength (r=0.87), thereby suggesting some hidden complementary relationship which concerns H ÷. To find evidence for this assumption and also to uncover possibly hidden and spurious relationships we used partial correlation techniques. Partial correlation coefficients measure the relationship between two variables while controlling for the possible effects of other variables. These are controlled by removing the
3025
linear relationship with the other variables before calculating the correlation coefficients between the two variables of interest. For example, we find that the correspondence between excess sulfate and nitrate disappears (0.55 ~ 0.03) when the relationship with ammonium is removed, indicating that the apparent correlation between nitrate and sulfate is due to the latter's positive correlation with ammonium. The result is also more in line with the Ouderkerk findings. It appears that correlation coefficients for hydrogen ions with excess sulfate and nitrate, grow (0.15 ~ 0.30, resp. 0.22~0.48) when the effect of ammonium is controlled for. When one of the strong acids is held constant, ammonium correlates more negatively with hydrogen ions ( - 0.09 ~ 0.25, resp. ~ - 0.40). After removing the possible effects of all the remaining major ions we obtain correlation coefficients as given in Table 4. Statistical strength has been increased considerably compared to Tables 2 and 3. Coefficients are of the correct sign as they should if excess sulfate, nitrate and ammonium were originally present as SO 2, NOx and NH 3. At Ouderkerk the strongest correlation with H ÷ is found for nitrate. Summarizing, the following mechanisms in the Schiermonnikoog samples are suggested by the correlation analysis: (1) the sea salt-related components form an interconnected group; (2) an interaction exists between the ammonium and the sulfate ions (also found at Ouderkerk); (3) an interaction exists between the ammonium and the nitrate ions (not apparent in the Ouderkerk data set); (4) an interaction between sea salt and sulfate exists that remains after correcting sulfate for the sea salt contribution (not measured at Ouderkerk); (5) most of the variation in the hydrogen ion concentrations is due to variations in nitrate, excess sulfate and ammonium. 3.3. Annual cycles and meteorological influences Precipitation weighted monthly concentrations were calculated in order to obtain annual cycles at both sites. The normalized concentrations (expressed as a percentage of the Xw^ in Table 1) are shown in Fig. 4. The ammonium content in the bulk samples is high in the first half of the year. The month with the lowest concentration is October. On Schiermonnikoog a pronounced maximum is found in February. At Ouderkerk the variability is larger which may be due to the smaller record length. Higher excess sulfate levels occur during the first 4 months of the year while the lowest values are found in July, August and September. Its pattern is very similar to ammonium but contrasts with that of nitrate: maximum concentrations in the early summer period and minimum values in December. No annual trend was evident in the hydronium concentrations but at both stations
3026
E.P. WEIJERSand H. F. VUOTS
Table 2. Correlation coeffÉcients (significance levels) between ion species in bulk samples collected on the island of Schiermonnikoog. Number of months of measurements is 152
H Ca Mg K Na NH4 CI SO4 NO 3 Ca* SO*
H 1.000 (0.000) 0.073 (0.382) -0.059 (0.484) 0.000 (0.998) -0.042 (0.616) -0.092 (0.269) -0.039 (0.643) 0.125 (0.132) 0.218 (0.008) 0.090 (0.286) 0.145 (0.082)
Ca 0.073 (0.382) 1.000 (0.000) 0.466 (0.000) 0.493 (0.000) 0.412 (0.000) 0.309 (0.000) 0.447 (0.000) 0.609 (0.000) 0.390 (0.000) 0.920 (0.000) 0.569 (o.ooo)
Mg -0.057 (0.484) 0.466 (0.000) 1.000 (0.000) 0.637 (0.000) 0.941 (0.000) 0.292 (0.000) 0.927 (0.000) 0.647 (0.000) 0.224 (0.006) 0.152 (0.063) 0.437 (o.ooo)
K 0.000 (0.998) 0.493 (0.000) 0.637 (0.000) 1.000 (0.000) 0.593 (0.000) 0.341 (0.000) 0.618 (0.000) 0.548 (0.000) 0.313 (0.000) 0.286 (0.000) 0.422 (o.ooo)
Na -0.042 (0.616) 0.412 (0.000) 0.941 (0.000) 0.593 (0.000) 1.000 (0.000) 0.250 (0.002) 0.974 (0.000) 0.608 (0.000) 0.172 (0.034) 0.074 (0.365) 0.374 (o.ooo)
NH, -0.092 (0.268) 0.309 (0.000) 0.292 (0.000) 0.341 (0.000) 0.250 (0.002) 1.000 (0.000) 0.269 (0.000) 0.647 (0.000) 0.790 (0.000) 0.264 (0.001) 0.685 (o.ooo)
CI -0.039 (0.643) 0.447 (0.000) 0.927 (0.000) 0.618 (0.000) 0.974 (0.000) 0.269 (0.001) 1.000 (0.000) 0.625 (0.000) 0.175 (0.031) 0.115 (0.161) 0.398 (o.ooo)
SO4 0.125 (0.132) 0.609 (0.000) 0.647 (0.000) 0.548 (0.000) 0.608 (0.000) 0.647 (0.000) 0.625 (0.000) 1.000 (0.000) 0.510 (0.000) 0.437 (0.000) 0.956 (o.ooo)
NOa 0.218 (0.008) 0.390 (0.000) 0.224 (0.006) 0.313 (0.000) 0.172 (0.034) 0.790 (0.000) 0.175 (0.031) 0.510 (0.000) 1.000 (0.000) 0.382 (0.000) 0.554 (o.ooo)
Ca* 0.089 (0.286) 0.920 (0.000) 0.152 (0.063) 0.286 (0.000) 0.074 (0.365) 0.264 (0.001) 0.115 (0.161) 0.437 (0.000) 0.382 (0.000) 1.000 (0.000) 0.498 (o.ooo)
SO~ 0.145 (0.082) 0.569 (0.000) 0.437) (0.000) 0.422 (0.000) 0.374 (0.000) 0.685 (0.000) 0.398 (0.000) 0.956 (0.000) 0.554 (0.000) 0.498 (0.000) 1.000 (o.ooo)
Table 3. Correlation coefficients (significance levels) between ion species in bulk samples collected at Ouderkerk. Number of months of measurements is 112
H Ca Mg K Na NHa CI SO4 NO s Ca* SO*
H Ca 1.000 -0.071 (0.000) (0.463) -0.071 1.000 (0.463) (0.000) 0.148 0.278 (0.138) (0.004) -0.116 0.329 (0.239) (0.001) 0.116 -0.060 (0.236) (0.532) -0.115 0.416 (0.237) (0.000) 0.161 0.213 (0.096) (0.025) 0.113 0.406 (0.242) (0.000) 0.480 0.405 (0.000) (0.000) -0.076 0.984 (0.433) (0.000) 0.093 0.412 (0.338) (0.000)
Mg 0.148 (0.138) 0.278 (0.004) 1.000 (0.000) 0.380 (0.001) 0.428 (0.000) 0.182 (0.062) 0.609 (0.000) 0.356 (0.000) 0.217 (0.026) 0.181 (0.065) 0.253 (0.009)
K -0.116 (0.239) 0.329 (0.001) 0.380 (0.001) 1.000 (0.000) 0.360 (0.000) 0.322 (0.001) 0.381 (0.000) 0.273 (0.004) 0.397 (0.000) 0.243 (0.012) 0.195 (0.044)
Na 0.116 (0.236) -0.060 (0.532) 0.428 (0.000) 0.360 (0.001) 1.000 (0.000) 0.002 (0.982) 0.816 (0.000) 0.142 (0.137) -0.017 (0.857) -0.198 (0.038) -0.018 (0.854)
maxima occurred in June coinciding with the time that nitrate shows its maximum. The a n n u a l pattern of chloride is very similar to those of sodium and magnesium. Cycles show peak values during the winter period (November, December and January) with a relative maximum in April.
NH,, -0.115 (0.237) 0.416 (0.000) 0.182 (0.062) 0.322 (0.001) 0.002 (0.982) 1.000 (0.000) 0.012 (0.901) 0.727 (0.000) 0.368 (0.000) 0.418 (0.000) 0.738 (0.000)
CI 0.161 (0.096) 0.213 (0.025) 0.609 (0.000) 0.381 (0.000) 0.816 (0.000) 0.012 (0.901) 1.000 (0.000) 0.241 (0.011) 0.023 (0.814) 0.067 (0.488) 0.078 (0.416)
SO4 0.113 (0.242) 0.406 (0.000) 0.356 (0.000) 0.273 (0.004) 0.142 (0.137) 0.727 (0,000) 0.241 (0.011) 1.000 (0.000) 0.228 (0.016) 0.378 (0.000) 0.980 (0.000)
NO 3
0.480 (0.000) 0.405 (0.000) 0.217 (0.026) 0.397 (0.000) -0.017 (0.857) 0.368 (0.000) 0.023 (0.814) 0.228 (0.016) 1.000 (0.000) 0.395 (0.000) 0.231 (0.014)
Ca* -0.076 (0.433) 0.984 (0.000) 0.181 (0.065) 0.243 (0.012) -0.198 (0.038) 0.418 (0.000) 0.067 (0.488) 0.378 (0.000) 0.395 (0.000) 1.000 (0.000) 0.415 (0.000)
SO* 0.093 (0.338) 0.412 (0.000) 0.253 (0.009) 0.195 (0.044) -0.018 (0.854) 0.738 (0.000) 0.078 (0.416) 0.980 (0.000) 0.231 (0.014) 0.415 (0.000) 1.000 (0.000)
The pattern of potassium, though being slightly different, still shows a typical sea salt characteristic (maximum in January). F o r calcium, the maximum value found in April coincides with the characteristic extreme observed in the sea salt patterns. However, in winter concentrations remain low.
F
M
A
M
J
J
A
S
0
N
D
F
M
A
M J
J
A
S
0
N
D
J
A
S
0
J L~6~6
J
N
D
D
,~
o
'2°
20~
J
J
F
F
M A
M A
M
M
J
J
J
J
A
A
S
S
0
0
D
2Ot
N D
120-1
1401
16oq
~8ot
F
M
A
M
J
J
A
S
0
N
D F
M
J
A
S
O N
60-I
6O.
J
~q
80
80
M A
M J
J
A
S
0
N
D
S c h i e r monnikoo(3
Schiermonnikoog
Fig. 4. P r e c i p i t a t i o n a v e r a g e d m o n t h l y c o n c e n t r a t i o n s e x p r e s s e d a s p e r c e n t a g e s o f X w ^ .
Ouderkerk
204 F
204
20" 201
60-1 40-1
40t
e01
too-1
40.
J
140t 12o-1
6O-1
I
i00
I00
,41 12
140 120
160-1
100t
N itrote
2O-1
40-1
2o-1
4oq
120
o =
J40
160
160
J
Excess coLclum
20t
180
40q
2O
~
60t
40-
~
60-1
6o-1
j
J
F
M
F M
E x c e s s suLfote
N
60-1 40t
80-1
M J
M
$2O I00 ~q
80-
A
M A
3;M~Mj
F
40t
604
801
I00t
i201
160
~00-1 80t
J
160 140
Ammonium
100-
120"1
o
140t
~80t
Potassium
140-
180-
J
4O Z0t
40~
6O
2ol
SOq
6O~
oo
I00
~ 120-
W(
J
120
120~
,B1°
160 140
140-
ChLoride
160-
180-
A
J
J
J
J
A
A
Ouderkerk
M
A M
S
S
0
0
D
N 0
N
E
=__
8==
0
O
a
=..
o"
o
¢3 o
3028
E.P. WEIJERSand H. F. VUGTS
Table 4. Partial correlations for hydronium with the other ion species at the stations Schiermonnikoog and Ouderkerk
NH 4 SO* NO 3 Ca* Mg K Na Cl
Schiermonnikoog
Ouderkerk
H -0.573 0.422 0.540 - 0.211 -0.168 -0.013 - 0.046 0.076
H -0.444 0.434 0.757 - 0.377 -0.059 -0.463 - 0.063 0.246
Though the XwA of the non-sea salt fractions are relatively small (Table 1), annual periodicity was apparent with higher values occurring in the summer period. The reason for the negative concentrations occasionally found in winter is unknown. Systematically overestimating the presence of sodium due to a non-sea salt source or an error in the measurement method affects the estimated non-sea salt contribution in a negative direction. However, this does not explain the observed annual variability. Annual periodicities with peak values in the first half of the year seem characteristic for most components in the European area (e.g. Moldan et a1.,1987). In general, our results reasonably agree with other reports, e.g. after analyzing EACN network data Soderlund and Granat (1982) found a maximum for ammonium in the early spring (March and April) and a minimum in August and September which is confirmed by a study from Buishand et al. (1988) for the Dutch area. The spring peak in sulfate concentrations is in agreement with the annual cycle (maximum in February-May, minimum in July-October) found by Rohde and Granat (1984). Skeffington (1984) found a late spring peak for nitrate (May-June, averaged over 4 years). In England Brimblecombe and Pitman (1980) found winter maxima and summer minima for chloride. To study the influence of monthly precipitation amount on concentration, correlation coefficients between the log-transforms of these variables were calculated. For every component a negative correlation was found. Being significant at the 1% level we found the following order: Ca 2 + * < SO 2- * < N O 3 < N H 2 . For excess calcium this is probably a consequence of the prevailing terrigenous nature of this element. The strength of its source will be strongly influenced by a seasonal pattern in the appearance of dry periods. The lowest rainfall amounts are measured in April which obviously affects the presence of sea salt material in the sample. Nevertheless the lowest correlation with precipitation amount is detected for these constituents. Levels of these components are enhanced when monthly rainfall amount rises ( > 70 mm) and when averaged monthly wind speed during rainfall increases ( > 6 m s - t ) . This combination often takes place in
winter when seawater temperature is higher than air and a tendency for fairly strong vertical mixing with higher wind speed just above the water surface exists. For excess sulfate and ammonium hardly any effect of the wind speed could be detected. Two counterproductive mechanisms are expected to be at work here. Ammonia emitted by a nearby source will remain longer in the surrounding air when wind speeds are low. On the other hand, a larger upward flux of ammonia will occur with increasing wind speeds. The third meteorological variable of interest is the direction of the wind. Monthly frequency distributions reveal that (south) westerly winds dominate most of the year. Wind roses of March and April do not show any dominancy whereas in February and May continental air masses cross the island relatively often. Contaminating influences from the mainland combined with the low precipitation amounts observed during these months may account for the higher concentrations. 3.4. Lon#-term behavior Figure 5 illustrates annual precipitation weighted concentrations on both stations for the most acidifying components and sodium, as representative oftbe sea salt species. Influence of rainfall amount is evident in the year 1976 which was extremely dry (530mm, normal 850 mm): for most components (except hydronium) the maximum annual concentration was measured. In 1980 the highest H + values were found at both stations, which coincides with relatively low ammonium levels. The annual variation of ammonium is by far the largest. On Schiermonnikoog weaker but corresponding variations in the ion concentrations of nitrate and sulfate can be seen. In the case of excess sulfate some similarity with the Ouderkerk pattern is deduced. Apparently effects from the mainland are also measured on the island. When 1976 and 1977, two drought years, are excluded, the levels of sodium (and the other sea salt) species barely changed over the monitoring period. A linear regression model has been used for a trend analysis on the monthly concentration data of the five components. Regression coefficients (RC) and significance levels (SL) are given in Table 5 for both Schiermonnikoog and Ouderkerk (period I: May 1972-January 1983) and for Schierrnonnikoog (period II: January 1983-December 1987). As linear regression is sensitive to the occurrence of outliers, the calculation was repeated on the logarithm of concentration values to see if drastic changes in significances (SLIog) took place. Significant levels (i.e. SL as well as SLIog are less than 1%) indicate increasing hydronium and nitrate concentrations in the Ouderkerk data record. The RC of ammonium and excess sulfate do not differ significantly from zero. The sea salt species show small
Composition of bulk precipitation on a coastal island 800
3029
sodium
600 400
•
_
• •
•
•
•
•
•
•
•
•
•
200-[]
[]
[]
[]
[]
[]
[]
[]
[]
0-ammonium
300 -•
200 -•
100
-
-
•
•
•
[] •
• []
• []
[]
[]
•
[]
[]
[]
• []
•
•
II
•
..
m
n
•
[]
0-E o E
nitrate
200 -•
100 f n ~
•
• []
[]
•
[]
0 excess sulfate 200 I IO0 i
)
~
•
•
•
[] m
[]
[] mm []
•
i
m
m
m
0 200 1
hydronium
. 73
74 •
75
76
77
78
Schiermonnikoog
79
80
811 []
82
83
84
85
86
87
Ouderke~
Fig. 5. Annualweighted averages of H +, NH~, NO~, SO~- *, and Na ÷ for the island of Schiermonnikoog and Ouderkerk. increases that are significant at the 10% level in the case of concentrations and at a level of less than 5% when the logarithmic value is taken. Apparently some effect of outliers has been measured here. Contrary to Ouderkerk, statistical evidence exists for relatively large increasing positive trends in the ammonium and sulfate concentrations on Schiermonnikoog. This might reflect the influence of the strongly intensifying cattle breeding culture during the observation period.
4. DISCUSSION AND C O N C L U S I O N S
The role of ammonia in atmospheric chemistry for the European area has recently drawn more attention (van Breemen et al., 1982; Moiler and Schieferdecker, 1985; Asman et al., 1987; ApSimon et al., 1987; Buijsman et al., 1987). The latter reference contains an enumeration of sources of atmospheric NH 3.
Animal waste and fertilizers are held responsible for 90% or more of the anthropogenic NH 3 emission. Some information on the emission density in The Netherlands is given in the inventory of Erisman (1989). Farmers on Schiermonnikoog use artificial fertilizers but under the influence of low pH values of the sandy soils the loss of ammonia from these fertilizers is expected to be relatively small. Adding animal manure (often sprayed on the land in a very dilute form) and artificial fertilizer to the agricultural grounds on Schiermonnikoog is also usage. The manure is known to have high pH values which favors the loss of NH 3. Close to sources, high concentrations ofNH 3 in air at ground level will occur because the plumes of these sources have not been mixed over the whole mixing layer yet (Asman and Janssen, 1987). This will lead to a relatively high dry deposition in the bulk collectors which are always open, and will cause the high NH,~ levels in our samples (not to be misinterpreted as a high contribu-
3030
E.P. WEIJERSand H. F. VUGTS
Table 5. Regression coefficients(RC) for concentrations (in /zmolmonth-t) and significance levels for both concentrations (SL) and the log-transformed concentrations (SLIos) of some major ions. The H ÷ concentration is calculated from the pH value. Periods of study: May 1972-January 1983 (I), and January 1983-December 1987 (II) Period I
RC
Ouderkerk SL
SLjog
H÷ NH2 SO2- * NO~Na ÷
0.545 0.316 0.066 0.374 0.476
-0.000 0.175 0.632 0.006 0.092
0.000 0.518 0.632 0.000 0.021
Period I
RC
Schiermonnikoog SL
SZlog
H÷ NH2 SO,2- * NO~Na ÷
0.372 1.001 0.482 0.617 0.810
0.005 0.013 0.016 0.001 0.307
0.041 0.032 0.004 0.000 0.216
Period II H* NH2 SO42-* NO~ Na ÷
RC 0.072 3.534 0.627 0.971 1.487
Schiermonnikoog SL SLIo8 0.745 0.000 0.076 0.010 0.415
0.602 0.000 0.078 0.030 0.252
tion of the local source to the NH~ concentration in precipitation). The simultaneous occurrence of certain economical and meteorological circumstances in the late winter months may explain the maximum concentrations of ammonium measured in the samples during this period. (1) Dumping of the surplus of stable manure on agricultural land often happens in this period when storage capacity is not large enough. The lack of storage volume has become more acute as numbers of livestock grew. Transport to the mainland for storage or use is not done for economical reasons. (2) Often the manure cannot be ploughed under on frozen and unworkable soils in winter. This will result in a net flux upwards of NH 3. The release of animal manure on snow-covered ground has recently been forbidden (winter 1989) by Dutch legislation to prevent the pollution of waterways by runoff as soon as thaw occurs. (3) Due to temperature differences between seawater and island soil occurring in the late winter period, relatively warm air caps the colder air above the island which results in low mixing heights and increasing concentrations just above the island. The emission of ammonia seems to influence the presence of sulfate and nitrate in the bulk samples. Correlation coefficients and annual patterns suggest that most of the ammonium was present as ammonium sulfate, which is probably formed by the inter-
action of ammonia with sulfur dioxide on the wetted funnel. A natural SO 2 source might be the inter-tidal fiats between the island and the mainland (about 1000 km2). Traditionally it has been assumed that the biological decomposition of sulfur-containingorganic matter results in emissions of hydrogen sulfide (H2S) but after the discovery of dimethylsulfide ((CH3)252) produced by algae in the surface seawater it is believed that this compound is the major natural component of sulfur emission in the atmosphere (Jorgensen, 1982). Some evidence for its presence might be derived from the observed correspondence between sea salts and excess sulfate which was not measured at the Ouderkerk station. A possible effect of such a local source will mainly be measured as dry deposition and not lead to higher SO,2- levels in collected rainwater as by the time the gases are oxidized, it will be far away from the island. In view of the difference observed in the annual cycles of ammonium and nitrate the presence of ammonium nitrate seems less obvious. The annual cycles reasonably agree with other reports. One ascribes the maximum values to larger levels of oxidants and to higher temperatures during spring and summer contributing to a faster oxidation of nitrate to acids. The additional nitrate amount may be due to a volatilization of nitrogen matter from the manure. A second cause may be some artefact in the sampling routine, i.e. a temperature-related nitrification of ammonium by bacteria which becomes more apparent with a field-residence time of a month. The reason why this effect is not observed in the Ouderkerk samples is probably the more acidic regime of the samples and the absence of nearby ammonia sources. This also explains that although the abundantly present ammonia is a base it does not cause a significant increase in the pH in the Schiermonnikoog samples with respect of Ouderkerk as the oxidation of ammonium to nitrate may have contributed some acid through: NH~ + 2 0 2 ~ N O 3 + H 2 0 + 2 H +.
Acknowledgements--We gratefully acknowledge Dr Ed Buijsman's critical comment on the first draft of this paper. Our co-worker, Dr Fred Cannemeijer, offered valuable comments regarding statistical evaluation. Many thanks to Tiny Baer and the laboratory staff for performing the analyses over such a long time. We are also grateful to the Provinciale Waterstaat van Noord Holland, Dienst voor de Milieuhygiene for providing us with their chemical precipitation data gathered at Ouderkerk. REFERENCES
ApSimon H. M., Kruse M. and Bell J. N. B. (1987) Ammonia emissions and their role in acid deposition. Atmospheric Environment 21, 1929-1946. Asman W. A. H. and Janssen A. J. (1987) A long-range transport model for ammonia and ammonium for Europe. Atmospheric Environment 21, 2099-2199. Asman W. A. H., Ridder T. B., Reijnders H. F. R. and Slanina J. (1982) Influence and prevention of bird-droppings in
Composition of bulk precipitation on a coastal island precipitation chemistry experiments. Wat. Air Soil Pollut. 17, 415-420. Breemen N. van, Burrough P. A., Velthorst E. J., van Dobben H. F., de Wit T., Ridder T. B. and Reijnders H. F. R. (1982) Soil acidification from atmospheric ammonium sulphate in forest canopy throughfall. Nature 229, 548-550. Brimblecombe P. and Pitman J. (1980) Long-term deposit at Rothamsted, southern England. Tellus 32, 261-267. Buijsman E., Maas H. F. M. and Asman W. A. H. (1987) Anthropogenic NH 3 emissions in Europe. Atmospheric Environment 21, 1009-1022. Buishand T. A., Kempen G. T., Reijnders H. F. R., Frantzen A. J. and van den Eshof A. J. (1988) Trend and seasonal variation of precipitation chemistry data in The Netherlands. Atmospheric Environment 22, 349-358. Erisman J.-W. (1989) Ammonia emissions in The Netherlands in 1987 and 1988, Report 228471006. National Institute of Public Health and Environmental Protection, Bilthoven, The Netherlands. Galloway J. N., Likens G. E. and Hawley M. (1985) Acid precipitation: natural versus anthropogenic components. Science 226, 829. Jorgensen B. B. (1982) The production and fate of reduced C, N, and S gases. In Atmospheric Chemistry (edited by E. Goldberg), pp. 225. Dahlem Konferenzen, SpringerVerlag, Berlin.
AE(A)
21:12-I
3031
van de Meent D., van Oosterwijk J. and Aldenberg T. (1984) Rijksinstituut voor de drinkwater voorziening, RIDVEWlN Meetnet regenwater 1978-1982. Samenvatting en statistische bewerking van de meetresultaten, ECOWAD8401 RIVM (in Dutch). Moldan B., Vesely M. and Bartonova A. (1987) Chemical composition of atmospheric precipitation in Czechoslovakia, 1976-1984--1. Monthly samples. Atmospheric Environment 21, 2383-2395. Moiler D. and Schieferdecker H. (1985) A relationship between agricultural NH 3 emissions and the atmospheric SO 2 content over industrial areas. Atmospheric Environment 19, 48-55. Rohde H. and Granat L. (1984) An evaluation of sulfate in European precipitation 1955-1982. Atmospheric Environment 18, 2627-2639. Sketiington R. A. (1984)The chemistry of bulk precipitation at a site in southeast England--lI. Relationships between ions and comparison with other sites. Atmospheric Environment 18, 1695-1704. Soderlund R. and Granat L. (1982) Ammonium (NH2) in precipitation--a presentation of data from the European Air Chemistry Network. Report CM-59. Department of Meteorology, University of Stockholm/International Meteorological Institute in Stockholm.