Environ. Sci. Technol. 2004, 38, 26-33
The Significance of the North Atlantic Oscillation (NAO) for Sea-Salt Episodes and Acidification-Related Effects in Norwegian Rivers A T L E H I N D A R , * ,† K J E T I L T Ø R S E T H , ‡ ARNE HENRIKSEN,§ AND YVAN ORSOLINI‡ Norwegian Institute for Water Research, Televeien 3, N-4879 Grimstad, Norway, Norwegian Institute for Air Research, P.O. Box 100, N-2027 Kjeller, Norway, and Norwegian Institute for Water Research, P.O. Box 173, N-0411 Oslo, Norway
Acidification of Norwegian surface waters, as indicated by elevated concentrations of sulfate and a corresponding reduction in acid neutralizing capacity and pH, is a result of emission and subsequent deposition of sulfur and nitrogen compounds. Episodic sea-salt deposition during severe weather conditions may increase the effects of acidification by mobilizing more toxic aluminum during such episodes. Changes in climatic conditions may increase the frequency and strength of storms along the coast thus interacting with acidification effects on chemistry and biota. We found that the North Atlantic Oscillation (NAO) is linked to sea-salt deposition and sea-salt induced water chemistry effects in five rivers. Particularly, toxic levels of aluminum in all rivers were significantly correlated with higher NAO index values. Further, temporal trends were studied by comparing tendencies for selected statistical indices (i.e. frequency distributions) with time. The selected indices exhibited strong correlations between the NAO index, sea-salt deposition and river data such as chloride, pH and inorganic monomeric aluminum, pointing at the influence of North Atlantic climate variability on water chemistry and water toxicity. The potentially toxic effects of sea-salt deposition in rivers seem to be reduced as the acidification is reduced. This suggests that sea-salt episodes have to increase in strength in order to give the same potential negative biological effects in the future, if acid deposition is further reduced. More extreme winter precipitation events have been predicted in the northwest of Europe as a result of climate change. If this change will be associated with more severe sea-salt episodes is yet unknown.
Introduction The causes and effects of acidification have been identified by research over the last few decades, but influences of * Corresponding author phone: +47 37 29 50 55; fax: +47 37 04 45 13; e-mail:
[email protected]. † Norwegian Institute for Water Research, Grimstad. ‡ Norwegian Institute for Air Research. § Norwegian Institute for Water Research, Oslo. 26
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 1, 2004
confounding factors are less well understood. Main uncertainties are connected to climatic variability and effects of climatic changes. One phenomenon that has been documented is the acidifying effects of sea-salt episodes (1-3) and the dramatic effects they might have on e.g. fish survival (4, 5). The sea-salt effect is ascribed to cation exchange processes in the soil, mainly involving exchange of Na+ with H+ and Al-ions and/or base cations. Cl- acts as a mobile anion (2, 6, 7) and passes through the soil relatively unaltered. The salt effect reduces the acidity of the soil temporarily due to Na+ retention but increases the acidity of the runoff (1). Empirical evidence for the sea-salt effect can be seen from concentrations of Na+ relative to Cl-. When the equivalent ratio of Na/Cl in runoff water is less than that of seawater, it can be assumed that ion exchange in the soil has occurred. Sea-salt episodes do not always reduce water quality. In acidified areas in Norway unusually high amounts of H+ and Al-ions were mobilized during a severe sea-salt episode in 1993 to compensate for the amount of Na+ that was adsorbed by ion exchange in the soil (4). In less acidified areas, adsorbed Na+ was compensated largely by mobilization of base cations or a combination of base cations, H+, and inorganic Al. A prerequisite for a sea-salt episode to be toxic is acid soils in combination with high sea-salt deposition in relation to normal deposition (3). A period of large amounts of dry fall out followed by rain may also result in reduced water quality (3). Knowledge of and experience with sea-salt episodes is largely from experimental studies and single episodes (2-4). Recently, impacts of deposition of marine ions for temporal trend determination have been analyzed for monitoring sites in the UK (8). Sea-salt episodes appear to occur in wintertime during extreme weather conditions (heavy storms). To better understand the causes and effects of sea-salt episodes, we thought it would be of great importance to assess links to seasonal variations and temporal trends in both acidification and climatic variability. The North Atlantic atmospheric circulation pattern or the North Atlantic Oscillation (NAO) is the main mode of atmospheric variability in the Euro-Atlantic sector during winter and spring. It relates to the strength of the jet stream across the Atlantic and largely governs the interannual variability of precipitation over Northern Europe. Its intensity is often expressed as an index (NAOI) calculated from seasonal-mean or monthly mean sea level pressure difference between Iceland and the Azores or Gibraltar (9, 10). The NAO, however, displays intramonthly variability. High positive values of NAOI during winter months, that is higher than usual pressure difference between the Iceland Low and Azores High, are strongly related to the climatic conditions in Europe (10, 11). Such pressure differences generally cause strong westerly winds, frontal precipitation and relatively high temperatures in the UK and Scandinavia. A relationship between NAOI and measured Cl- concentrations has been found for one of several monitored streams (Afon Hafren in mid-Wales) in the UK based on a period of 15 years (8). In this paper, water chemistry data for five rivers in western Norway, sampled since 1980-1984, are linked to observations of precipitation chemistry and variations in the NAOI. The main goal was to examine to which extent the effects of seasalt episodes can be related to changes in acidification and to climatic variability and thus see whether a climate change, giving more extreme weather conditions, may influence 10.1021/es030065c CCC: $27.50
2004 American Chemical Society Published on Web 11/20/2003
TABLE 1. Location, Catchment Size and Specific Discharge for the Rivers
river Vikedal Sæta (tributary to Gaular) Nausta Trodøla (tributary to Nausta) Øyensåa
location of river mouth
distance from coastline, km
catchment area, km2
specific discharge, L s-1 km-2
59°31′ N 5°58′ E 61°20′ N 6°09′ E 61°34′ N 5°49′ E 61°34′ N 5°56′ E 64°15′ N 11°10′ E
40-55
119
86.6
70-90
181
77.2
45-75
274
79.7
60
10
79.7
35-45
263
FIGURE 1. Map showing locations for the five rivers and sulfur wet-deposition contours for the year 2001. Deposition is given as g S m-2. acidification conditions in Norwegian watercourses in the future. Deposition. Precipitation chemistry has been continuously monitored at a number of rural sites, from 1971 to 1980 as part of the SNSF (Acid Precipitation - Effects on Forest and Fish) project (12) and then as part of the Norwegian Monitoring Program for Long-Range Transported Air Pollutants (13). A selection of the sites is operated with daily sampling for precipitation chemistry thus comprising a unique data set that allows for linking of observation data with the synoptic advection patterns. Samples are taken using bulk-samplers, and the 24-hour temporal resolution also serves to reduce the influence of dry deposition to the precipitation samplers. The precipitation samples were analyzed by standardized methods at the Norwegian Institute for Air Research (NILU). Rivers. The Norwegian Monitoring Program for LongRange Transported Air Pollutants (13) includes data from monitoring of 16 rivers on a regular basis. Three main rivers and two tributaries were selected for this study (Table 1, Figure 1). They are located close to the coast, thus representing typical rivers for Atlantic salmon production. In areas close to the coast, precipitation is strongly influenced by sea-salts, whereas areas more distant from the shore are gradually less influenced. Criteria for selecting rivers were regular sampling for at least 15 years and that they would represent a gradient in
45
acid deposition (see Figure 1). No river liming or other significant human activities should take place in their catchments. The rivers were normally sampled on a monthly basis, with more frequent sampling (biweekly) during the snowmelt period in April-May. River Trodøla is a tributary to River Nausta, but it was selected due to a long record of weekly sampling. Samples were collected on prewashed plastic bottles and mailed to the Norwegian Institute for Water Research (NIVA) in Oslo. The samples normally arrive at NIVA within 1-3 days from sampling. The river samples were analyzed by standardized methods. The nonmarine part of sodium (Na*) is calculated on the basis of the concentrations of Na+ and Cl- and the relation between Na+ and Cl- in seawater. Inorganic monomeric aluminum, operationally defined as labile Al (LAl) because it is retained in the ion-exchange column, is calculated as the difference between reactive Al and nonlabile Al. Characteristic water chemistry data for the five rivers are presented in Table 2. North Atlantic Oscillation Index. Although seasonalmean or monthly mean NAO indexes are commonly used, we used standardized NAO daily indexes, based on the NCEP (National Center for Environmental Prediction) meteorological reanalyzes (14). The daily indexes could then be averaged over weekly to seasonal time scales. Data Analysis. The chemistry data for the rivers have been compared with the weather situations as reflected in the content of Na+ and Cl- in deposition for the whole monitoring period, starting in the early 1980s and up to 2001. Data from several stations included in the monitoring network for deposition have been used. As the main indicator for seasalt deposition we have used Mg2+ deposition, for which sea-spray is the only significant source in the coastal areas of Norway. In rivers, concentrations of Cl- and nonmarine sodium (Na*) are used as sea-salt indicators. Cl- from seaspray passes relatively unaltered through the soil, whereas Na+ is readily adsorbed by ion-exchange when the concentration and amount in deposition is high. Due to insignificant soil sources for Na+, deviation of the Na/Cl-relationship in the rivers relative to that in the sea strongly indicates a seasalt episode that might have negative effects. Further, river Na+ and Cl- concentrations together with the two fish toxicity indicators pH and LAl concentration have been related to the NAOI on a daily, a weekly and a monthly basis. This was done by using NAOI either from the same date as the water chemistry sampling (daily; NAOID), from the preceding 7 days (weekly; NAOIW) or from the preceding 30 days (monthly; NAOIM), respectively. Temporal trends were studied by analyzing changes in the frequency distributions for the various variables. This approach provides a picture of the interdependency of these variables. Frequency distributions also allow for comparing data between years for which the temporal resolution is different. As sea-salt episodes, and in particular those causing VOL. 38, NO. 1, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
27
TABLE 2. Mean Water Chemistry Data for the Rivers for Their Respective Monitoring Periods rivers parameter
unit
Vikedal
Sæta
Nausta
Trodøla
Øyensa˚a
pH Ca2+ Mg2+ Na+ K+ SO42ClNO3alkalinitya ANCb labile-Al nonlabile Al reactive Al total organic carbon total nitrogen nonmarine base cations nonmarine sulfate
µequiv L-1 µequiv L-1 µequiv L-1 µequiv L-1 µequiv L-1 µequiv L-1 µequiv L-1 µequiv L-1 µequiv L-1 µmol L-1 µmol L-1 µmol L-1 µmol L-1 µmol L-1 µequiv L-1 µequiv L-1
5.57 35.4 29.6 95 5.1 43.7 110 10.9 3.9 -0.2 0.8 0.7 1.5 88 17.5 39.1 32.6
5.60 22.0 14.8 51 5.6 25.0 59 5.9 4.5 4.9 0.7 0.9 1.6 102 12.3 23.7 19.0
5.79 27.4 20.6 67 7.2 27.1 79 4.8 8.7 14.2 0.3 0.9 1.2 134 11.3 30.4 18.4
5.58 18.5 20.6 72 5.6 25.0 82 4.5 4.2 6.1 0.4 0.9 1.3 112 9.4 20.2 16.5
6.05 51.4 56.8 219 7.2 37.5 251 1.4 26.4 43.7 0.2 1.6 1.8 364 11.2 49.7 11.7
a
Equivalence point alkalinity.
b
ANC ) (Ca + Mg + Na + K) - (SO4 + NO3 + Cl).
FIGURE 2. Relative deviation from average Mg2+ deposition (bars) and relative deviation from average winter NAOI (open circles) for the years 1973-2001 for four deposition stations in Norway; Birkenes, Lista, and Vatnedalen in south and Tustervatn in the middle of Norway. adverse effects in river water chemistry, are associated with more extreme events, we have focused on the higher percentiles of the distribution.
Results and Discussion Deposition and River Trends. The amount of sea-salts deposited atmospherically in catchments depends strongly on episodes with strong winds from the west and southwest. Figure 2 shows how the sea-salt contribution, as illustrated by deviation from the annual mean Mg2+ deposition, has varied since 1973 at four deposition stations located from south to north in Norway. Relatively high positive deviations from the mean were evident for the period 1989-1993 and the year 2000, indicating unusually high sea-salt deposition. A significant decrease in atmospheric sulfur deposition has been documented for Norwegian monitoring stations (15). The concentration of S in deposition has decreased by more than 60% in western Norway during the period 19802001 and as much as 71-73% in the Trøndelag area further north. 28
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 1, 2004
The nonmarine SO42- concentrations (SO4*) in the five rivers decreased significantly over the period 1986-2001 (Figure 3). Reductions of about 40% in the southernmost rivers and 50-60% in River Øyensa˚a correspond well with the decrease in S deposition. Data from River Nausta from 1980 and also the more discontinuous data from the other rivers until 1984 or 1986 indicate that the concentrations were relatively stable until the middle of the 1980s. NO3concentrations were fairly constant in all rivers in the whole period. SO4* concentration is calculated on basis of measured SO42- concentration and the relation between SO42- and Clin seawater. Cl- is relatively unaffected by soil processes, whereas SO42- adsorption during sea-salt episodes may result in extremely low calculated SO4* concentrations, thus obscuring a more monotonic temporal trend. This effect is particularly important for the mean value of the years 1989, 1990 and 2000, as may be seen in Figure 3. In contrast, the year 1996 was particularly dry and the winter was cold, resulting in a relatively high mean SO4* concentration. We
TABLE 3. Correlation Coefficients for 75 Percentiles of Annual NAOI, Mg Deposition and River Water Chemistry (Cl, pH, and Labile Al) for Each Winter Season (November-April) for the Period 1984-2001
NAOI Mg dep. Cl, Trodøla pH, Trodøla Cl, Sæta pH, Sæta
Mg dep.a Haukeland
Cl Trodøla
pH Trodøla
LAl Trodøla
Cl Sæta
pH Sæta
LAl Sæta
0.85b
0.73c 0.86b
-0.77b -0.62c -0.49d
0.76b 0.66c 0.50d -0.81b
0.77b 0.87b
-0.81b -0.64c
0.71c 0.57d
-0.63c
0.67c -0.86b
a Mg-deposition was monitored at Haukeland 60 km southwest of River Sæta (Figure 1). of significance is shown: if p < 0.01. d Level of significance is shown: if p < 0.05.
b
Level of significance is shown: if p < 0.001. c Level
deviation of the Mg deposition. The mobilization of LAl from catchment soils depends apparently on both the acidification pressure on the soil and the sea-salt deposition. Multiple regression of LAl concentrations and the predictor variables SO4* and Cl-, representing acidification and sea-salt episodes, respectively, showed that Cl- contributed significantly (p < 0.05) in all regression models, whereas SO4* contributed significantly only for rivers Vikedal, Trodøla and Øyensa˚a:
LAlVikedal (µg L-1) ) -48 + 8.5 Cl + 1.18 SO4* (r2 ) 0.45, n ) 234) LAlSæta (µg L-1) ) 3.98 + 7.35 Cl (r2 ) 0.38, n ) 316) LAlNausta (µg L-1) ) 3.34 + 1.86 Cl (r2 ) 0.14, n ) 364) LAlTrodøla (µg L-1) ) -6.2 + 3.71 Cl + 0.39 SO4* (r2 ) 0.28, n ) 892) LAlØyensåa (µg L-1) ) -4.1 + 0.66 Cl + 0.46 SO4* (r2 ) 0.13, n ) 214)
FIGURE 3. Concentrations of nonmarine sulfate (SO4*; closed circles) and NO3- (open circles) for the rivers in the period 1980-2002. found a strong positive correlation (r2 ) 0.34; p < 0.01) between Cl- and SO42- for River Øyensa˚a due to the high marine SO42- contribution to the measured SO42- concentration. An almost similarly strong, but negative, correlation between Cl- and SO4* was also found, however, indicating that sulfate is adsorbed in the soil during the most severe sea-salt episodes. In fact, four of 14 calculated concentrations of SO4* at Cl > 400 µequiv L-1 were negative. The reduction in SO4* concentration was accompanied by increased pH and reduction in the concentration of LAl during the 1990s (Figure 4). The monitoring data indicate that the concentrations of LAl were at maximum in 19891990, a few years after the time when the sharp decrease in SO4* concentration started, but at the time of high positive
The ability of sulfate to be adsorbed in the soil during sea-salt events, as already pointed out and as found also in the UK (8), may partly explain the problems of obtaining significant contributions from SO4* to the models. Keeping the Cl- concentration constant a modeled reduction in LAl may be calculated for the rivers Vikedal, Trodøla and Øyensa˚a over the 17 years of monitoring that corresponds with the reduction in SO4* concentrations. The standard errors of the estimates are relatively large and the contribution from SO4* to the model relatively minor, making calculations of the reduction of LAl difficult. Increases in Cl- concentration may compensate for a reduction in SO4* concentration in order to maintain the levels of LAl, which means that the sea-salt episodes have to be more severe. Taking away the sea-salt part of the models, represented by Cl-, reduced the correlations dramatically, showing the importance of the sea salt episodes for potential toxic water chemistry for fish in the rivers. These regression analyses indicate that both the change in acidification and the sea-salt episodes are important for the variation in LAl. The sea-salt episodes seem to represent a driving force for episodic high concentrations of LAl. Acidification and Climate Variability. Strong positive correlations were found between wet deposition of sea-salts, sea-salt episodes in the watercourses and the North Atlantic Oscillation pattern (Tables 3-5, Figures 2 and 5). Comparing percentile values for the individual winter seasons makes it possible to study covariations between data series with dissimilar time resolutions. NAOI and sea-salt deposition data are available on a daily basis, whereas the water chemistry data are based mainly on monthly and VOL. 38, NO. 1, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
29
FIGURE 4. Concentrations of pH, inorganic monomeric aluminum (LAl) and nonmarine sodium (Na*) for the five rivers. (LAl is in µg L-1 and Na* in µequiv L-1).
TABLE 4. Correlation Analysisa-e Between Water Chemistry Variables for Rivers and NAOIf Vikedal (n ) 146)
Na Cl pH LAl
Trodøla (n ) 434)
Nausta (n ) 195)
Sæta (n ) 135)
Øyensa˚a (n ) 118)
NAOID
NAOIW
NAOIM
NAOID
NAOIW
NAOIM
NAOID
NAOIW
NAOIM
NAOID
NAOIW
NAOIM
NAOID
NAOIW
NAOIM
nsd nsd a a
a c a a
a a a a
a a a a
a a a a
a a a a
nsd c a b
c b a a
a a a a
nsd b b a
b a a a
a a a a
nsd nsd nsd nsd
nsd nsd nsd nsd
c b c c
a Level of significance is shown: if p < 0.001. b Level of significance is shown: if p < 0.01. c Level of significance is shown: if p < 0.05. d Level of significance is shown: ns (nonsignificant) if p > 0.05. e Analyses are based on NAOI and water chemistry data for November-April for each year. Number (n) of water chemistry data for each variable except LAl is shown in parentheses. N for LAl is lower for Vikedal (115), Nausta (165), and Øyensåa (97), due to the introduction of the analytical method in 1984. f NAOI was calculated on a daily (NAOID), a weekly (NAOIw) and a monthly (NAOIm) basis, see text.
weekly (River Trodøla) samples. Based on the observations, the 75 percentile values of Mg2+ deposition for the individual winter seasons correlated better with the corresponding percentile of the NAOI compared to the annual mean values and the 99-percentiles. The reason for this is probably that sea-salt deposition events are poorly represented by mean values, whereas the 99-percentiles are too much affected by single events that are not part of the main variations in the data sets. Correlations based on the 75 percentile values are thus presented in Table 3. The correlation between NAOI, deposition and river water chemistry is weakest when we use daily values of NAOI (Table 4). This was expected as the NAOI is calculated as an actual pressure difference distant from the location of the monitoring sites. Obviously, time delays between the measured sea 30
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 1, 2004
level pressure differences in the Atlantic Ocean, as are the basis for the index, and the resulting weather conditions, depositions and associated effects in Norway will occur. Stronger correlation for River Trodøla compared with the other rivers may be due to an artifact caused by more data from the weekly sampling in this river. Application of atmospheric transport models to assess the influence of the advection and the formation of sea-salt aerosols on the observed data would improve the understanding of factors causing sea-salt episodes and their link to the NAOI and is recommended for further studies. This would also allow a more direct link between sea-salt episodes and climate models and their downscaling, to provide regional scenarios on the impact of sea-salt episodes. The main objective for this study was, however, to carry out an
TABLE 5. Categories of Different Concentration Levels of Labile Aluminum (LAl) and Corresponding Mean Values of NAOIa,b river
LAl-cat., µg L-1
NAOID mean sdm
0-10 -0.07 11-20 0.41 21-40 0.19 >40 1.25 Sæta 0-10 -0.22 11-20 0.10 >20 0.29 Nausta 0-8 -0.01 9-15 0.05 >15 0.63 Trodøla 0-10 0.03 11-20 0.09 21-30 0.68 >30 0.67 Øyensåa 0-8 0.32 9-15 0.18 >15 0.39 Vikedal
a a a b a ab b a a b a a b b a a a
NAOIW mean sdm
NAOIM mean sdm
-0.25 0.41 0.04 0.96 -0.20 0.01 0.33 -0.04 0.05 0.81 -0.03 0.19 0.76 0.63 0.33 -0.02 0.67
-0.29 0.27 0.10 0.83 -0.22 0.05 0.39 -0.06 0.13 0.69 -0.04 0.21 0.66 0.65 0.24 0.32 0.55
a b ab c a ab b a a b a b c c ab a b
a b b c a b c a a b a b c c a ab b
n 28 24 39 25 46 50 41 94 46 26 201 164 53 17 56 23 19
a NAOI was calculated on a daily (NAOI ), a weekly (NAOI ) and a D W monthly (NAOIM) basis, see text. Analyses are based on NAOI and water chemistry data for November-April for the period 1986-2001. Number (n) of water chemistry data for each LAl-category is shown. N is high for Trodøla due to weekly sampling. b Significant different (p < 0.05) means (sdm) of NAOI within each river and NAOI-category are indicated by small letters, such that different letters mean significantly different mean values of NAOI.
initial investigation of the climate variability from statistical analysis of observed data. Figure 5 shows that the correlation between winter NAOI and both the median and 75 percentile values of sea-salt deposition is high. However, for a number of winter seasons, high sea-salt values are seen without a correspondingly high NAOI. One example here is the winter 1996-1997 for which sea-salt deposition in western Norway was very high. Looking at the individual observations it became evident that a single episode in March 1997 could explain the somewhat poorer correspondence. Also here, the sea-salt deposition was a result of strong winds and precipitation, but the origin of the air masses in part of this period was from the north, as can be seen from air mass trajectories (not presented). The eastAtlantic pattern was exceptionally strong, and the NAO was not the dominant pattern over the Europe-Atlantic region. The deviation observed here was not unexpected, because other weather conditions than the more typical pattern during high positive NAOI, with strong southwesterly winds, may also cause storms and sea-salt deposition at the western coast. We also explored how the potential toxicity of the water, indicated by LAl concentrations over a certain critical value, was related to NAOI. This was done by making separate LAlcategories for each river and by using all LAl concentrations for each category and the corresponding NAOIs for these sampling occasions. We calculated the mean NAOIs for all of the LAl-categories. Significantly different (p < 0.05) means of NAOI between high and low LAl-categories were identified by use of one-way ANOVA (Table 5). The differences of NAOI mean values were especially obvious when comparing the highest and lowest LAl categories, whereas the difference of the means between lower LAl categories was not clear for all rivers. By comparing the means of daily, weekly and monthly NAOI-values a stronger correlation between monthly NAOI values and water chemistry was found. We categorized LAl into concentration ranges in order to get relatively comparable number of observations in each category and in order to make categories suitable for discussion of potential toxicity for fish (Table 5). By defining
a LAl concentration of 20 µg L-1 for Atlantic salmon in these clear-water rivers as a critical chemical limit, the rivers had toxic water on 55,