Chemical composition of surface-and groundwater in fast-weathering silicate rocks in the Seiland Igneous Province , North Norway

The geochemical composition of nepheline-syenite, carbonatite, ultramafic rocks and gabbro samples from the Seiland Igneous Province is presented in terms of whole-rock chemistry, major and selected trace elements, and mineralogy. Additionally, the hydrochemistry of around 100 surface water, groundwater, rainfall and seawater samples from the area is presented. Surface water and groundwater solute chemistry can be explained in terms of reaction of precipitation with bedrock minerals: the waters thus acquire a clear chemical signature dependent on the catchment’s lithology. The mineralisation of the waters (total dissolved solids or TDS) is generally low, although it is somewhat elevated in carbonatiteor nepheline-syenitedominated terrains, in comparison with ultramafic, gabbro or granitic terrains. Additionally, each lithology exhibits its specific water type. Thus, waters in a nepheline-syenite-dominated terrain tend to be Ca–Na–HCO3 rich with relatively high pH, whereas waters in carbonatite terrains are Ca–HCO3 dominated with only slightly elevated pH. Waters in ultramafic rocks exhibit strongly elevated Mg concentrations with circumneutral pH, whereas waters from gabbros are variable in terms of their major cation (Mg, Ca, Na) composition. Mineral stability diagrams aid our understanding of the evolution of rainwater by chemical interaction with rocks in each lithology. In gabbro and ultramafic terrains, such diagrams suggest that kaolinite is likely to be the primary alteration product and that saturation with respect to calcite will not be reached; instead, the waters will evolve until saturation with respect to stilbite (a Na–Ca zeolite mineral) is attained. Waters in ultramafic rocks will then follow this stable phase boundary until saturation with respect to chlorite is achieved. The gabbro waters, on the other hand, could also follow the stable phase-boundary to stilbite until saturation with respect to paragonite (a sodium mica) is reached, depending on the plagioclase-to-clinopyroxene ratio (Plg/Cpx). Surprisingly, it is found that olivine (i.e., forsterite) does not control the water chemistry in gabbros and ultramafic rocks..


Introduction
The overall motivation of this investigation has been to get a better understanding of the chemical evolution of surface water and groundwater in fractured crystalline bedrocks.Solutes and gases in surface water and groundwater can derive from many different sources.Main contributions to the chemical composition include atmospheric input, interaction with the biomass and soil, and mixing with deeper, mineralised groundwater, dissolution of minerals from bedrock and anthropogenic input (Garrels & Mackenzie, 1971;Raiswell, 1984;Drever & Hurcomb, 1986;Tranter et al., 1993;Brown et al., 1994;Aquilina et al., 1997;Stober et al., 1999).Separation of the different contributions is difficult in densely populated low-altitude catchments covered by vegetation (Hilberg & Riepler, 2016).
In this study, the composition of surface water and groundwater under arctic climate conditions in a barely populated area with a thin or deficient biomass and Stober, I., Teiber, H., Li, X., Jendryszczyk, N. & Bucher, K. 2017: Chemical composition of surface-and groundwater in fast-weathering silicate rocks in the Seiland Igneous Province, North Norway.Norwegian Journal of Geology 97, 63-93.https://dx.doi.org/10.17850/njg97-1-04.soil cover is investigated.In these catchments far north of the Arctic Circle, the diverse potential contributions to the chemical composition of surface water and groundwater are very limited.In more or less bare-rock areas, the chemical weathering of rocks and minerals causes primarily solute acquisition of surface waters.Thus, the chemical reaction of atmospheric precipitation with exposed fractured rocks is the main cause for the observed composition of surface water.Since the area of investigation is surrounded by sea, a marine influence is to be expected.
The Seiland Igneous Province in North Norway has an extremely varied bedrock geology, including nephelinesyenite, carbonatite, gabbro and ultramafic rocks (e.g., Heier, 1961;Bennett, 1974;Bruland, 1980;Gardner, 1980;Strand, 1981;Robins, 1996;Gautneb, 2009).Since these rocks are composed of relative fast-weathering minerals compared with granites or gneisses, this diverse bedrock geology is expected to be reflected in the composition of waters in the brooks and lakes of the investigation area.
plagioclase + water (± CO 2 ) = zeolite / clay mineral + (Ca 2+ +) Na + + SiO 2aq (3) Thus, the model process of faster weathering minerals should help us to gain a better insight into the alteration processes of crystalline rocks with slower weathering components.
On the island Stjernøya of the Seiland Igneous Province in northern Norway (Fig. 1), there is a huge outcrop of a nepheline-syenite body with one of the largest (sub) surface nepheline mines in the world (e.g., Heier, 1961;Bruland, 1980;Robins, 1996;Roberts, 2007).In the nepheline-syenites, relatively fast reaction kinetics were to be expected, to observe directly the development of the hydrochemical properties of waters due to alteration processes (WRI).To collect water and rock samples within the nepheline-syenite mine, run by Northcape Minerals AS, was a unique possibility.By reason of the mine's vertical depth extension of about 700 m, it is possible to study the influence of increasing depth, e.g., prolonged reaction time, extended contact surfaces, etc., on the hydrochemistry of waters.Analogue studies in the likewise 700 m-deep Clara mine, situated in gneisses of the Black Forest (southwest Germany), showed increasing solutes and a change in water type with increasing depth (Stober et al., 2002).Thus, hydrochemistry, flow path, contact time, and geochemistry of the rock and alteration products provide insight into total reaction rates (e.g., Anderson, 1996;Drever, 1997).By studying the weathering process of nepheline-syenite it was expected to deduce a model for water-rock interaction (WRI) in aluminium-silicate rocks.
Another reference system to study relatively fast weathering rocks relates to the presence of carbonatites, which occur in surface outcrops in close proximity to the mine on Stjernøya (Bruland, 1980;Strand, 1981;Gautneb, 2009).In the carbonatite, calcite is the only carbonate phase.Calcite is a very fast weathering mineral.
Olivine, present in ultramafic rocks and in olivinegabbros of the Seiland Igneous Province, also weathers at a very fast rate compared to other Mg-Fe-silicate minerals, such as hornblende and mica (Lasaga, 1984(Lasaga, , 1998)).The alteration process of olivine (eq.2) and its influence on hydrochemistry is considered to be very important.The intention was to find out if this process can be regarded as a model for Fe-Mg-silicates or if there are other minerals controlling the weathering process more decisively.In olivine-free or -poor gabbro catchments, the weathering process of pyroxene and plagioclase is expected to have a dominant control of the chemical properties of these waters.
Thus, the overall aim and motivation of the study has been to better understand the chemical evolution of surface water and shallow groundwater in fractured aluminium-silicate rocks.All the investigated rock-types in the Seiland Igneous Province, i.e., nepheline-syenite, carbonatite, gabbro and ultramafic rocks, are exposed to more or less identical hydrological conditions, which makes the island uniquely suited for investigations on weathering effects in different fractured, crystalline bedrocks.The study is mainly focused on surface waters with a relatively short contact time in the rocks.Nevertheless, investigations in the 700 m-deep underground mine on Stjernøya and in two tunnels (through gabbro and granite) were carried out to see how the infiltrating water chemistry evolves with depth.
In contrast to this study, Pfeifer (1977), Clark & Fontes (1990), Marques et al. (2008), Chavagnac et al. (2013) and Miller et al. (2016) investigated upwelling springs deriving from ultramafic rocks.The spring waters showed high TDS and pH values.According to Marques To our knowledge, investigations on how water will evolve in carbonatite-or nepheline-syenite terrains under these climatic conditions have not previously been carried out in detail.However, Zakharova et al. (2007) described the composition of groundwater in the Karelia region and the Kola Peninsula, NW Russia, and linked it to that of the bedrock, but no detailed investigations were carried out describing the weathering process and the evolution of the waters.Either way, the total of solutes (TDS) in these areas of NW Russia are in the same order of magnitude as the water samples collected by us in the Seiland Igneous Province.
This paper also presents the first systemic investigation of the hydrochemical properties of surface waters in the different rock-types of the Seiland Igneous Province, i.e., nepheline-syenite, carbonatite, gabbro and ultramafic rocks.
et al. (2008), these waters belong to a Ca-OH water type, most probably depleted in Mg as a result of precipitation of brucite and serpentine minerals at the surface.
In contrast to this study, most of the above investigations were not focused on a detailed development of rainwater in contact with relatively fast-weathering crystalline basement rocks.Additionally, most examinations were carried out in the so-called Temperate Zone, i.e., under moderate temperature conditions, in most cases even under Mediterranean climatic conditions, whereas our research has been done in the Polar Regions, i.e., under arctic climate conditions.Griffin et al., 2013).
Topography, hydrology and geology of the investigated area The Seiland Igneous Province consists of a deepseated alkaline magmatic suite of dominantly gabbroic composition and is located in the county of Finnmark, northern Norway, and hosted within the northernmost part of the Caledonian orogenic belt (Fig. 1).The island Seiland lies more or less in the centre of this igneous province, surrounded by Sørøya, Stjernøya and the Øksfjord Peninsula.
Situated far north of the Arctic Circle at latitude of 70° N, for two month of the winter the sun is always below the horizon.Correspondingly, during the summer the county benefits from two months of midnight sun.The topography of the Seiland Igneous Province is very rough with steep mountain sides down to the sea and peaks up to about 1.000 m a.s.l.The coast is indented by large fjords and the whole area is more or less above the treeline with stunted birch in sheltered fjord areas.The central parts of the islands are glaciated, in part with huge plateau glaciers like the Øksfjordjøkelen (45 km²) or the Seilandsjøkelen (14 km²).The Øksfjord plateau glacier calved directly into the sea (Jøkelfjorden) until the last century, but is today slowly receding.
In Finnmark, the Sámi culture is still thriving; especially notable is the reindeer husbandry with summer grazing at the coastal areas and winter grazing on the heathland plateau inland.Due to the mineralogy of the bedrock, soils of the Seiland Igneous Province are generally carbonate-rich.Especially on Stjernøya the soils are rich in carbonate and potassium minerals and are therefore very fertile (Heim et al., 2010).Thus, the island is more or less a green grassland.In summer large flocks of reindeer are now brought by the Sámi to the island for grazing, whereas earlier the reindeer had to swim across the Stjernsund.
Annual precipitation is about 830 mm, with somewhat lower rain-or snowfall in spring and higher in the autumn.Mean annual temperature is about 3.1°C.The area is very sparsely inhabited and the population is steadily declining.Stjernøya, for example, now has no road or ferry connections.
The approximately elliptical Seiland Igneous Province, about 100 km x 50 km in size, consists of different intrusions such as mafic and ultramafic plutons, mafic dykes, alkaline complexes and carbonatite bodies (Bennett, 1974;Roberts et al., 2010).The intrusions were emplaced during a time span from 570 to 520 Ma (Roberts et al., 2006(Roberts et al., , 2010;;Menegon et al., 2011), in an inferred intracontinental rift setting, most likely related to the incipient stages of opening of the Iapetus Ocean (Siedlecka et al., 2004;Roberts et al., 2006;Menegon et al., 2011).Thus, the Seiland Igneous Province is supposed to represent the eroded roots of an intracontinental rift igneous complex, similar to the East African Rift Valley.The occurrence of nepheline-syenites and carbonatites within the province is one supportive argument (Mjelde, 1983;Kirkland et al., 2008;Roberts et al., 2010).In general, the tectonic evolution of the Seiland Igneous Province and whole of Finnmark is complex and still under discussion.
The Caledonian fold and thrust belt emerged from the collision of the continents Laurentia and Baltica in Silurian times, where the margin of Baltica is considered as having been subducted beneath Laurentia (Torsvik et al., 1996;Roberts, 2003).As a result of this collision a stack of nappes was produced on the Precambrian basement of Baltica.The allochthonous nappes are basically composed of Neoproterozoic to Silurian metasedimentary rocks intruded by diverse igneous bodies and dykes.This overthrust package, consisting of four major nappe complexes, caused crustal thickening and associated metamorphism, similar to the Alpine orogeny (Bucher, 1991).The Seiland Igneous Province occupies the westernmost part of one of these nappes, the Kalak Nappe Complex.
Globally, nepheline-syenite and carbonatite are relatively uncommon rocks; however, in the Seiland Igneous Province they are comparatively widespread as large complexes.One of the best known carbonatitic bodies is located NNE of Lillebukt in the south of Stjernøya forming the northern slope of the mountain Nabbaren, where the nepheline-syenite is mined by Northcape Minerals, and is extending to Smaldalen in the north.In addition, nepheline-syenite dykes are found throughout the Seiland Igneous Province.Syenite rocks, carbonatites and mafic dykes are the youngest igneous rocks in the province and are believed to be related to the nephelinitic magmatism.The nepheline-syenite pegmatites represent the latest phase of magmatic activity at c. 520 Ma (Pedersen et al., 1989) and are thus older than the Caledonian orogeny.
The Lillebukt Alkaline Complex covers a 13 km 2 area situated in the south of the island Stjernøya.It is dominated by syenite, nepheline-syenite, apatite-  (after Roberts, 1973).water samples and of selected rock samples, important within this context.Sampling was carried out mainly on the islands of Stjernøya, Seiland and on the Øksfjord Peninsula and was achieved by using a small motorboat to gain access and to collect samples in the different lithological units.Care was taken to acquire representative rock samples of each lithological unit and to obtain water samples from the differentbut well defined catchment basins with equal lithological units, in order to obtain general information on the surfacewater composition dependent on the varying geology.
Water and rock samples were also taken from inside the nepheline-syenite mine of Northcape Minerals, i.e., within the mountain Nabbaren on Sternøya, and in two tunnels, Stetind-and Øksfjordtunnel, beneath overlying ground reaching up to about 1000 m and 350 m, respectively.
Information concerning the rock-and water samples is listed in Tables 1 & 2A-E, together with the major-and trace-element geochemistry of the rock samples (Table 1) and the hydrochemical results of the water samples (Table 2A-E).
rich hornblende-clinopyroxenite dykes, hornblendite, metagabbro and silicocarbonatite and can be regarded as an irregularly deformed and metamorphosed ringcomplex with the carbonatite as the latest intrusion (Skogen, 1980).The central part of the silicocarbonatite intrusion is dominated by biotite.Along the margin of the carbonatites, there are many apatite-rich hornblendeclinopyroxenite dykes which are surrounded by fenites and are believed to represent injections from an olivine melanephelinite magma undergoing fractional crystallisation (Strand, 1951;Heier, 1961;Bruland, 1980;Mjelde, 1983;Robins, 1996).

General information
Overall, an extensive and broad collection of rock and water samples from the Seiland Igneous Province was carried out during fieldwork in the summers of 2003 and 2009.Fig. 2 shows the localities of all collected                           precipitation was not expected.Nevertheless, all water samples were stored in a refrigerator until they were analysed in the laboratory.
Rain-and sea-water (Table 2A; Fig. 2) were collected as reference samples for the field study and to have the opportunity to consider possible influences concerning the input via precipitation and/or sea spray, when studying silicate weathering processes.Rainfall was collected during single storm events using a polyethylene bottle with a huge polyethylene funnel.Sea-water sampling was carried out by boat at about 1 m depth below the sea surface also using polyethylene bottles.
The water samples from boreholes in the mine were collected after a longer flow period, to avoid contamination with water stored in the borehole.However, all boreholes showed 'artesian' behaviour, i.e., free outflow.Fracture seepage samples from the mine and the two tunnels were collected directly at the 'outflow' points.Surface water samples from brooks, rivers or springs in talus were collected directly in the flowing water.The thickness of the accumulation of broken rock debris (talus) was low.
The locations of the water samples collected in graniteor gneiss-dominated terrains are given in Table 2E.The samples collected in the granite-dominated terrain originate from the immediate vicinity of the Stetind (68°09'54''N, 16°35'33''E, elevation: 1392 m) and the Stetind tunnel (about sea level).Samples taken in gneissdominated terrains were collected in rivers (Table 2E) just south of the area presented in the geological map (Fig. 2).

Sampling methodology
Water samples for chemical analyses were collected in 500 ml acid-cleaned polyethylene bottles (Fig. 3).From each location a duplicate was taken.Care was taken to fill the bottles completely to avoid gas exchange.Temperature, pH-value, Redox potential and electrical conductivity (EC) were measured in the field.
Immediately after sampling, alkalinity was determined by titration with 0.1 N HCl to a certain pH end point by using a pH-meter, whereas waters with pH >8.3 had to be titrated to pH = 8.3 and pH = 4.3.According to the very low mineralisation (low EC) of all water samples,

Results of investigated rock samples
In this chapter, a selection of the rock samples collected during fieldwork is described, comprising rocks typical of the Seiland Igneous Province including carbonatite, nepheline-syenite, gabbro and ultramafic rocks.Fig. 2 shows the locations of these rock samples.Most of the mineralogical and geochemical investigations were carried out within the context of two master theses (Teiber, 2010;Jendryszczyk, 2013) and one PhD thesis (Li, 2013) at the University of Freiburg under the supervision of the first and fifth author.Based on the medium-and coarse-grained appearance of all samples the rocks can generally be classified as plutonic.
The mineralogy and petrochemistry of carbonatites were previously investigated by Strand (1951Strand ( , 1981) ) and Bruland (1980).According to Woolley & Kempe (1989) they could be characterised as calcium-rich silicocarbonatites.The collected carbonatite samples (Fig. 2) are medium-to coarse-grained plutonic rocks consisting of varying amounts of calcite, amphibole, biotite, apatite and Fe-Ti oxides, whereas calcite is the only carbonate phase.Table 1 shows the results of the XRF analyses of the whole-rock chemistry, where according to custom the major elements are expressed as oxides in wt.% and the trace elements in ppm.An average whole-rock composition of 26 wt.%SiO 2 and 18.3 wt.% CaO has been measured.The MgO content varies from 2.1 to 3.8 wt.% and P 2 O 5 between 0.8 wt.% and 1.4 wt.%, whereas phosphorus is incorporated in apatite (Jendryszczyk, 2013).The carbonatite samples are quite rich in alkalis (>4.8 wt.%) and contain high amounts of Al 2 O 3 (8.44-14.69wt.%), corresponding to the large proportion of observed aluminosilicate minerals.Furthermore, in the silicocarbonatites a high enrichment of trace elements such as barium (Ba) and strontium (Sr) is detected (Jendryszczyk, 2013).These results are in line with the investigations of NGU (Geological Survey of Norway) presenting mean values based on 65 samples (Gautneb, 2009;Heim et al., 2010).They found that the silicocarbonatites contain about 30% biotite, 40% calcite and 7.5% apatite and detected unusually high Ba-and Sr-concentrations in the carbonatites, with barium mainly in biotite and strontium mainly in calcite.
Two different calcite phases were observed by Jendryszczyk (2013): a primary crystallised SrO-FeOrich and MnO-poor phase with about 2 wt.%SrO and a secondary SrO-FeO-poor and MnO-enriched phase, partly replacing the primary one.In addition, MgO is about 10 times higher in the primary phase (~0.35 wt.%) than in the secondary phase.Biotites are Ba-rich with 0.24-3.77wt.% BaO and contain about 28 wt.%FeO and 8 wt.%K 2 O. Apatites and two types of amphibole are also major mineral phases in the silicocarbonatites.Other accessory minerals like feldspar, nepheline, pyrite and zircon show an irregular distribution and are not present in all samples (Jendryszczyk, 2013).
Alteration materials such as 'white crusts' on rock surfaces were collected by using a knife and storing the samples in small plastic bags.
Care was taken to sample unaltered rock material by using a heavy hammer and a chisel to avoid weathering crusts or fractures in the collected rocks.

Analytical methods
Optical polarisation microscopy and X-ray fluorescence analyses were performed on the rock samples.The 'white crust' has been exclusively investigated by X-ray diffraction analysis.The composition of minerals in the rock samples was determined by electron probe microanalysis (EPMA) using the CAMECA SX100 instrument at the IMPG (Institute of Mineralogy Petrology and Geochemistry, University of Freiburg).
EPMA is a tool to determine non-destructively the chemical composition of small volumes of solid materials.All quantitative analyses were made using the wavelength dispersive spectrometers of the instrument.
The acceleration voltage was 15 kV, the beam-current 10 nA and the counting times 10 s.Analyses were made using a focused electron beam (2 µm spot size) except for Na and K measurements in mica and alkali-feldspar where a defocused beam was used (c. 10 µm).The instrument was calibrated for each element analysed using well characterised natural materials as standards.Data reduction was performed by utilising software provided by the manufacturers of the instrument.The results of the whole-rock chemistry, major and trace elements, are presented in Table 1.
The chemical composition of the water samples was determined in the laboratory of the IMPG after filtration through cellulose acetate membrane filters with a 0.45 µm pore size, using ion chromatography for the anions (Dionex 120) and atomic absorption spectroscopy for cations (AJ Vario 6).Detection limits for Ca, Sr, F, Br and NO 3 are 50 µg L -1 , for Na, K, Li and Mg 20 µg L -1 , for Cl 500 µg L -1 , for PO 4 100 µg L -1 and for SO 4 800 µg L -1 .Si was not analysed.The total of dissolved solids (TDS) has been calculated from the analyzed concentrations.The analytical data of the water samples are presented in Table 2A-E.
The overall charge error of the analyses expressed as electroneutrality (EN) was calculated from (meq L -1 ): EN = 100 * (sum of cations -sum of anions) / (sum of cations + sum of anions) (4) The average EN for all 100 samples was <4%.In 6 samples the EN was >10%  1).The results of the CIPW Norm calculation are in good agreement with the findings of the polarisation microscopy.
The higher MgO contents in some samples (Table 1) seem to be caused mainly by small amounts of olivine.The CaO contents can be assumed to be related, on the one hand, to the amount of clinopyroxenes and amphiboles detected by means of the polarisation microscopy and, on the other hand, to the amount of plagioclase present in the gabbro samples.The peridotite samples 1110 and 1113 are characterised by very low alkali and silica contents (<0.3 and <40 wt.%, respectively) in contrast to the gabbro samples (0.7-3.7 wt.% and 41-51 wt.%;Table 1).Ni shows higher concentrations in the peridotites (c.1100 ppm), respectively in the olivine-bearing gabbros where Ni is supposed to be mainly incorporated in olivine.Cr is also found with higher concentrations in the two peridotites (c.2000 ppm), presumably incorporated in the spinels or in other minerals like pyroxene.The Sr contents are higher in the classified gabbro samples (200-1000 ppm) relative to the ultramafic samples (<50 ppm).For example, Sr may probably substitute for Ca in the plagioclases.In addition, Ba contents are generally also higher in the gabbros than in the ultramafic rock samples.
Electron probe microanalysis (EPMA) of 36 plagioclases in the rock samples 1028, 1281 and 1046, characterised as olivine-gabbro, and in the troctolite sample 1224 has shown noticeable variations in chemical composition from sample to sample, but the geochemical range of the major elements is quite low (Teiber, 2010).All plagioclases in the olivine-gabbros (Table 1) show a significantly higher anorthite (≤80 mol.%) than albite component, whereas the plagioclase in the troctolite sample 1224 has significantly lower An and Ab contents.However, all plagioclases are Ca-rich, and are in line with the CIPW Norm analyses.
Analogous investigations were carried out on 21 pyroxenes in the above-mentioned olivine-gabbros (1028, 1281), in the troctolite (1224), and in the peridotite sample 1113.Taking into account the high Ca and Mg content, the pyroxenes could be classified On a macroscopic scale, Bruland (1980) observed intrusive carbonatite veins within the nepheline-syenite.However, the collected nepheline-syenite samples show some few, small, white reaction veins with bilateral, slightly reddish, reaction zones, whereas the fresh nepheline-syenite has a greyish colour.(Li, 2013).The analytical data are in good agreement with the data (mean value and variance of 65 analyses) published in Gautneb (2009).
The composition of the mineral nepheline can generally be expressed as a solid solution between the three end-members: ideal sodic nepheline (Ne -NaAlSiO), kalsilite (Ks -KAlSiO 4 ) and quartz (Qz -SiO 2 ).The mineral chemistry of the investigated nepheline showed variations in composition between Ne 80 Ks 20 Qtz 1 and Ne 77 Ks 18 Qtz 4 without systematic differences.An average Na/K ratio in the nepheline of 3.95 was detected with a variance of 3.63-4.33.The investigated K-feldspars showed a Na-and K-rich lamella texture and thus two main composition groups without an An component: Ab 15-30 and Or 65-86 (An -anorthite, Ab -albite, Ororthoclase).Additionally, Li (2013) detected some ternary-phase feldspars with a variable composition and a low An component.According to the international nomenclature (Leake et al., 2004), the amphiboles are Caand Al-rich hastingites and ferro-paragasites.Most surface water samples were collected in small brooks along the coastline, but some samples were taken in small lakes and brooks farther inland on the islands or the peninsula.Three samples were collected from within road-tunnels and 10 samples within the nephelinesyenite mine at Lillebukt in the south of Sternøya (Fig. 2; Table 2B-E).Since the investigated area is very sparsely inhabited, the water samples in this study are not affected by anthropogenic activity with the exception of a few samples from the nepheline-syenite mine.
The temperature of the collected water samples is low, generally about 6°-22°C (Table 2B-E).The pH value varies between acidic with pH = 5.45 (W65) and alkaline with pH = 9.39 (W5).All water samples are negligibly mineralised with TDS between 7 mg L -1 (W66) and 118 mg L -1 (W75).The highest TDS values, >80 mg L -1 , have been detected in waters circulating in carbonatite rocks and in waters of the nepheline-syenite underground mine.TDS values >50 mg L -1 are found in springs from talus material of gabbro and ultramafic rocks and in brooks passing across a carbonatite-dominated terrain or in waters collected in the nepheline-syenite underground mine.The lowest TDS values, <20 mg L -1 , originate from brooks in granitic or metamorphic rocks or sandstones, from small brooks in compact gabbro or ultramafic rock complexes and from glacier lakes.
Dissolved silicon (Si) was not analysed, such that in the discussion section various assumptions have had to be made regarding the likely saturation status of the waters with respect to silica minerals.However, there is no evidence that surface water or groundwater are saturated with respect to chalcedony or quartz, since silica sinter was never observed during sampling.
In the following, the water samples are grouped in different so-called Schoeller-diagrams (Fig. 4A-F (Teiber, 2010).
The XRD pattern of the weathering surface (white crust) on the olivine-gabbro sample 1046 suggests that the crust consists of Ca-and Na-bearing zeolites (Teiber, 2010).According to Bucher & Stober (2000), the formation of this mineral most probably originates from the rock sample itself, due to weathering of Ca-rich plagioclase.

Results of the water analyses
Standard sea-water is highly mineralised (TDS about 38 g/kg) in the open temperate and tropical oceans as a consequence of little rainfall and high evaporation (Table 2A), whereas in cold seas with melting ice the mineral concentration is significantly lower (TDS about 30 g/ kg).Generally, sea water has a slightly alkaline pH in the range 7.5 to 8.4.Two seawater samples (WS6, WS54) were collected in 2003 and 2009 at different locations (Fig. 2; Table 2A).The pH values of the samples are in the order of 8.2 and the TDS about 30 g/kg, typical for ocean waters atlatitude >70°N.The dominant ions in the collected sea-water samples are Na and Cl with elevated Mg concentrations, which are characteristic values for sea-water (Nordstrom et al., 1979).Surprisingly, both samples have different TDS, 29.7 g/kg and 31.1 g/kg, respectively.Sample WS6, collected within the Kvaenangen fjord, has a significantly lower concentration than sample WS54, sampled in the ocean at a greater distance from the fjords (Fig. 2).The difference might be an effect of inflow of water from rivers and brooks and/or of density layering in the more sheltered fjord compared to the open ocean, where water circulation and mixing is more intensive.Nevertheless, both sea-water samples (WS6, WS54) show Na/Cl ratios in the same order, 0.80 and 0.81, respectively.
higher TDS are especially caused by higher calcium, sodium and magnesium concentrations.Nevertheless, potassium is slightly increasing too with TDS.The waters collected in ultramafic rocks belong to a Na-Mg (Ca)-HCO 3 -or Na-Mg-HCO 3 -type, when residence time is short (water samples from compact rocks), and they seem to develop to a Mg-Ca-Na-HCO 3 -type with a more prolonged residence time and enhanced contact surface area (ultramafic talus material; Fig. 4C).
The TDS in water samples collected in 'ordinary' gabbro-dominated terrain (Fig. 2; Table 2C) is generally low, in most cases even <30 mg L -1 .In gabbroic talus material waters show a significantly higher TDS, up to 45 mg L -1 .Elevated TDS values (46 mg L -1 , 48 mg L -1 ) were also recorded in the two samples collected in the Øksfjordtunnel (W5, W99).The pH of the gabbro samples is circumneutral, with the exception of the two samples collected in the Øksfjordtunnel which show alkaline values.Water type in gabbro rocks is quite variable, thus Na-, Mg-and Ca-dominated waters are found, all with HCO 3 as a main anion component (Fig. 4D).Since TDS is generally low, Cl is sometimes also one of the main anions.Waters collected in clinopyroxenegabbro, foliated syeno-gabbro or olivine-gabbro rocks (Fig. 2) strongly resemble waters in 'ordinary' gabbro material (Fig. 4E, F), with the exception that the TDS of waters in clinopyroxene-gabbro generally might be slightly higher.The two water samples collected in syenogabbro have very low TDS (Fig. 4F).Whereas waters collected in clinopyroxene-gabbro mostly have higher Ca concentrations than waters circulating in 'ordinary' gabbro (Fig. 4E, D), the low-concentrated samples from the syeno-gabbro have relatively high Na concentrations.Water samples collected in norite resemble those in syeno-gabbro, but show a slightly elevated TDS (Fig. 4G).
Since conditions are oxidizing in surface water or shallow groundwater, Fe and Al (not analysed) are generally low as pH is circumneutral in waters collected in gabbro-and ultramafic rock catchments.However, filtration of water samples in the laboratory might remove any precipitates or colloids if present, leading to an underestimation of dissolved Fe contents in the waters.
Water samples collected in granite and gneiss catchments reveal rather diverse Schoeller-patterns (Fig. 4H; Table 2E).The majority of the samples were collected in gneisses, which have higher TDS and are dominated by Ca and HCO 3 (W69, W70, W74, W78), whereas waters collected in granites are lower concentrated and seem to be dominated by Na and Cl (W60-W62).
The Piper-diagram (Fig. 5) shows the distribution of the main components of all collected water samples.According to Fig. 5 the waters are subdivided in certain subgroups dependent on the lithological properties of the rocks, in which the waters are circulating.Thus, waters in carbonatite-and nepheline-syenite-dominated respect to the geology of the catchment area, where they were collected.It is important to mention that the collected samples may not exclusively be affected by the geology of the catchment as shown in the map, since in some cases there are additional small outcrops with a different lithology which cannot be shown in the scale of the map (Fig. 2).Thus, some water samples are hydrochemically affected by these 'small outcrops' .In addition, within the nepheline-syenite mine relatively high nitrate concentrations of up to 17 mg L -1 were measured deriving probably from the explosives used in mining.
As mentioned earlier, the Lillebukt Alkaline Complex (Fig. 2) consists mainly of carbonatites, nepheline syenites, metagabbros and hornblendites.The chemical properties of the water samples within the nephelinesyenite mine inside the mountain Nabbaren clearly show a dependency on the thickness of the overburden.Thus, TDS is increasing with depth.The TDS near the top of the mountain (c.700 m a.s.l.) is about 41 mg L -1 , whereas deep inside the mountain (c. 100 m a.s.l.) the TDS is 118 mg L -1 .This overall increase, with respect to the cations, is mainly caused by an increase in sodium but also by calcium and potassium, while magnesium shows no depth dependency (Fig. 4A; Table 2B).The water samples of the nepheline-syenite mine have the highest pH values of all samples.Water samples in the mine are alkaline with pH values generally between 8.05 and 9.36, with the exception of one dripping water sample from of the tunnel ceiling (pH = 7.48).
Surface waters sampled within the carbonatite rocks of the island Sternøya (Fig. 2; Table 2B) have TDS of about 100 mg L -1 and are thus the samples with the highest TDS observed in surface waters.Waters are mainly composed of calcium and hydrogen carbonate, whilst magnesium and sulphate contents are low (Fig. 4B).PH of these waters fluctuates between 7.72 and 8.05, reflecting a calcite buffering system.In the carbonatite rocks pH is significantly lower than in the waters of the nephelinesyenite mine.
On the island Seiland, many samples were collected from small brooks and lakes situated in ultramafic rocks (Fig. 2; Table 2D).Waters sampled in ultramafic rocks typically have elevated magnesium concentrations (Fig. 4C).The pH of the ultramafic water samples is circumneutral (6.65-7.69),whereas waters sampled in the ultramafic talus material (W16, W17, W22) have pH values in the upper range of the interval.TDS values of the water samples are between 11.9 and 68.1 mg L -1 and thus generally lower than the waters collected in the nepheline-syenite mine or in carbonatite rocks.The 3 water samples from ultramafic talus material show TDS of more than twice (51.4-68.1 mg L -1 ) the values of samples from compact ultramafic material (11.9-32.4mg L -1 ), probably caused by the enhanced contact surface area and residence time in the talus material.The much

Discussion
Generally, TDS are, amongst other things, a function of residence time, reactive surface area, degree of alteration and the mineralogical composition of the rock.Prolonged residence time and enhanced water-rock interaction as a result of a larger reactive surface area and increased degree of alteration of talus material are expected for waters circulating in a comparable terrain.Thus, enhanced TDS are observed there.
terrains contain very low Mg-and high Ca-portions, whilst waters circulating in gabbros or ultramafic rocks show proportionally much higher Mg and lower Ca.Waters in granites and gneisses are somewhere in between.In contrast to the Piper-diagram (Fig. 5) the Schoeller-diagrams (Fig. 4A-H) show additionally the waters' mineralisation, while a Piper-diagram has the advantage that many more analyses could be plotted in the figure.

Marine influence
All surface water samples showed significant amounts of chloride between 0.91 mg L -1 and up to 15.69 mg L -1 , even though the minerals of the rocks of the Seiland Igneous Province have no chloride bound in their mineral lattice.
The Cl/Br ratio (mg-basis) of the two sea-water samples (WS6, WS54) is 273 and 281, respectively, corresponding to Cl/Br investigations of recent seawater worldwide (e.g., Braitsch & Herrmann, 1963;Stober & Bucher, 1999a;Rao et al., 2005).Since the TDS of most water samples collected in the Seiland Igneous Province are quite low, bromide could be detected in only a few samples (15 Br data) with values of Br ≤0.06 mg L -1 .Most of these samples showed the measured sea-water signature in Cl/Br.Two samples had slightly higher Cl/Br values, but bromide analysed in these samples was very low (Br = 0.02 mg L -1 ), actually below the detection limit.Thus, the chloride in the water samples with reliably detectable bromide concentrations had a sea-water signature, i.e., the origin of chloride is most probably marine.
To obtain some further information concerning the marine influence on the collected water samples, a closer look was taken at the Na/Cl ratio (mol-basis), which is Na/Cl = 0.8 for seawater (Fig. 6).The rainwater sample WR76 has the same Na/Cl ratio as seawater.As mentioned before, rainwater sample WR59 is of lower The distribution of species and saturation states of the waters were calculated using the geochemical code Phreeqc (Parkhurst & Appelo, 1999).The saturation index (SI) relates the ion activity product (IAP) to the equilibrium constant (K) of the dissolution reaction of a considered mineral at a given P and T: SI = log(IAP/K).However, using the code Phreeqc and the llnl database (Parkhurst & Appelo, 1999), all collected water samples (with exception of the two sea-water samples) are undersaturated (SI <0) with respect to all possible solids, dependent on the analysed, respectively analytically detectable parameters in the water samples (Table 2B-E).Taking the whole hydrogeological situation of the relatively short circulation time into account, most water samples could be undersaturated with respect to quartz.
According to Bruland (1980) and Gautneb (2009), carbonatite vein intrusions and/or calcite-rich mafic dykes are occurring in nepheline-syenites.In gneisses, fractures filled with precipitates of calcite or Ca-zeolites are quite common (Bucher & Stober, 2000).Thus, some water samples in these lithological units show a possible additional source of Ca, probably from locally occurring Ca-carbonate minerals (e.g., calcite: CaCO 3 ).component of nepheline-syenite and carbonatite in water is calcium, whereas the content of magnesium is very low (Fig. 7).The high calcium component in the nephelinesyenite waters is most probably caused by the presence of carbonatite vein intrusions and/or calcite-rich mafic dykes in the nepheline-syenite, as observed during fieldwork and described by Bruland (1980) and Gautneb (2009).Water samples from ultramafic rocks show a very high magnesium component (central part in Fig. 7).In gabbros, including clinopyroxene-, olivine-and syenogabbros, and norites, calcium or the earth alkali (Ca + Mg) dominate the water-type (left side of Fig. 7).The right-hand side of Fig. 7 shows the cation distribution of waters circulating in gneisses and granites nearby and outside the Seiland Igneous Province.

Waters in carbonatites and in the nephelinesyenite mine
Taking into account the marine influence on all water samples of the Seiland Igneous Province, the subsequent investigations on water-rock interaction are based on the hydrochemical data reduced by the marine component.
The dominating water-rock reaction in carbonatites is given by calcite (eq.5), as a result of its relatively high dissolution rate and frequency in the rocks.
quality.The Na/Cl ratio of all surface and subsurface water samples is equal to or higher than in the seawater samples (Na/Cl = 0.8).The highest Na/Cl ratios are from waters collected in the nepheline-syenite mine (W31, W32, W39).High values are also found in waters circulating in ultramafic rocks (e.g., W20; Fig. 6).
Thus, the source of chloride in the collected, fast-circulating water samples is marine, as a result of the marine component in rainwater and the influence from blown spray from the sea.Sodium, on the other hand, has both a marine component (Na/Cl = 0.8) and a 'rock' component (Na/Cl >0.8), deriving from alteration processes that have affected the rocks.Note: Na/Cl (mol-basis) of all surface waters and groundwater is >0.8 (Fig. 6).
Taking into account that parts of the Na concentrations measured in the water samples are of direct marine origin, Ca or Ca + Mg are the main cations in nearly all collected waters, even in samples from norites and syeno-gabbros (Fig. 4G, F).By neglecting the marine component of the sodium, the alkaline earth metals (Ca + Mg) in most water-samples of the Seiland Igneous Province are the main cations (Fig. 7), with the exception of two samples from the deeper part of the nephelinesyenite mine (W31, W32).
Considering this marine influence, the main alteration in thin-sections the albitisation of nepheline.She also found a co-precipitate of calcite with thomsonite (Ca 2 NaAl 5 Si 5 O 20 · 6H 2 O).Nonetheless, this process needs a Ca-and HCO 3 -rich fluid, deriving probably from the carbonatite veins in the nepheline-syenite.
According to the findings of Bruland (1980) and by reason of the general investigations of Lasaga et al. (1994), Froger & Schweda (1998) and Zhou (2010), relatively high dissolution rates are expected of nepheline and significantly lower of K-feldspar.Generally, possible simplified reaction-equations of water-rock interaction (WRI) for waters within the mine are the dissolution of the end-member nepheline and K-feldspar, both forming kaolinite (eq.6, 7).
Both reactions are consuming H + , leading to an increase in the pH value.The theoretical considerations are corresponding with the measurements; all water samples collected in nepheline-syenite rocks within the mine had elevated pH values of up to 9.36 (Table 2B).
The whole-rock analyses of the nepheline-syenite (Table 1) showed Na/K ratios (mol-basis) of about 1.36, whereas in the collected water samples (Table 2B) in most cases Na*/K is roughly about 10 and sometimes even higher (Na* corresponds to Na, reduced by the marine component).Nepheline has a Na/K ratio of about 3.95, whereas in the K-feldspars the Na/K ratio is difficult to determine on account of the observed Na-and K-rich lamella texture.The nepheline-syenite contains about 40% K-feldspar and 35% nepheline.Since the molar mass of K-feldspar at 278.33 g/mol is much higher compared to that of nepheline, 142.05 g/mol, the Na/K ratio of K-feldspar is in the order of 0.85.Thus, in order to get Na*/K = 10 in the water, an additional K-sink is needed, i.e., a precipitation of K-rich mineral(s).However, the required K-sink could not be identified.
Thus, the sodium-potassium ratio is one of the most fundamental relationships in the nepheline-syenite waters.The Na*-K stability diagram (Fig. 9) shows the dependency of the Na*-and K-activity dependent on pH (expressed as H + ) at 15°C.The calculation has been made for two saturation states of quartz (SI Qtz = 0.0, -0.5).Taking into account an assumed undersaturation of the nepheline-syenite waters with respect to quartz by reason of the SiO 2 -poor nepheline, the collected water samples plot into the kaolinite window.Independent of the degree of undersaturation with respect to quartz, even highly developed waters will not plot in the muscovite window (Fig. 9).
The water data can be fitted to one continuous linear trend line with a slope of m = 1 (violet dotted line The influence on water chemistry according to the alteration of other mineral components like biotite, amphibole or apatite is much smaller (e.g., Lasaga et al., 1994).Thus, the magnesium concentration as a result of the alteration of biotite and amphibole is very low (Figs. 4F & 7) and phosphor, with apatite as possible source, is below the detection limit (Table 2B).
By reason of the generally relatively low TDS in all collected water samples, strontium could only be detected in higher concentrated waters, like the carbonatite and the nepheline-syenite waters (Table 2B).
There seems to be a correlation between Ca and Sr (Fig. 8).This correlation is probably caused by the alteration of the primary SrO-rich calcite in the carbonatite and in the carbonatite veins of the nepheline-syenite and calciterich mafic dykes with high strontium concentrations, as described above.The slope of the correlation line in Fig. 8 with m = 0.0256 corresponds to the SrO content of about 2 wt.% in the primary calcite phase, described by Jendryszczyk (2013).Bruland (1980) studied recent near-surface weathering effects as a consequence of precipitation in alteration zones of nepheline-syenites and found that there is no difference in appearance between the alteration products of nepheline and K-feldspar.Dorfman (1958) and Bruland (1980) stated that the alteration of nephelinesyenite is a process of several stages.Bruland (1980), in addition, investigated the whole-rock composition of alteration products from alteration zones in the nepheline-syenite and found illite/hydromuscovite with a K 2 O concentration of up to 9.7%.These alteration products are normally associated with low-grade metamorphism.However, Geis (1979) found that in the alteration zones in the nepheline-syenite, nepheline was replaced by zeolite aggregates.Li (2013) observed may occur, the latter preferentially in olivine-gabbros.Typical for the ultramafic rocks are the high Mg and Fe but low K and SiO 2 contents and that the rocks contain hardly any Ca-rich plagioclase.
Thus, waters collected in gabbros are expected to be either Ca-dominated or Mg-dominated, or a combination of both, the latter depending essentially on the olivine content of the gabbros in the catchment.Waters collected in ultramafic rocks should have -in consequence of the mafic minerals in the rocks -significantly higher Mg and lower Ca concentrations than waters circulating in a gabbro terrain.These common assumptions correspond to the hydrochemical composition of the water samples collected in gabbro-and ultramafic rock catchments.
Fig. 10 shows the strong correlation between TDS* and the (Ca + Mg) concentration in these water samples, whereas waters from gabbro catchments generally show somewhat higher Ca and somewhat lower Mg concentrations than waters from ultramafic rock catchments.Please note that the TDS* is reduced by the marine Na and Cl components, as described above.
in Fig. 9).Thus, waters will develop along this line, a parallel to the metastable extension of the paragonitemuscovite phase boundary, until saturation with respect to muscovite is achieved (paragonite is a sodium mica, with indicative formula NaAl 2 [(OH) 2 AlSi 3 O 10 ], related to muscovite).They will then evolve along this vertical phase boundary also until saturation to paragonite is reached.
Similar studies regarding the incongruent dissolution of feldspars as well as irreversible reactions involving minerals and aqueous solutions were carried out by Helgeson (1974) and Helgeson et al. (1968).

Waters in gabbro and ultramafic rock catchments
Analyses of rock samples have shown, as described above, that the main components of the gabbro samples are Ca-and Mg-rich clinopyroxene and plagioclase with a generally higher An than Ab component.Small amounts of orthopyroxene may also be present.Additionally, minor amounts of Ca-amphibole and Mg-rich olivine  2B) in dependency of pH for 15°C, atmospheric pressure and different saturation states of quartz, calculated with the thermodynamic program MacIIPtax (Brown et al., 1989) on the basis of thermodynamic data of Helgeson et al. (1981), Berman (1988), Liou et al. (1991) and the phase plotter of Lieberman (1991).The two different saturation states of quartz (0.0, -0.5) correspond to the chalcedony saturation states -0.28, -0.78.
plagioclase and its low dissolution rate in comparison to An (e.g., Lasaga et al., 1994), the alteration of the albite component (eq.12) should not influence the water signature in the gabbros to any great extent.In ultramafic rocks, the predominant water-rock interaction (WRI) is expected to be the weathering of olivine (forsterite, eq.11) and clinopyroxene (diopside, eq.9), followed by that of orthopyroxene (enstatite, eq.10).
Possible WRI equations in surface waters (eq.8 and 12 from Stober & Bucher, 1999b) 2010), high dissolution rates are expected from anorthite, followed by forsterite, then enstatite and diopside, and with lower rates probably from albite and amphibole.So, taking also into account the frequency of occurrence of the specific mineral components, the most important and dominant alteration reaction in the gabbros is expected to be that of the anorthite component of plagioclase (eq.8), followed by clinopyroxene (diopside, eq.9) then orthopyroxene (enstatite, eq.10).In olivinerich gabbros the alteration reaction of olivine (eq.11) could dominate the reactions described with eq. 9 and/ or eq.10.Because of the relative low Ab component in W99) had significantly elevated pH-values (9.39, 8.40), demonstrating proton-consuming reactions, since the atmospheric CO 2 inside the mountain is consumed by near-surface WRI reactions.
During the alteration process in gabbros, H 2 O is continuously withdrawn from the water according to eq. 8 (main important alteration reaction) leading to an increase in TDS and, thus, in the long term following Fig. 11 theoretically to the formation of stilbite, a Na/Ca zeolite mineral (occurring in both sodium-and calciumrich forms) binding free H 2 O into the structure.
Water composition in gabbro terrains indicates that dissolution of the major minerals, Ca-rich plagioclase and clinopyroxene (also in some olivine-rich localities These simplified considerations suggest that the Ca concentrations in waters collected in gabbros should generally be higher than the Mg concentrations and that the Na concentrations should be generally very low, whereas in ultramafic rocks relatively high Mg concentrations in the water are to be expected, but with a certain amount of Ca too.That is precisely what is observed in the analysed waters (Figs. 7 & 10).All reactions (eq.8-12) lead to HCO 3 as the dominant anion, as measured in the water samples.However, a significant increase of pH with increasing TDS in consequence of the 'minor important' reactions (eq.9-11) was not detectable, neither in the gabbro nor in the ultramafic rock waters.All surface-water samples showed circumneutral pH.On the other hand, the two subsurface gabbro-water samples from the Øksfjordtunnel (W5,  2C; black dots -brooks, turquoise -talus, pink -tunnel) in dependency of pH for 15°C, atmospheric pressure and different saturation states of quartz, calculated with the thermodynamic program MacIIPtax (Brown et al., 1989) on the basis of thermodynamic data of Helgeson et al. (1981), Berman (1988), Liou et al. (1991) and the phase plotter of Lieberman (1991).The three different saturation states of quartz correspond to the chalcedony saturation states -0.28, -0.58, -0.78.Calcite saturation according to PHREEQC (Parkhurst & Appelo, 1999) waters, the calcium-sodium and the calcium-magnesium relationships are the fundamental ties, whereas in waters circulating in ultramafic rocks the calcium-magnesium relationship is the most important.
The Ca-Na* stability diagram (Fig. 11) shows the dependency of the Ca and Na* activity dependent on pH (expressed as H + ) at 15°C.Taking into account the assumed undersaturation of the gabbro-waters with respect to quartz, the water samples plot into the kaolinite window, depending on the degree of undersaturation.The calculation has been done for several saturation states of quartz (SI Qtz = 0.0, -0.3, -0.7).
The gabbro water data can be fitted to one continuous linear trend (violet dotted line in Fig. 11) with a slope of forsterite), essentially dominates the chemical evolution of the waters and that the reaction with the gabbro rock matrix is the major source of the cations.Accordingly, plagioclase with its An and Ab components is expected to supply Ca and Na and clinopyroxene to contribute Mg and Ca, and in olivine-rich gabbros an additional Mg input is assumed.
In ultramafic rocks, olivine and clinopyroxene (and to some degree orthopyroxene) are expected to be the dominating factors enhancing Mg and Ca in the waters.
Many aspects of the water composition and evolution will be better understood by utilising isobaric and isothermal activity diagrams of the phase relationships in the composition space of particular interest.In gabbro  2C; black dots) and ultramafic rocks (Table 2D; hollow dots) in dependency of pH for 15°C, atmospheric pressure and different saturation states of quartz, calculated with the thermodynamic program MacIIPtax (Brown et al., 1989) on the basis of thermodynamic data of Helgeson et al. (1981), Berman (1988), Liou et al. (1991) and the phase plotter of Lieberman (1991).The three different saturation states of quartz correspond to the chalcedony saturation states -0.28, -0.58, -0.78.Calcite saturation according to PHREEQC (Parkhurst & Appelo, 1999).

Waters in granite and gneiss catchments
Waters collected in granite and gneiss catchments are generally poorly mineralised (Stober, 1995;Frengstad, 2002;Frengstad & Banks, 2007).Thus, the water samples in the investigation area had very low TDS too, especially under the arctic climatic conditions.Waters in gneiss-and granite-dominated terrains belong to the waters having basically the lowest TDS of all sampled waters.Neglecting the marine influence, the water type is dominated by Ca and/or Na* (Na reduced by the marine component) and by HCO 3 (Figs. 4H,5 & 7).Water chemistry in granite and gneiss catchments is mainly influenced by the alteration of plagioclase, whereas the water type is dependent on the An/Ab ratio in the plagioclase (Bucher & Stober, 2000;Banks & Frengstad, 2006).In the gneisses the water type is thus Ca-dominated, whereas in the granites preferentially Na-dominated.

Conclusions
During two research campaigns, water and rock samples were collected in the Seiland Igneous Province, North Norway.The focus of the sampling was on silicate rocks which are easily weathered such as nepheline-syenites, carbonatites, ultramafic rocks and gabbros.Fieldwork comprised a detailed field study of selected outcrops.Care was taken to collect water samples in defined catchments with the same lithology.For comparison reasons, also water samples in granites and gneisses were collected.Precipitation and sea-water was also collected as reference material for the field study in view of possible influences in consequence of rainfall and sea spray.Field observations, geochemical compositions of the different rock samples, and chemical analyses of the collected water samples are used to model the interaction of water (rainwater) with unstable (easily) weatherable silicate rocks, to better understand silicate weathering processes in general.
In this paper, the hydrochemistry in different crystalline basement rocks of the Seiland Igneous Province were studied.The focus was on relatively fast-weathering rock processes visible in the hydrochemical properties of surface water and groundwater.Thus, first of all the mineralogical and geochemical compositions of the different rock types were investigated: nephelinesyenite, carbonatite, gabbro and ultramafic rocks.Knowing the different components of each rock-type, the main components (minerals) responsible for the water chemistry could be identified.The hydrochemical properties of the diverse (rock-specific) catchments differ from each other.All collected waters had relatively low TDS, between 7 mg L -1 and 118 mg L -1 .To carry out further hydrochemical investigations, the marine m = 2, given as a result of the alteration of plagioclase (An and Ab components).Thus, waters in a pure plagioclase terrain will develop along the An-Ab line until saturation with respect to stilbite is reached, and then follow along this stable phase-boundary until saturation with respect to paragonite or -depending on the saturation state of quartz -Na-beidellite is reached.Expressed as X An (Ca/ (Ca + Na)), the gabbro waters are close to X An = 0.70 and thus in between An 50 and An 80 on an activity basis.The best fit is achieved for an assumed saturation state of quartz SI Qtz = -0.3 (Fig. 11).Thus, saturation with respect to calcite will not be reached (Fig. 11).All water samples plot in the kaolinite field, with the exception of two samples collected in the Øksfjordtunnel which have different boundary conditions.
Equivalent investigations with the Mg and Ca activity (dependent on pH, expressed as H + ) at 15°C show a strong relationship between calcium and magnesium for waters in gabbro and in ultramafic rocks, respectively (Fig. 12).The calculation has been done for different saturation states of quartz: SI Qtz = 0.0, SI Qtz = -0.3 and SI Qtz = -0.7.The data of the gabbro and the ultramafic waters follow a trend-line with a slope m = 1 in the Mg-Ca stability diagram (violet dotted lines in Fig. 12).Taking into account the assumed undersaturation of the waters with respect to quartz (SI Qtz ≈ -0.3), all collected water samples -with exception of the two gabbrosamples collected in the Øksfjordtunnel--plot into the kaolinite field of this diagram (Fig. 12).Waters in a pure clinopyroxene rock will develop along the Di-En line until saturation with respect to stilbite is reached, and then follow along this stable phase-boundary until they are saturated with respect to chlorite.Assuming SI Qtz = -0.3,saturation with respect to calcite will not be achieved.Expressed as diopside (CaMg[Si 2 O 6 ]) and enstatite (Mg 2 Si 2 O 6 ) components, all gabbro and ultramafic rock waters lie in a small clinopyroxenecontrolled band between Di 70 En 20 and Di 90 En 1 (orange dotted lines in Fig. 12), whereas the Di/En ratio in gabbro waters is higher than in ultramafic rock waters.
Thus, olivine (i.e., forsterite), respectively orthopyroxene, does not control the water chemistry in the gabbros, respectively in the ultramafic rocks, to any significant extent, because otherwise the data would follow a trendline with a slope of m = 2, getting saturated with respect to chlorite.Consequently, in both catchments magnesium is derived mainly from clinopyroxene.
Thus, plagioclase and clinopyroxene control the hydro chemistry in gabbro-dominated terrains and in ultramafic rocks clinopyroxene is controlling the chemistry of the waters.
Oregon).The waters showed TDS values even up to 1.500 mg L -1 with Mg and HCO 3 as dominant ions and a pH up to 8.9.
The investigations of Bucher et al. (2015) in surface waters of the Ronda peridotite, southern Spain, also showed relatively high TDS (275-563 mg L -1 ) and elevated pH values (8.36-8.54).In all waters Mg and HCO 3 were also the dominant ions.The hydrochemistry of the waters develops when atmospheric carbon dioxide dissolves in the water and forms bicarbonate, as olivine and serpentine minerals in the near-surface react with the waters and release magnesium ions and silica.Zhou (2010), on the other hand, carried out investigations in brook waters in serpentinite and peridotite catchments in the high alpine area of Zermatt-Saas, Switzerland.The climatic conditions there are more or less similar to those of our area in northern Norway.The waters that Zhou (2010) collected are very poorly mineralised with average TDS of 21 mg L -1 .Her water samples also showed Mg and HCO 3 as the dominating ions, and pH values are in most cases slightly elevated.She found that with increasing TDS SiO 2 , Na and K are progressively increasing and assumed that these waters are originating predominantly from Mg-silicates as exposed in rocks in the catchment.Drever & Zobrist (1992) studied indirectly the climatic effect on the weathering process and surface-water composition by investigating the weathering process as a function of elevation in the southern Swiss Alps.They found a more or less exponential decrease of the weathering effect on water chemistry with increasing elevation.Their findings can explain the generally surprisingly low TDS of the water samples in relatively fast-weathering rocks of the Seiland Igneous Province in North Norway under arctic climatic conditions.
Further investigations will focus on deep-seated waters, from tunnels and wells, especially in gabbro-and ultramafic rock-dominated terrains.These waters should have longer residence times in the subsurface and are thus more highly developed than surface waters.It is planned to compare the hydrochemistry of these waters with the results presented in the stability diagrams.Thus, we also intend to sample alteration products of the various rock types.In addition, a detailed isotopic investigation of groundwater is planned in order to provide information concerning the age and origin of the collected waters.
influence in the water samples, represented by sea spray or rainfall, had to be eliminated.
All waters are undersaturated with respect to quartz.
Water chemistry in a carbonatite-dominated terrain is regulated by the weathering process of calcite.In nepheline-syenite, the sodium-potassium ratio is one of the fundamental relationships, because of the nepheline and K-feldspar components of nepheline-syenite.
The chemical development of rainwater in different lithologies could be shown and explained with the aid of mineral stability diagrams.
The investigations of Gascoyne & Kamineni (1993) on groundwater in a gabbro/anorthosite complex in Massey, Ontario (Canada), showed a Ca-Na-HCO 3 water type with elevated Mg concentrations, a slightly elevated pH value and TDS of 150 mg L -1 .The authors could show that the waters plot into the stability field of kaolinite, as do the waters analysed in this investigation.With increasing depth, i.e., prolonged residence time, Gascoyne & Kamineni (1993) observed that the water type changed into a Na-HCO 3 water with higher TDS and pH values.
Since the residence times of the investigated surface waters and shallow groundwaters are much shorter, the TDS in the Seiland gabbro waters are less and more strongly dominated by Ca and Mg.On the other hand, with increasing residence time and contact area for water on rock-surfaces, i.e., waters in talus material or in tunnels, these waters showed slightly elevated TDS and pH.Thus, the investigations of Gascoyne & Kamineni (1993) are in good agreement with our results.
In contrast to our findings in North Norway, Marques et al. (2008) presented hydrochemical properties of surface water with much higher TDS (850 mg L -1 ).The investigated brook is flowing through ultramafic rocks, dunites, in Central Portugal.While the dominating ions are the same as in the ultramafic rock catchments of Seiland, Mg and HCO 3 , the brook water in Central Portugal has a significantly elevated pH.The pH values of our samples in ultramafic rocks are circumneutral (6.65-7.69),whereas the waters sampled in the ultramafic talus material showed a pH in the upper portion of this range.Concerning the general setting in the investigated area of Marques et al. (2008), with a lot of upwelling highly concentrated waters from diverse depths in the entire area and near the brook, it was assumed that the chemistry of the brook water might be influenced by nearby upwelling springs, mixing with the brook water.Quite possibly the more favourable climatic conditions for weathering in Central Portugal have helped to intensify the alteration process and thus increase TDS (and pH) in the water.
Equivalent investigations were carried out by Barnes et al. (1967) and Barnes & O'Neil (1969) on surface waters and shallow groundwaters in serpentinites, peridotites and dunites in the western United States (California,

Figure 2 .
Figure 2. Geological map of the Seiland IgneousProvince showing all water and selected rock sample sites(after Roberts, 1973).

Figure 5 .
Figure 5. Piper-diagram of all collected water samples showing various subgroups.Anions and cations presented in eq.%.

Figure 7 .
Figure 7. Relative cation distribution in the water samples.Na is reduced by the marine component (Na*).

Figure 8 .
Figure 8. Correlation between Ca and Sr in carbonatite (black dots) and nepheline-syenite (red dots -within the mine, light red dotspring waters).Detection limit of Sr is 50 µg L -1 .

Figure 10 .
Figure 10.TDS* vs. the calcium concentration, magnesium concentration and the sum of both (Ca + Mg) in gabbro (filled symbols) and ultramafic rock catchments (hollow symbols).TDS* is reduced by the marine component (Na, Cl).

Table 1 .
Major and selected trace elements of rocks in the Seiland Igneous Province.

Table 2A .
Hydrochemical analyses of sea-water and rainfall samples.For standard ocean water see: www.lenntech.com/composition-seawater.htm.

Table 2B .
Hydrochemical analyses of water samples collected in the nepheline-syenite mine and in nearby carbonatite-dominated terrain on the island Stjernøya of the Seiland Igneous Province.

Table 2C .
Hydrochemical analyses of water samples collected in the gabbro-dominated terrain of the Seiland Igneous Province.

Table 2E .
(continuation).Hydrochemical analyses of water samples collected in the gabbro-dominated terrain of the Seiland Intrusive Complex.
. High EN values were predominantly calculated for very low TDS waters, because small analytical errors may lead to high EN values.With the exception of just two samples, all collected and analysed samples were used for this paper.are gabbros and samples 1110 and 1113 are ultramafic rocks, because they contain no or almost no plagioclase (Plg).The gabbro samples 1246, 1046, 1028, 1054, 1281, 1115 and 1245 could be characterised as olivine-gabbros, because they contain the minerals plagioclase (Plg), pyroxene (Px) and olivine (Ol).Gabbro sample 1027 could be classified as anorthosite because of its very high Plg-content and sample 1224 as troctolite, showing a very low Px content.In the gabbros, all calculated plagioclases show very high An/(An + Ab) ratios (55-79 vol.%).In contrast to the gabbros, the ultramafic rock samples 1110 and 1113 are composed mainly of olivine (Ol) and pyroxene (Px) and thus could be characterised as peridotites, showing very high MgO contents in the XRF analyses (Table

Table 1
Si 2 O 5 (OH) 4 -kaolinite, SiO 2eq -silica in waterThe mineralogical composition of the rock samples has shown that the plagioclases in the gabbros are generally Ca-rich, i.e., the An component (CaAl 2 Si 2 O 8 ) is relatively high (Ca/(Ca + Na) Pgl = 55 to 100 vol.%).The pyroxenes in the gabbros could be classified as clinopyroxenes ((Mg, Fe) Ca Si 2 O 6 ) and in the ultramafic rocks additionally as orthopyroxenes.The cation proportions (expressed as