Processes Governing Alkaline Groundwater Chemistry within a Fractured Rock ( Ophiolitic Mélange ) Aquifer Underlying a Seasonally Inhabited Headwater Area in the Alada L lar Range ( Adana , Turkey )

The aim of this study was to investigate natural and anthropogenic processes governing the chemical composition of alkaline groundwater within a fractured rock (ophioliticmélange) aquifer underlying a seasonally inhabited headwater area in the Aladağlar Range (Adana, Turkey). In this aquifer, spatiotemporal patterns of groundwater flow and chemistry were investigated during dry (October 2011) and wet (May 2012) seasons utilizing 25 shallow hand-dug wells. In addition, representative samples of snow, rock, and soil were collected and analyzed to constrain the PHREEQC inverse geochemical models used for simulating waterrock interaction (WRI) processes. Hydrochemistry of the aquifer shows a strong interseasonal variability where Mg–HCO3 and Mg–Ca–HCO3 water types are prevalent, reflecting the influence of ophiolitic and carbonate rocks on local groundwater chemistry. R-mode factor analysis of hydrochemical data hints at geochemical processes taking place in the groundwater system, that is, WRI involving Caand Si-bearing phases; WRI involving amorphous oxyhydroxides and clay minerals; WRI involving Mg-bearing phases; and atmospheric/anthropogenic inputs. Results from the PHREEQCmodeling suggested that hydrogeochemical evolution is governed by weathering of primary minerals (calcite, chrysotile, forsterite, and chromite), precipitation of secondary minerals (dolomite, quartz, clinochlore, and Fe/Cr oxides), atmospheric/anthropogenic inputs (halite), and seasonal dilution from recharge.


Introduction
Achievement of a sustainable aquifer management requires an improved understanding of the complex natural processes generating the observed composition of groundwater, as well as all anthropogenic activities hindering its safe use and availability [1].This is critically important, especially in headwater areas, since they typically constitute 70-80% of the total catchment area [2] and represent starting point of the terrestrial water cycle [3].Our understanding of the mountainous headwater systems and the impacts of anthropogenic activities on headwater-scale has been largely impeded by their small size, large numbers, remote locations, rugged terrain, harsh climate conditions, and lack of road access, logistics, and available data [4][5][6][7].Despite their importance, in the literature, there is no clear definition as to what constitutes a "headwater area" [4,8].It is generally agreed upon that these areas are unique and fragile recharge environments near the topographical drainage divides where flow lines of zeroto first-order catchments originate [7,9].Yet, because of the problem of scale dependency, most of these low-order stream channels are rarely documented on the topographic maps [4,10]; hence, they are frequently omitted from the ordering schemes (e.g., [11,12]).In reality, these montane headwater systems serve as the transport medium for delivering water, sediment, nutrients, and other materials to downstream areas, especially during intermittent rainfall and snowmelt events [4,6,10].Recently, Bishop et al. [5] called aptly this understudied and ignored realm as "Aqua Incognita," the unknown waters.A number of studies (e.g., [2,6,10,[13][14][15]) have shown that hydrological and hydrogeochemical processes occurring in headwater systems have critical control on the quantity and biophysicochemical quality of the underlying shallow groundwater and downstream systems, all of which are intimately linked through the hydrologic cycle.Furthermore, these areas and associated hyporheic zones have also important ecosystem functions, providing unique habitats for diverse flora, fauna, and microbiota [16][17][18] that are imperative for a fully functioning system.
During the last several decades, relatively poorly developed and remote highlands of the Aladaglar Range of eastern Taurides have become increasingly valued for their climatological and bioecological diversity (e.g., [19][20][21]), nearpristine water and air quality (e.g., [21,22]), scenic and aesthetic beauty (e.g., [20]), and recreational/touristic opportunities (e.g., [23]).Long before the recent appreciation of all these natural treasures/qualities, these highland areas (called yayla) were occupied essentially as summer camping grounds by the nomadic people (called Yörük) who commonly made their living by livestock rearing (primarily goat) and to a lesser extent small-scale family farming [24,25].While the pure nomadic lifestyle is still alive in some areas, currently, yaylas are mostly frequented by the city dwellers, especially during the summer season, due to their comfortable climate (e.g., cool and less humid) as compared to the Mediterranean coastal zone (i.e., C ¸ukurova region) [19,21].As a response to this new trend, numerous seasonal settlements were created in the headwater areas, which in turn have not only significantly altered demographic, cultural, and socioeconomic characteristics of the region [21] but also had a marked impact on the natural environment [23,24,[26][27][28].
Recent modeling studies conducted in the Seyhan River basin also raised concerns over the anthropogenic climate change, which is projected to aggravate the pressure on the hydrologic system in the forthcoming decades [29][30][31].This paper presents the first detailed analysis of hydrological and hydrochemical data obtained from two snapshot sampling campaigns carried out in the Kızılgedik seasonal settlement, which is located in a serpentinized ophiolitic terrain in the headwaters of the Seyhan River basin.The specific objectives of the present study were (i) to define the mineralogy and geochemistry of the rocks and soils found in the area; (ii) to determine water levels and groundwater flow directions in the ophiolitic complex aquifer; (iii) to investigate possible effects of anthropogenic inputs to the underlying shallow aquifer; and (iv) to shed light on the predominant hydrogeochemical processes using inverse geochemical modeling approach.

Study Area
2.1.Physiographic Setting, Climate, and Land Use.The study area, located ∼100 km north of the Mediterranean Sea coastline in Adana province (Turkey), lies within the Aladaglar Range of eastern Taurides (Figures 1(a) and 1(b)) and is a part of the Seyhan River basin (area 21,700 km 2 ).The specific area studied is bounded by the latitudes 37 ∘ 31  55.50  N and 37 ∘ 32  28.70  N, and longitudes 35 ∘ 25  10.75  E and 35 ∘ 25  52.51  E. This area encompasses a rugged mountainous terrain, with altitudes ranging from 1030 to 1310 m above mean sea level (msl), and it is characterized by topographic gradients between 0.13 ∘ and 45.9 ∘ (with a mean slope of 16.6 ∘ and E-SE aspect).The climate is continental to some extent [19] and influenced by both the Mediterranean and central Anatolian weather systems, bringing temperate, dry summers, and cold, wet winters to the area [20].Based on the available climate data  recorded at the Pozantı meteorological station (see Figure 1(b)), the average annual air temperature is 13.5 ∘ C and temperatures occasionally exceed 31 ∘ C in summer and rarely drop below −6 ∘ C in winter [32].The area receives an average annual precipitation slightly higher than 725 mm and 85% of it occurs between November and May [32].The precipitation occurs in winter, as rain and snowfall, but in summer as occasional thunderstorms.
In Kızılgedik site, there are 85 individual houses accommodating some 300 people during the peak season (June to September).However, population remains insignificant during the rest of the year (i.e., off-season).Currently, the area does not have a sewerage network and each property has its own cesspit in the garden.Traditionally, cesspits are built square in form (dimensions: 2 m × 2 m × 1.5 m) and lined with loose-fitting stones allowing wastewater to percolate into the ground (Figure 2).There was no piped water supply until 2011, where majority of the residents still rely on large-diameter hand-dug wells (HDWs) (Figure 2) for their domestic water needs and irrigation.Typical of serpentinitic terrains, the natural vegetation in the settlement area is limited to sparse shrubs and herbaceous vegetation (a.k.a.serpentine barrens [35]), whereas domesticated plants and trees are mainly found around residential houses.Additionally, patches of mixed conifer forests (e.g., pine, juniper, larch, fir, and cedar) are often found in the hills surrounding the settlement area.

Soil Not to scale
In the study area, Cenozoic sedimentation begins with the Oligocene-Late Miocene Karsantı formation (terrestrial and lacustrine pebbly sandstone, mudstone, coaly claystone, and marl) [43].This formation is separated from the underlying Mesozoic tectonostratigraphic units by a distinct unconformity [53].This nonmarine deposition ended in the Early Miocene by a transgression from the Adana basin [54].The Early-Middle Miocene (Aquitanian-Burdigalian) Kaplankaya formation (shallow marine marl, claystone, sandstone, and sandy limestone) [45] records the first marine transgression in the Karsantı basin [37]; the base of this unit also lies above an irregular unconformity surface [55].Kaplankaya formation is partly overlain by and passes laterally into the Early-Middle Miocene (Burdigalian-Serravallian) Karaisalı formation (fossiliferous reefal limestone) [43,56].Karaisalı formation occupies topographically higher parts of the study area (e.g., Korum Mountain) and is highly susceptible to karstification, as evidenced by the prevalence of sinkholes.A detailed synthesis of both regional and local tectonic framework and evolution of the Karsantı basin can be found in [34,38].
Hydrologically, the study area resides within several zeroorder catchments that lie at the ultimate extremes of the local drainage network, where overland flow is only seen after heavy rainfall events and during snowmelt episodes in illdefined surface flow paths (e.g., rills, gullies, and swales).The aquifers within the study area can be classified into two primary groups based on host rock and structural characteristics, as follows: (i) carbonate rock aquifer and (ii) ophiolitic complex aquifer.Groundwater occurrence within the carbonate rocks is not known due to absence of monitoring wells, but secondary porosity created by fractures/faults, together with karstic features such as sinkholes, may allow significant groundwater circulation and enhanced recharge.Groundwater found within the ophiolitic complex aquifer is of utmost importance for the livelihood of the headwater environment and local residents, although ophiolitic rocks of this region have been considered impermeable in earlier studies [57,58].In the study area, the ophiolitic complex aquifer is compartmentalized by a distinct set of faults, trending in SW-NE direction.In this aquifer, groundwater is tapped from the highly fractured portion of the ophiolitic mélange by shallow HDWs with depths not exceeding 10 m.In the study area, recharge to the aquifers takes place through several ways, such as (i) infiltration from the runoff from precipitation and snowmelt events; (ii) lateral and downwards groundwater flow from the overlying geological formations (mostly karstic in nature); (iii) infiltration from irrigation water; and (iv) wastewater percolation from the cesspits.The main recharge areas are positioned to the N and SW of the study area (Figures 3(a) and 3(b)).Water levels in the ophiolitic complex aquifer respond relatively quickly to the recharge events, due to highly fractured nature of the upper portions of the ophiolitic mélange.The discharge from the aquifers occurs in different ways, including (i) subsurface outflow to adjacent valleys that moves through

Rock/Soil Sampling and Analytical Methods.
A rock and soil sampling campaign was carried out in November 2013 in order to relate the groundwater chemistry with lithology.Selection of the sampling sites (see Figure 4) was largely based on spatial distribution of the major geological units and field observations made during prior on-site surveys.Fresh rock samples ( = 10) were collected at various locations (i.e., R1-R10), generally as composite chip samples (∼1 kg) from available outcrops with a rock hammer.Topsoil composite samples ( = 8) were collected at a depth of 0-15 cm with a stainless steel spatula, after removing stones, plant/root debris, and foreign objects.At each soil sampling site (i.e., S1-S8), representative composite samples were obtained by pooling four subsamples (∼250 g) taken on the corners of a 1 m 2 square [59].All samples were placed in labeled selflocking polyethylene bags and transferred to the laboratory for further processing.In the laboratory, the rock and soil samples were air-dried at room temperature for several days, disintegrated and homogenized in an agate mortar and then passed through a 2 mm sieve.Samples for X-ray diffraction (XRD), wavelength dispersive X-ray fluorescence (WDXRF), and loss-on-ignition (LOI) analyses were prepared by the usual powder method procedures as described by Buhrke et al. [60].Rock and soil samples were ground to powder and homogenized by RS 200 tungsten carbide vibratory disc mill (Retsch, Germany) and then fine grinding was accomplished using an agate mortar and pestle.
The main mineral phases occurring in the rock and soil samples were characterized by powder XRD technique using a Rigaku SmartLab X-ray diffractometer (Rigaku Corporation, Japan) with Cu K radiation at an accelerating voltage of 40 kV and a tube current of 30 mA.XRD patterns in diffractograms were obtained from 5 ∘ to 60 ∘ in 2 with a step width of 0.02 ∘ , at a scanning speed of 4 ∘ min -1 , using 1 mm receiving slits, a 10 mm length limiting slit, and a 2/3 ∘ incident slit.The software PDXL and the PDF-2 database (http://www.icdd.com)were employed for mineral phase identification.The chemical composition of the rock and soil samples were determined by a Rigaku ZSX Primus II WDXRF spectrometer (Rigaku Corporation, Japan) with a 4 kW rhodium target, using an acceleration voltage of 30 kV and a current of 100 mA.The major oxides (SiO 2 , TiO 2 , Al 2 O 3 , Fe 2 O 3 , MnO, MgO, CaO, Na 2 O, K 2 O, P 2 O 5 , and SO 3 ) and trace elements (Co, Cr, Ni, and Sr) in bulk solids were quantified using the standardless analysis program SQX [61].
WDXRF analyses were carried out on pressed-powder pellets that were prepared by thoroughly mixing 10 g of each sample with 4 g of cellulose binder (SPEX SamplePrep PrepAid5, USA) with a particle size of ≤20 m.The mixture was pressed into 38 mm diameter pellets using a manually operated hydraulic press.After pressing, the pellets were dried in oven at 100 ∘ C for 12 h, before the WDXRF analysis.The LOI was determined as the weight loss percentage after burning 4 g of powdered dry sample in an electric muffle furnace at 950 ∘ C for 1 h [62,63].All analyses were performed at the Advanced Technology Education, Research and Application Center (ME İTAM), Mersin University (Turkey).

Groundwater Level Measurement.
The groundwater sample collection and water level measurements were completed within two days, in two separate field campaigns, covering all the wells.The field campaigns took place in October 2011 and May 2012.For convenience, the terms "dry season" and "wet season" will be used throughout the rest of the paper to refer to measurements/sampling made on shallow hand-dug wells (HDWs) during October 2011 and May 2012, respectively.The wells found in the area are typically largediameter (ca.0.8-1.2m) HDWs ranging in depths from 3.29 to 9.54 m.All the wells are directly completed in the highly fractured upper portion of the Kızılcadag ophiolitic mélange and none was identified to have a casing or lining within the saturated zone (Figure 2).These relatively shallow wells are generally equipped with hand pumps and exploited for domestic purposes and/or irrigation water supply, chiefly during summer months.The depth to water in the HDWs ( = 25) was determined manually by means of a flat tape water level sounder (Akım Hydrometry, Turkey) with a precision of 1 mm.Water table elevations (with respect to msl) were calculated in a Geographic Information System (GIS) environment by subtracting depth to water measurements from the topographic elevations obtained from the digital elevation model (DEM with a grid size of 10 m).Groundwater level maps were created by employing the ordinary kriging spatial interpolation algorithm available in the Geostatistical Analyst extension of the ArcGIS 9.3.1 software [64].

Water Sampling and Analytical Methods.
Groundwater samples were collected from identical HDWs ( = 25) in dry and wet seasons (Figure 4).In this study, well purging was not attempted due to presence of large volume of water in the HDWs.Groundwater samples were collected from a depth of a few meters below the water table by lowering a plastic bailer into the HDWs.Additionally, snow samples were collected shortly after two major snowfall events on January 2012 ( = 6) and February 2013 ( = 5) from various locations of the site for physicochemical characterization of the precipitation.Sampling and analytical techniques followed the suggestions by APHA-AWWA-WEF [65] and were similar to those described earlier in the literature [1,66].The field parameters (pH, redox potential (Eh), dissolved oxygen (DO), electrical conductivity (EC), and temperature) were measured in situ using a WTW Multi 340i/SET multiparameter instrument (Wissenschaftlich-Technische Werkstätten, Germany).The probes were calibrated daily in the field using standard procedures before sampling as per manufacturer's instructions.Groundwater samples were immediately filtered on site through a disposable nylon membrane syringe filter (Econofilter) with a pore size of 0.45 m (Agilent Technologies, Germany).In brief, at each site, two 250 mL aliquots were collected in clean HDPE bottles for cation and anion analyses.Aliquots taken for cation analysis were acidified at the field (below pH 2.0) with 65% extra pure HNO 3 (Merck, Germany) to prevent biological activity and precipitation of cationic species.All the samples were stored in an icebox containing gel-filled ice packs to prevent possible evaporation effects.Then, they were transported to the laboratory and refrigerated at 4 ∘ C until analysis.
Analyses for total concentrations of five major elements (Ca, Mg, Na, K, and Si) and 17 trace elements (Al, As, B, Ba, Br, Co, Cr, Cu, Fe, Li, Mn, Mo, Ni, Sr, Ti, V, and Zn) in the acidified aliquots were carried out in the ME İTAM, Mersin University (Turkey) by Agilent 7500ce ICP-MS (Agilent Technologies, Japan) equipped with Octopole Reaction System.The purity of argon gas used in the ICP-MS was 99.998% or higher.The external standard calibration method was applied to all determinations using 6 Li, 45 Sc, 72 Ge, 89 Y, 115 In, 159 Tb, and 209 Bi internal standard mix.Five-point calibration curves were created by analyzing NIST single-element reference standards prepared by serial dilution of stock solutions.The concentrations of ammonia (NH 4 + ), nitrate (NO 3 − ), nitrite (NO 2 − ), sulfate (SO 4 2− ), orthophosphate (PO 4 3− ), chloride (Cl − ), and fluoride (F − ) in the unacidified aliquots were determined at the Mersin University Geological Engineering Department with Hach Lange DR 2800 spectrophotometer (Hach Lange GmbH, Germany).Carbonate (CO 3 2− ) and bicarbonate (HCO 3 − ) in water samples were determined by volumetric titration with 0.01N standard H 2 SO 4 using phenolphthalein and methyl orange indicator solutions, respectively.The ultrapure water (obtained from the ELGA Purelab UHQ system; Veolia Water Solutions, UK) used in the analytical processes had a resistivity of 18.2 MΩ cm at room temperature.The accuracy of the analytical results was estimated by calculating the percent charge balance errors (%CBE), as described by Freeze and Cherry [67].Calculated %CBE average −0.55 for the dry season dataset and −0.30 for the wet season dataset, with standard deviations of 0.86 and 0.40, respectively.No samples in the database have a %CBE greater than ±2.31.

Statistical Analysis and Data
Processing.The water chemistry data were subjected to basic and multivariate statistical analyses utilizing the open source statistical software R ver.3.1.2[68].The basic statistical analyses include descriptive statistics (minimum, maximum, mean, median, and standard deviation), Pearson product-moment correlation coefficient (), and Kolmogorov-Smirnov (K-S) test.K-S test [69,70] was used to assess normality of data variables.R-mode factor analysis (R-mFA) was employed for the multivariate statistical analysis of the water chemistry data.R-mFA can help in extraction of hidden information on the factors controlling groundwater chemistry, by only retaining the key components of the dataset.As a data reduction technique, R-mFA reduces a large number of variables to a minimum number of uncorrelated (i.e., orthogonal) new variables called factors by linearly combining measurements made on the original variables [71].Only normalized and standardized variables were utilized in the R-mFA as suggested by Güler et al. [72].In R-mFA, rotation of factors was carried out using the "varimax raw" method, where Kaiser criterion [73] was utilized to determine the number of factors.Detailed technical description of R-mFA technique and best practices can be found elsewhere [71,[74][75][76][77].
The GIS spatial database used in this study was created using (i) 1 : 25,000-scale geological maps published by Alan et al. [36]; (ii) 1 : 25,000-scale topographic map sheet (Adana M34c3) published by Turkish Ministry of National Defense; (iii) high-resolution (2.44 m) QuickBird satellite imagery acquired in 2012; and (iv) geographic coordinate measurements made during on-site surveys using a Magellan Triton 2000 GPS unit.The spatial data layers were georeferenced within GIS environment using the WGS84 datum (UTM Zone 36N), then integrated, manipulated, analyzed, and visualized using ArcGIS 9.3.1 software and its extensions, namely, 3D Analyst, Geostatistical Analyst, and Spatial Analyst [64].

Geochemical Modeling.
The geochemical code PHREEQC Interactive ver.3.1.4[78] was used for determination of aqueous speciation and saturation indices, as well as for performing inverse modeling calculations related to representative end-member water types.The saturation index (SI) of a mineral phase is defined using where  is the ion activity product for a given mineral phase and   is the equilibrium constant of its solubility product at temperature .The SI parameter describes three saturation states.These are (i) undersaturated (SI < 0), (ii) in equilibrium (SI = 0), and (iii) supersaturated (SI > 0) states.All geochemical calculations and water-rock interaction (WRI) modeling were performed using the Lawrence Livermore National Laboratory thermodynamic database, that is, LLNL.dat [79].

Results and Discussion
4.1.Rock/Soil Mineralogy and Geochemistry.The mineralogical and chemical composition of rocks and soils found in the study area may imprint a unique character to the regional groundwater and will be used to constrain the selection of mineral phases that will be utilized in WRI modeling [80].
Results from XRD analyses (Table 1) were compared with the results from WDXRF analyses (Table 2) to provide a reliable characterization of the mineral phases in the rock and soil samples.XRD analysis of serpentinite rocks making up the Kızılcadag ophiolitic mélange (i.e., samples R1, R2, and R3) revealed that lizardite is the dominant mineral phase with trace amounts of antigorite, olivine, chromite, calcite, and phlogopite (Table 1), whereas XRD analysis of exotic blocks (i.e., samples R7, R8, and R9) dispersed in the ophiolitic mélange shows the presence of four predominant mineral phases, such as lizardite, quartz, calcite, and dolomite (Table 1).These exotic blocks (i.e., limestone, siltstone, and sandstone) are also associated with minor and trace amounts of secondary mineral phases, such as hematite, ankerite, magnesite, dickite, and vermiculite (Table 1).As reflected in XRD results, carbonate rocks of the Early-Middle Miocene Kaplankaya and Karaisalı formations (samples R4 and R5, resp.) are composed almost entirely of calcite, whereas those of Late Triassic-Early Jurassic Etekli formation (sample R6) are composed predominantly of calcite, with trace amounts of dolomite.Chrysotile (i.e., fibrous asbestos) is the most common mineral phase found in veins and shear zones crosscutting the serpentinite rocks (Figure 5), along with dolomite and trace amounts of clinochlore (i.e., sample R10).
In the study area, dolomite mostly occurs as white to pink veins which show fracture-seal texture (Figure 5).Note that sampling of chrysotile veins was intentionally avoided due to hazardous nature of this mineral; therefore, chrysotile was not detected in the XRD analysis.b Lithology refers to the principal geological formation exposed at the surface in the sampling site (see Figure 3 for geological formation descriptions).c Source refers to the principal rock type or parent material occurring in the sampling site.
The soils found in the area are generally shallow in depth (ca.0-30 cm) and discontinuous and mostly direct weathering product of the serpentinite and carbonate rocks underneath.The mineralogy of serpentinitic soils (samples S1-S7), identified by XRD, was dominated by mineral phases such as lizardite and antigorite (Table 1).In some samples, in addition to these mineral phases, secondary phases such as quartz, hematite, magnesite, kaolinite, vermiculite, clinochlore, and diopside were present (Table 1).The presence of vermiculite and lack of smectites in the upper parts of the soil profiles indicate that soils are well drained and have been formed under temperate climate conditions [81].Additionally, presence of expansive clays, such as vermiculite, in the soil matrix implies high level of cation exchange capacity [82,83] of serpentinitic soils.On the other hand, a soil sample (i.e., S8) taken from an area overlying an exotic block found within the Kızılcadag ophiolitic mélange shows the presence of dolomite as the main phase, with lesser amounts of calcite and quartz.
The samples R4, R5, R6, and R7 (all limestone formations) composed almost entirely of calcite (Table 1) are found to be low in SiO  2).The distinct differences in Cr and Ni concentrations between soils can be taken as an indication of differences in degree of weathering and/or mineralogical compositions of the parent rocks [83,86].Cr most commonly occurs as an accessory mineral (e.g., chromite) in serpentinites, whereas Ni primarily exists as impurity on the crystal structure of mineral phases in serpentine [87,88].On the other hand, S8 sample, determined by XRD to contain dolomite, calcite, and quartz (Table 1), shows high concentrations of CaO, MgO, and SiO 2 (Table 2), confirming the presence of these mineral Lithology refers to the principal geological formation exposed at the surface in the sampling site (see Figure 3 for geological formation descriptions).b Source refers to the principal rock type or parent material occurring in the sampling site.c LOI = loss-on-ignition (reported in wt.%); bdl = below detection limit.
phases.The concentrations of oxides such as Na 2 O, K 2 O, P 2 O 5 , and SO 3 are generally very low (<0.10 wt.%) in all types of rock and soil samples (Table 2), reflecting the nutrientpoor character of the serpentinitic terrain.The results from XRD and WDXRF analyses suggest that occurrence of mineral phases and elevated concentrations of some elements in the soils of the study area is mostly due to geogenic sources and representative of the geological formations occurring in the area.

Groundwater Levels and Flow Directions.
Water level maps for dry season and wet season are presented in Figures 4(a) and 4(b), where average depths to groundwater were 4.83 and 2.79 m below ground surface (bgs), respectively.The decrease in average depth to water (2.04 m bgs) between dry season and wet season can be attributed to increased recharge through precipitation, as well as snowmelt, and insignificant amount of groundwater extraction during the off-season (September to May).Even though the groundwater levels in individual HDWs show significant fluctuations (from 0.74 to 5.31 m) between dry season and wet season, no discernible spatiotemporal variations were observed on groundwater flow directions and gradients (Figures 4(a) and 4(b)).Equipotential maps constructed for the ophiolitic complex aquifer indicate the direction of the groundwater movement to be mainly from N to NE/SE near well K2 and from SW to NE near well K6 (Figures 4(a ).An interesting feature in the area is the depression cone formed around well K14, where most of the upgradient flow appears to be directed towards the depression (during both dry and wet seasons), even though no heavy pumping of groundwater ever occurred in or near this well.The depth to water in well K14 was recorded as 5.22 and 3.49 m bgs in dry season and wet season, respectively.Well K14 is located near (∼60 m) a major fault zone juxtaposing Kızılcadag ophiolitic mélange and carbonate rocks of Etekli formation (Figures 4(a) and 4(b)), where open fractures and karstic features developed within these units might have collectively provided highly conductive pathways for subsurface outflow underneath the adjacent dry valley.Interestingly, the same fault zone acts as a flow barrier at SW part of the study area, as evidenced by groundwater flow direction that is aligned parallel to the fault zone, displaying a combined conduitbarrier behavior (e.g., see [89]).

Hydrochemical Characteristics of the Water Samples.
The summary statistics of the seasonal physicochemical composition of groundwater and snow samples are presented in Table 3 and complete dataset is provided as supporting information in Table A.1 (Supplementary Material available online at https://doi.org/10.1155/2017/3153924).The average electrical conductivity (EC) values and total dissolved solids (TDS) contents of shallow groundwater are 419 S cm −1 and 288.2 mg L −1 in dry season, whereas the average values of these parameters decline over 10% and 28% in wet season, respectively (Table 3).Dissolved oxygen (DO) and redox potential (Eh) measurements indicate predominantly oxidizing conditions during both dry and wet seasons, with a tendency towards slightly reducing conditions in dry season (Table 3).The pH values vary from 7.9 to 9.4 in dry season and from 7.4 to 9.3 in wet season (Table 3), indicating the slightly to very alkaline nature of the groundwater.The pH values display somewhat lower values in wet season due to supply of low pH recharge water from rain and snowmelt (e.g., mean snow pH = 5.78).Groundwater temperature of shallow HDWs vary slightly (depending on the depth to water) and range from 12.6 to 20.2 ∘ C in dry season and from 11.0 to 16.6 ∘ C in wet season (Table 3).
In the ophiolitic complex aquifer, a significant seasonal variation in groundwater trace element and major ion chemistry is evident from the summary statistics (Table 3).Generally speaking, the concentration values were higher in the dry season than in the wet season (except for Ca 2+ , Cl − , PO 4 3− , Br, and Cr), indicating relatively rapid recharge from precipitation events.At this site, trace elements could be divided into low (<1.0 g L −1 ; Co, Mo, Cu, V, and As), moderate (1.0-10 g L −1 ; Ba, Cr, Li, Ni, and Mn), and high (>10 g L −1 ; Zn, Al, B, Sr, Ti, Br, and Fe) concentration ranges according to their average abundances in the shallow groundwater, considering the entire dataset (e.g., combined dry and wet season samples).The relative abundance of trace elements was ranked (considering median concentrations) in the order B > Fe > Ti > Br > Sr > Zn > Al > Li > Ba > Ni > Mn > Cr > As > Mo > V > Cu > Co for dry season samples, whereas they ranked in the order Br > Sr > Ti > B > Al > Zn > Fe > Ni > Cr > Li > Ba > As > V > Cu > Mn > Co > Mo for wet season samples (Table 3).The concentrations of major cations and anions found in dry season and wet season groundwater samples (along with mean snow composition) are plotted on the Piper diagram [90] in order to determine main water types and depict the hydrogeochemical evolution path (Figure 6).From this figure, it is evident that the dominant ions in all samples are Mg 2+ , Ca 2+ , and HCO 3 − , which is typical of ophiolitic and carbonate terrains [91,92].Nevertheless, many of the groundwater samples contained very low Na + , Cl − , and SO 4 −2

concentrations, both in dry and wet seasons (see Table A.1) (Supplementary Material).
Based on Piper diagram, three hydrochemical facies have been identified, including Mg-HCO 3 , Mg-Ca-HCO 3 , and Ca-Mg-HCO 3 (Figure 6).About 72% and 40% of groundwater samples from dry season and wet season, respectively, belong to Mg-HCO 3 type.The rest of the groundwater samples were mostly classified as Mg-Ca-HCO 3 type, except for two wet season samples (e.g., K7 and K16), which were classified as Ca-Mg-HCO 3 type.The linear scattering of the wet season water samples along the Ca-Mg axis in the  Piper diagram (Figure 6) indicates that during the wet season Ca 2+ domination increases, where some of the Mg-HCO 3 type dry season samples (e.g., K2, K4, K5, K8, K10, K13, and K17) evolve towards Mg-Ca-HCO 3 type.The decline in values of EC and TDS (Table 3), together with increased Ca 2+ domination in wet season samples (Figure 6) can be explained by a mixing mechanism, where more evolved Mg-HCO 3 type dry season groundwater is diluted with less evolved Ca-HCO 3 type recharge water during the wet season.
Indeed, most of the groundwater samples showing a shift in water type during wet season are found in the SW part of the study area where a local recharge area (i.e., Korum Mountain), dominated by carbonate rocks, provides water to the shallow ophiolitic complex aquifer.
In the study area, dry season groundwater from several HDWs (i.e., K4, K5, K6, K10, K18, and K21) is probably affected by the aerobic denitrification processes mediated by the facultative anaerobes [93], which is evidenced by their low NO 3 − (0.7-6.2 mg L −1 ) and relatively high DO (2.07-5.46mg L −1 ), Fe (100.7-1018g L −1 ), and Mn (5.76-78.6 g L −1 ) concentrations (see Table A.1). Fe and Mn are probably mobilized by dissolution of amorphous Fe/Mn oxyhydroxides found in the shallow aquifer system and soils.As mentioned previously, Cr displays higher concentrations in wet season than dry season (Table 3), which is probably due to low pH recharge waters reacting with chromite (FeCr 2 O 4 ), an accessory mineral found in serpentinites (Table 1

Factors Governing the Groundwater Chemistry in the
Kızılgedik Area.In the Kızılgedik area, R-mFA multivariate statistical method was utilized in order to explain factors and key processes affecting alkaline groundwater compositions within the ophiolitic complex aquifer.For that purpose, the combined groundwater dataset ( = 50) from dry and wet seasons was subjected to R-mFA to account for the observed spatiotemporal variability in water chemistries.K-S normality test results showed that the variables Mg 2+ , Na + , Si, Al, Fe, Mn, Ni, and Sr are log-normally distributed ( < 0.05), while the variables Ca 2+ , Cl − , HCO 3 − , and NO 3 − are normally distributed ( > 0.2).Therefore, a logarithmic transformation was applied for the variables that display a lognormal distribution.Subsequently, all variables were standardized by calculating their -scores to scale the data to a range of approximately −3 to +3 standard deviations (), centered about a mean () of zero [95].
In this study, by applying R-mFA method, 12 variables (Ca 2+ , Mg 2+ , Na + , Cl − , HCO 3 − , NO 3 − , Si, Al, Fe, Mn, Ni, and a The proportion of a variable's variance explained by the factor structure. Sr) combined to produce four factors explaining 83.52% of the variance of the original dataset.The remaining factors, explaining only 16.48% of the data variance, were ignored as "noise," since they do not meet the Kaiser criterion [73].The orthogonality of the factors was ensured by rotating them using the varimax raw method.The results of the R-mFA are presented in Table 4, where the factors are arranged in decreasing order of importance.

Factor 1: WRI Involving Ca-and Si-Bearing Phases.
Factor 1 explains 27.68% of the data variance and is mainly associated with the chemical parameters Ca 2+ , Si, and Sr (with loadings −0.854, −0.817, and −0.901, resp.)(Table 4).Sources of these chemical parameters may be attributed to naturally occurring WRI processes involving Ca-and Sibearing phases (e.g., calcite and quartz), which are known to exist in the formations found at topographically higher recharge zones (e.g., Korum Mountain).Strong association of Sr with Factor 1 is probably related to its significant concentrations in carbonate rocks.Sr is well known for its association with carbonates such as limestones and dolomites [96], where it can readily substitute for Ca 2+ and Mg 2+ due to their similar electrochemical behavior.For instance, measured Sr concentrations in limestone samples (R4 and R5) range between 701 and 1192 ppm, whereas exotic blocks (R7 and R8) dispersed within the ophiolitic mélange display much lower values (i.e., 240 and 491 ppm) (Table 2).

Factor 2: WRI Involving Amorphous
Oxyhydroxides and Clay Minerals.The chemical parameters Al, Fe, Mn, and Ni (with loadings 0.926, 0.909, 0.766, and 0.746, resp.)contribute most strongly to the Factor 2 that explains 25.64% of the data variance (Table 4).The high concentrations of these elements are likely due to WRI processes involving amorphous Fe/Mn oxyhydroxides and clay minerals found in the shallow aquifer system and soils.In such environments, autochthonous microorganisms mediate complex biogeochemical reactions using a number of electron acceptors to decompose organic material [97].Site specific geochemical data (Table 2) confirm that Al, Fe, Mn, and Ni are abundant in the rock and topsoil samples, where total concentrations (in wt.%) may reach up to 7.41% (as Al 2 O 3 ), 19.07%(as Fe 2 O 3 ), 0.41% (as MnO), and 0.83% (as NiO), respectively.In natural groundwater (pH < 10), Ni 2+ is preferentially adsorbed onto Fe(III) and Mn(IV) oxyhydroxides [98], which are possibly present as an amorphous phase or as coatings on grains in the aquifer.The close association between Fe, Mn, and Ni is a function of their similar atomic radius and high adsorption capacities of Fe/Mn oxyhydroxides [82].Furthermore, Fe 2+ , Mn 2+ , and Ni 2+ may take part in cation exchange processes between the aqueous phase and aquifer materials, considering the abundance of clay minerals in the study area (see Table 1).

Factor 3: WRI Involving Mg-Bearing
Phases.Factor 3, explaining 18.84% of the data variance, displays significant positive loadings (0.913 and 0.853, resp.) on chemical parameters Mg 2+ and HCO 3 − (Table 4).This indicates that the mineralization of groundwater in this area probably occurs through the WRI processes involving Mg-bearing silicates such as serpentine polymorphs (e.g., lizardite, antigorite, and chrysotile), olivine, diopside, and phlogopite and solubility of carbonate minerals (e.g., dolomite).Although the fractured rock aquifer is almost entirely composed of Mg-silicates, no association between Mg and Si was evident from the R-mFA results (Table 4).This lack of association might be explained by the weathering reactions involving serpentine polymorphs, during which either Mg 2+ ion is preferentially Cesspit factor (m) Figure 7: Plot of cesspit factor (Euclidean distance from groundwater sampling site to the nearest pit) versus groundwater nitrate (NO 3 − ) concentrations in dry and wet seasons.Euclidean distances were calculated using the Spatial Analyst extension of the ArcGIS 9.3.1 software [64].
removed with respect to Si 4+ ion or a secondary Si-bearing phase (e.g., quartz or amorphous silica) is precipitated.

Factor 4:
Atmospheric and Anthropogenic Inputs.Factor 4, explaining 11.36% of the data variance, exhibits significant positive association with Na + , Cl − , and NO 3 − (with loadings 0.740, 0.714, and 0.873, resp.), which are all known to be closely related to dry/wet atmospheric deposition and anthropogenic inputs.In the study site, the substantial amount of these chemical constituents may be derived from the domestic waste (sewage) percolation through cesspits, especially during dry season.However, during wet season, when population is insignificant, dry/wet atmospheric deposition is another process to contribute these constituents to the underlying fractured rock aquifer.In the study area, nitrate concentrations are not well correlated with distance to cesspits.This is depicted in Figure 7, which shows the scatter plot of cesspit factor (Euclidean distance from groundwater sampling site to the nearest pit) versus nitrate concentrations.Aquifer flow is dominated by fractures rather than typical porous media flow meaning that distance may not be good measure.In addition, nitrate will undergo denitrification under aerobic conditions such as those found in the local groundwater (Table 3), making nitrate nonconservative.Thus, direct correlation between nitrate and chloride is not expected unless they have a common source.In fact, nitrate and chloride are not strongly correlated ( = 0.47 for dry season and  = 0.59 for wet season) which is consistent with nonconservative nitrate from cesspits and conservative chloride from the dry/wet atmospheric deposition.The contribution of Cl − from atmospheric deposition can be determined by using the Cl − /Br − data.The source of salt (i.e., halite) in atmospheric deposition is sea spray.As it can be seen in Figure 8, the Cl − /Br − ratio is strongly correlated with Cl − for both dry season and wet season ( = 0.98) and the ratio values are near seawater (200) or lower.Cl − /Br − ratios between 50 and 150 are typical of atmospheric precipitation with lower values possible if Br − is preferentially adsorbed on clays [99].There is no indication of a Cl − source from the cesspits as domestic sewage has Cl − /Br − ratio between 300 and 600 [99].Therefore, we conclude that most of the Cl − is derived from atmospheric deposition in the Kızılgedik site.

Summary of Spatial and Seasonal Water Chemistry
Variations.In the Kızılgedik site, the overall groundwater chemistry is mostly Mg-HCO 3 .As noted in Figure 6, there is some variation with season where the dry season samples are more magnesium rich and the average values for all solutes (except for Ca 2+ , Cl − , PO 4 3− , Br, and Cr) are higher in the dry season compared to the wet season (see Table 3).The arrows in Figure 4 define southern (K6, K7, K8, K17, K15, and K14), central (K3, K19, K11, K12, K13, K14, and K15), and northern (K2, K1, K25, K23, and K24) flow paths.There are variations between wells, but no systematic increase in solutes along flow paths.The larger watershed (to the east, e.g., Korum Mountain) is mainly composed of carbonate rocks (limestone) and karstic in nature.This terrain is expected to generate water chemistry dominated by calcium and bicarbonate.While that signature can explain the Ca-HCO  The effect of anthropogenic inputs to the underlying shallow aquifer can be seen in the concentrations of the chemical constituents NO 3 − , NH 4 + , and NO 2 − (Table A.1).We also see that Fe and Mn are elevated in some HDWs located at the central and northern sectors of the study area (i.e., K1, K18, K19, K21, and K25).The increase in concentrations of these constituents is probably due to domestic waste discharges through cesspits during dry season when population drastically increases compared to the off-season.For instance, dry season NH 4 + concentrations in K1, K19, and K25 (Table A.1) were more than one to eight times the standard value (0.5 mg L −1 ) set by EC [100].Additionally, dry season concentrations of NO 2 − in K1 (0.68 mg L −1 ), Fe in K18 (205.6 g L −1 ), and Mn in K21 (78.6 g L −1 ) were slightly higher than the EC [100] standards for drinking water.
Based on the data, there is limited impact from cesspits to local groundwater chemistry, except for five wells where the standards are exceeded.

Geochemical Modeling of Water-Rock Interaction (WRI).
The geochemical modeling of the WRI starts with the speciation of all water analyses according to the thermodynamic database (e.g., LLNL.dat).All the phases identified in XRD, except vermiculite and lizardite, are found in the database.Nevertheless, the LLNL database did contain antigorite and chrysotile, which are different forms of serpentine.Additionally, generic olivine data was not available, but the LLNL database did contain forsterite, the magnesium endmember of olivine.Therefore, all the identified primary and secondary minerals, except vermiculite, can be included in the model.The data used in the geochemical modeling included temperature, pH, Na, K, Ca, Mg, alkalinity (as HCO 3 ), SO 4 , Cl, F, NO 3 , NH 4 , Fe, Mn, Cr, Al, Si, and Sr (Table 3).Other trace elements were not included to simplify the modeling.PHREEQC was used for speciation of all water samples ( = 50) and to calculate saturation indices (SI) of selected minerals.
4.6.1.Saturation Data.The saturation indices (SI) of dry season and wet season groundwater samples are shown in Figure 9 for selected primary (i.e., calcite, forsterite, chrysotile, and phlogopite) and secondary (product) minerals (i.e., dolomite, quartz, kaolinite, and clinochlore).Minerals with negative SI values (undersaturated) are likely to dissolve, while minerals with positive SI values (supersaturated) are likely to precipitate.Calculated SI values for antigorite and chromite (both not shown) are very positive indicating that the minerals would never dissolve or the thermodynamic data are not very accurate.The primary mineral chrysotile is supersaturated during the dry season (mean SI = 3.14) but becomes undersaturated in most water wells during the wet season (mean SI = −1.47)(Figure 9).Olivine (forsterite) is undersaturated in both dry and wet seasons (Figure 9) with average SI values of −3.94 and −7.07, respectively.Similar to chrysotile, phlogopite is supersaturated during the dry season (mean SI = 4.74) but becomes undersaturated in many well samples during the wet season (mean SI = −0.47)(Figure 9).The variation in saturation during the wet season with greater undersaturation for the same sample location is consistent with the dilution of groundwater with recharge lowering TDS and promoting dissolution of primary minerals (weathering).Calcite is close to saturation at most locations during both seasons (mean SI = 0.38 during dry season and mean SI = −0.05during wet season) (Figure 9), indicating limited calcite dissolution or precipitation at the site.During the dry season, however, some secondary minerals such as dolomite, clinochlore, and magnesite (not shown) show a strong pattern of supersaturation indicating potential precipitation and lower degree of supersaturation during the wet season, while others such as quartz, kaolinite, and Fe(OH) 3 (not shown), are near equilibrium during both seasons (Figure 9).The saturation indices show the main WRI reactions involve dissolution of forsterite and in some cases during the wet season also of phlogopite and chrysotile, together with dolomite, clinochlore, kaolinite, and quartz precipitation.Dissolved silica concentration appears limited by formation of kaolinite and quartz.This accounts for the lack of (statistical) association between Mg and Si in groundwater.While both solutes are supplied from the Mg-silicates, the loss is governed by different minerals with different solubilities.The silica concentration and pH values for all dry and wet season samples are plotted in Figure 10, the mineral stability diagram for the MgO-Al 2 O 3 -SiO 2 -H 2 O system.The samples plot on or near the phlogopite-dolomite-clinochlore boundaries as expected based on the saturation indices.This indicates that the major solutes Mg 2+ and HCO 3 − (Factor 3 from R-mFA) may be controlled by water-rock equilibria between these mineral phases.

Inverse Geochemical Modeling.
Inverse geochemical modeling is a method to quantify the WRI reactions that produce the observed groundwater chemistry.One solution is defined as the starting point and second solution as the end point.The model calculates the mole transfers of minerals and gases required to produce the second solution using dissolution or precipitation reactions [102].
One set of models used snow (see Table 3) as the initial solution and dissolved or precipitated minerals and gases to reach the dry season average water composition (Table 3; models 1-3 of Table 5).A second set of simulations (models 4-6 of Table 5) start with the dry season average water composition and reach the wet season average composition.Table 5 shows the results of the inverse geochemical modeling where the values are in moles per kg of water, and positive values indicate dissolution (or ingassing) while negative values indicate precipitation (or outgassing).The models all consume CO 2 and O 2 dissolve most the primary minerals (serpentine, olivine, calcite, and chromite) as well as minor amounts of halite.Considering that the site is located 100 km north of the Mediterranean Sea coastline, the source of halite (NaCl) is most likely the dry/wet atmospheric deposition of sea salt aerosol particles entrained into the atmosphere by wind action at the sea-air interface.Site specific groundwater and snow Cl − /Br − ratios (Figure 8) together with the lack of a strong correlation between NO 3 − and Cl − ions suggest that domestic waste discharge from the cesspits is unlikely to be a major source of halite in this seasonally inhabited headwater area.All snow to dry season models need ferrihydroxide, clinochlore, chromium oxide, and quartz precipitation.However, model 1 dissolves chrysotile, but not forsterite to account for the dissolved magnesium, while the other models dissolve forsterite.Models 1 and 2 precipitate dolomite, whereas model 3 does not, as it uptakes less CO 2 (Table 5).The two dry to wet season models (models 4 and 5) have the reactant water to account for the dilution effect of recharge (Table 5).Models 4-6 are similar; none have clinochlore or chromium oxide precipitation (there is very little chromite dissolved) but one has dolomite precipitation.Altogether, the models for dry and wet seasons show the major solutes are basically controlled by dissolution of primary minerals and formation of secondary minerals that are observed to be present in the headwater area.Explicitly, the modeling results demonstrated that five reactants or dissolving phases and five products or precipitating phases (see Table 5) are sufficient to explain seasonal differences occurring in groundwater compositions.The difference between dry and wet season groundwater compositions appears to be mostly due to increased influx of water (e.g., recharge from rain and/or snowmelt) during the wet season.However, it should be underlined that the inverse models often provide nonunique solutions, meaning that for any hydrochemical evolution step more than one model is usually found (see Table 5).
In a broad sense, multivariate statistical method used in this study (R-mFA) was able to discern the factors (i.e., Factor 1: WRI involving Ca-and Si-bearing phases; Factor 2: WRI involving amorphous oxyhydroxides and clay minerals; Factor 3: WRI involving Mg-bearing phases; and Factor 4: atmospheric/anthropogenic inputs) representing major geochemical processes taking place in the groundwater system.This indicates that, as an initial step, R-mFA can be of a great assistance while formulating inverse geochemical  models and deciphering the relative importance of a variety of factors (geologic, hydrologic, topographic, physiographic, climatic, anthropogenic, etc.) influencing groundwater composition.In general terms, the results obtained from saturation index (SI) calculations have shown that saturation states of forsterite, quartz, chromite, and halite in individual groundwater samples do not change drastically between seasons, whereas saturation states of the chrysotile, calcite, dolomite, and clinochlore in individual groundwater samples show significant variations.This indicates that mineral phases found in the headwater area respond differently to the increased influx of recharge during the wet season, which is also reflected in the mass transfer amounts of reactants and products calculated by geochemical models (Table 5).

Conclusions
In this study, graphical, multivariate statistical (e.g., R-mFA), GIS, and PHREEQC inverse geochemical modeling tools were employed to investigate the natural and anthropogenic processes governing the alkaline groundwater chemistry of a seasonally inhabited headwater area located in an alpine environment (Adana, Turkey).Dry season (October 2011) and wet season (May 2012) groundwater samples from shallow hand-dug wells on ophiolitic mélange, as well as snow, rock, and soil samples collected throughout the site, were used to evaluate hydrogeochemical processes occurring in this headwater environment.
Results from the XRD and WDXRF analyses of collected rock and soil samples have shown that major mineral phases found in the area include lizardite, antigorite, calcite, dolomite, and quartz, together with minor amounts of hematite and vermiculite, and trace amounts of clinochlore, magnesite, kaolinite, olivine, chromite, ankerite, dickite, diopside, and phlogopite.In this headwater area, karstic features such as sinkholes, together with faults displaying a combined conduit-barrier behavior, appear to play an important role in controlling groundwater flow patterns and recharge/discharge mechanisms.However, no discernable seasonal differences were observed for hydraulic gradients and groundwater flow directions in the ophiolitic complex aquifer.As illustrated by the Piper diagram, Mg-HCO 3 and Mg-Ca-HCO 3 water types are prevalent (with pH values ranging from 7.4 to 9.4), indicating influence of ophiolitic and carbonate rocks on the local groundwater chemistry.The difference between the dry and wet season samples was mainly the dilution of solutes during the recharge period.Results obtained from R-mFA corroborate these results but also point to atmospheric/anthropogenic inputs (e.g., Na + , Cl − , and NO 3 − ).The PHREEQC inverse geochemical models were used to quantify the processes that affect the groundwater evolution between end-member water types.The results of PHREEQC inverse modeling revealed that the groundwater chemistry is strongly dependent on the local serpentine mineralogy that contains reactive minerals such as lizardite, antigorite, olivine, chromite, calcite, and phlogopite.The major solutes (e.g., Mg 2+ , Ca 2+ , and HCO 3 − ) appear to be the product of water-rock interactions of these minerals and local recharge.The byproducts of the weathering reactions include Fe-oxides, clinochlore, silica, and dolomite, all found in the aquifer.The presence of local mineralization and the near equilibrium between minerals and groundwater within this small watershed imply rapid water-rock reactions.The presence of cesspits that discharge sewage directly into the aquifer does cause local exceedance in drinking water standards for nitrite and ammonia, but nitrate concentrations remained below regulatory standards.
Using a methodology employing these diverse tools can provide an increased understanding of the chemical nature of groundwater, its geochemical evolution, and the influence of spatially and seasonally variable anthropogenic inputs from point sources (i.e., cesspits) in the headwater areas with increased human influence.The results obtained from this study can be used to improve public policies and water quality regulations and to promote sustainable development and effective management of water resources in alpine headwater areas, which are usually considered as nearly pristine and out of human influence.However, continuous monitoring is needed to fully characterize the hydrological and hydrochemical dynamics in the headwater environments, which is currently a neglected research topic throughout much of the world.

Figure 2 :
Figure 2: Schematic drawing showing typical designs of hand-dug wells and cesspits in the study area.

Figure 3 :
Figure 3: (a) Detailed geological map (adapted from [36]) of the region surrounding the Kızılgedik site (shown in the red box) overlaid on a digital elevation model (DEM) with a grid size of 10 m and (b) generalized geological cross sections along the lines A-A  (SW to NE) and B-B  (NW to SE) in (a) showing the subsurface lithology and major tectonic structures (faults shown in red color).

KkmKkmFigure 4 :
Figure 4: Hydrogeological map of the Kızılgedik study area (see red box in Figure 3(a)) showing locations of groundwater ( = 25), rock ( = 10), and soil ( = 8) sampling sites.The hand-dug well (HDW) codes (i.e., K1-K25) refer to both water table measurement and groundwater sampling sites.The water table elevations are given at intervals of 2 m above mean sea level (msl) for both (a) dry season (October 2011) and (b) wet season (May 2012).Arrows depict the general direction of shallow groundwater flow in the fractured ophiolitic complex aquifer.

Figure 5 :
Figure 5: Field image of a large fracture system (crosscutting the highly fractured serpentinite rock in Kızılcadag ophiolitic mélange) filled/sealed with chrysotile (i.e., fibrous asbestos) and secondary dolomite.Hammer for scale is 33 cm long.
) and 4(b)), both of which are located near the local recharge areas.The hydraulic gradients calculated from equipotential maps vary between 0.047-0.235m/m for dry season and 0.043-0.223m/m for the wet season, showing no significant seasonal gradient changes in the study area.Steep hydraulic gradients are restricted to the SW mountainous part (near well K6), where highly fractured Kızılcadag ophiolitic mélange is overlain by relatively thin (∼3 m) and low hydraulic conductivity aquitard (i.e., Kaplankaya formation) composed of shallow marine marl, claystone, sandstone, and sandy limestone (Figures 4(a) and 4(b)

a
Electrical conductivity.b Total dissolved solids (calculated as sum of all ions).c Dissolved oxygen.d Mean snow chemistry.Snow samples were collected in January 2012 ( = 6) and February 2013 ( = 5); Min = minimum, Max = maximum, Med = median, and SD = standard deviation.

4 )Figure 6 :
Figure 6: Piper diagram[90] showing the major ion composition (in %meq L −1 ) of the groundwater samples collected in dry season (October 2011) and wet season (May 2012) from the ophiolitic complex aquifer in the Kızılgedik area.Mean snow chemistry ( = 11) of the site represents the starting point in the hydrogeochemical evolution path of groundwater samples.

Figure 8 :
Figure 8: Plot of Cl − concentration versus Cl − /Br − ratio for dry and wet season groundwater samples.Precipitation data reflects the average of 11 snow samples collected from the site on January 2012 ( = 6) and February 2013 ( = 5) after two major snowfall events.

Figure 9 :
Figure 9: Saturation indices of the (a) dry season and (b) wet season groundwater samples with respect to various primary and secondary mineral phases.Saturation indices were calculated by the geochemical code PHREEQC Interactive ver.3.1.4[78] using the accompanying LLNL thermodynamic database [79].

Figure 10 :
Figure 10: Mineral stability diagram for the MgO-Al 2 O 3 -SiO 2 -H 2 O system.The aqueous silica (log  SiO 2(aq) ) concentrations versus pH values for the dry season and wet season groundwater samples are plotted on the diagram using The Geochemist's Workbench5 software[101].Solid lines represent saturation at 14 ∘ C and 1-bar pressure for the selected mineral phases.

Table 1 :
Mineralogical composition of the selected rock (R1-R10) and soil (S1-S8) samples from the Kızılgedik area as determined by X-ray diffraction (XRD) analysis a .

Table 2 :
Chemical composition of the selected rock (R1-R10) and soil (S1-S8) samples from the Kızılgedik area as determined by wavelength dispersive X-ray fluorescence (WDXRF) analysis.Sample number Lithology a

Table 3 :
Summary of seasonal physicochemical composition of groundwater (from identical hand-dug wells;  = 25) and snow ( = 11) samples collected in the study area.

Table 4 :
Factor loadings for groundwater samples ( = 50) of the Kızılgedik area collected during dry and wet seasons.Significant loadings (>0.700) are in bold and underlined (rotation method: varimax raw; extraction method: principal components). 3

Table 5 :
[78]lts of PHREEQC[78]inverse geochemical models related to representative end-member water types (i.e., precipitation (snow), dry season groundwater, and wet season groundwater; see Table3) of the Kızılgedik site (Adana, Turkey) showing the mass transfer of reactants and products.Values are in moles per kilogram H 2 O (positive values indicate dissolution or ingassing and negative values indicate precipitation or outgassing).