The use of hydrogeochem ical analyses and multivariate statistics for the characterization of thermal springs in the Constantine area,Northeastern A lgeria

Riad Kouadra·Abdeslam Dem doum·Nabil Chabour·Rebiha Benchikh

Abstract This paper deals w ith the results of a hydrogeochem istry study on the thermalwaters of the Constantine area,Northeastern Algeria,using geochem ical and statistical tools.The samples were collected in December 2016 from twelve hot springs and were analyzed for physicochemical parameters(electric conductivity,pH,total dissolved solids,temperature,Ca,Mg,Na,K,HCO3,Cl,SO4,and SiO2).Thewatersof the thermal springs have temperatures varying from 28 to 51°C and electric conductivity values ranging from 853 to 5630μS/cm.Q-mode Cluster analysis resulted in the determ ination of twomajor water types:aCa—HCO3—SO4 typew ith amoderate salinity and a Na—K—Cl type with high salinity.The plot of the major ions versus the saturation indices suggested that the hydrogeochemistry of thermal groundwater is mainly controlled by dissolution/precipitation of carbonate minerals,dissolution of evaporite m inerals(halite and gypsum),and ion exchange of Ca(and/or Mg)by Na.The Gibbs diagram shows that evaporation is another factor playing a m inor role.Principal Component Analysis produced three signif icant factors which have 88.2%of total? Riad Kouadra kouadrariad@gmail.com Abdeslam Demdoum slimdem@yahoo.fr Nabil Chabour nabilchabour@gmail.com Rebiha Benchikh rebiha.bencheikh@um c.edu.dz variance that illustrate the main processes controlling the chem istry of groundwaters,which are respectively:the dissolution of evaporiteminerals(halite and gypsum),ion exchange,and dissolution/precipitation of carbonateminerals.The subsurface reservoir temperatures were calculated using differentcation and silica geothermometersand gave temperatures ranging between 17 and 279°C.The Na—K and Na—K-Ca geothermometers provided high temperatures(up to 279°C),whereas,estimated geotemperatures fromK/Mggeothermometers werethelowest(17—53°C).Silica geothermometers gave the most reasonable temperature estimate of the subsurface waters overlap between 20 and 58°C,which indicate possible mixing with coolerMg groundwaters indicated by the Na—K—Mg plot in the immature water f ield and in silica and chloride mixing models.The results of stable isotope analyses(δ18O andδ2H)suggest that the origin of thermal water recharge is precipitation,which recharged from a higher altitude(600—1200m)and inf iltrated through deep faultsand fractures in carbonate formations.They circulate at an estimated depth that does not exceed 2 km and are heated by a high conductive heat f low before rising to the surface through faults thatacted as hydrothermal conduits.During their ascent to the surface,they are subjected to various physical and chem ical changes such as cooling by conduction and change in their chem ical constituents due to them ixing w ith cold groundwaters.

Keywords Hydrogeochem istry·Thermalwaters·Multivariate statistical analysis·Silica geothermometers·M ixingmodels·Cold groundwaters

1 Introduction

Geothermal energy is a renewable resource that can be tapped by many countries around the world;it can be harnessed from underground reservoirs formed by hot rockssaturatedw ithwaterand/orsteam.In Algeria,several studies(Dib Adjoul 1985,2008;Fekraoui 1988;Rezig 1991;Bouchareb-Haouchine 1993;Kedaid and Mesbah 1996;Lahlou M im i et al.1998;Saibi et al.2006;Kedaid 2007;Bahrietal.2011,Bouchareb-Haouchine 2012;Saibi 2009,2015;Belhai et al.2016,2017;Foued et al.2017;Djemmaletal.2017)have shown thatNorthern A lgeriahas a large number of hot springs;over 240 thermal springs havebeen discovered in Algeria(Lahlou M imietal.1998).Historically,these thermal springswere discovered by the Romans and their exploitation is currently lim ited to medical purposes due to their physicochem ical properties such as high m ineralization,temperature,dissolved gas concentration,and radioactivity.Most of these hydrothermal systemsare located in the Northeastern partof A lgeria(Fig.1a)as it hosts thermal springs w ith average temperatures discharge of about 60°C and geothermal gradients ranging from 3 to 4°C/100m(Bouchareb-Haouchineetal.1994).Isotopic and gas studies(Souag 1985;Rezig and Marty 1995;Kedaid and Mesbah 1996)indicate ameteoric origin for these thermal waters and suggest that the chem ical composition in this thermalgroundwater depends mainly on three factors:(1)the composition of recharging water,(2)the reactivity of the geological formations in aquifers,and(3)anthropogenic activities.

A multivariate statistical technique is a useful tool for providing insight into groundwater geochem ical data to better understand the origin,type,and main factors that affect the chem istry of groundwaters.Many researchers have applied itw ith success.Todorovic et al.(2016)and Blake etal.(2016)applied it to assessand characterize the chem istry of thermal and m ineral waters of Serbia and Ireland,respectively,and to identify the differentprocesses inf luencing the hydrochem istry of the springs.In Northeastern A lgeria,specif ically in the Constantine area,many thermalsprings such as Hamma Bouziane,SidiM cide,and Telghma have discharge temperatures ranging between 31 and 60°C.In the past 50 years,many authors(Durozoy 1960;Souag 1985;Dib Adjoul1985;Issaadi1992;Djebbar 2005)have shown that these springshave exhibited a large variability in their chemical properties and temperatures.This work aimed to update and evaluate the hydrochemistry,identify the dom inant processes controlling it,estimate the temperatures and depths of the subsurface reservoirs of the thermal waters in Constantine area,and construct a conceptualmodel explaining themechanisms responsible for watermineralization,hydrochem ical variations,and recharge patterns,relying on geochemical and statisticalmethods.

Fig.1 a Geologicalmap of the study area(Vila 1980),b and c geological cross sections through prof iles(AB and CD see Fig.1a)

2 Geological setting

The studied area is situated in Northeast of Algeria(Fig.1a).It is located from between Latitudes 36°00′to 36°35′N,and Longitudes 6°00′to 7°00′E.The altitude variesbetween 500m and 1000m.Its climate is sem i-arid w ith an annual precipitation of 500mm,mostly falling between Novembersand January.Theaverage atmospheric temperature is 15.6°C.From a geological point of view,the study area is part of the external zone of the Eastern Algeria A lpine Belt(Vila 1980;Fig.1a).It consists of formations ranging from the Triassic to M io-Plio-Quaternary.The Triassic is characterized by gypsum,clay,and dolomite formations.In the south,there is the neritic carbonate platform so-called‘‘The Constantinois carbonate platform,''which starts from Upper Jurassic to Senonian and mainly consists of neritic limestone(Vila 1980),(Fig.1b).It is covered by a geologic unitof limestone-marl alternations from Campanian to Maastrichtian and from Ypresian to Oligocene.In the north,there is a lithological succession consisting of three units,namely,from the bottom to the top:(1)Cretaceous to Eocene of Tellian nappes includingmarl,sandstone,and clay of Barrem ian—Aptian and Priabonian,(2)Cretaceous-m iddle Eocene called?f lysch nappes?,and consisting in Senonian and Oligocene limestone,(3)Oligocene to Burdigalian(Num idian f lysch)represented by a clay-sandstone series overlain by M io-Pliocene to Quaternary continental formations(Fig.1c).

The studied areabelongs to the Alpine structuraldomain(unstable)characterized by high tectonic and seismic activities that occurred between the Mesozoic and Cenozoic periods.Verdeil(1982)regrouped the thermal springs of Northern Algeria and Tunisia into four thermal zones according to some geological and thermal considerations:(1)thermal zone(Oran-Relizane-Tiaret)(2)thermal zone(Cherchel-Te′ne′s-Gafsa-Benghazi-A lexandria)(3)thermal zone(Constantine-El Kef-Maktar)and(4)thermal zone(Annaba-Tebersouk-Takrouana).These zones are located in carbonate formations in the Northeastern part and dolomites in the Northwestern part(Saibi2015).Twomain tectonic stylesare distinguished in the Constantine area:(1)brittle tectonics in limestone massifs(The Constantinois carbonate platform),which consistof carbonate formations of the Mesozoic up to 1000 m thick(intensely karstif ied),reston the saline rocks of Triassic age,and are constituted by the Jurassic-Cretaceous dolom ite and limestone thatare characterized by the presenceofNW—SE conjugated faults,and(2)ductile tectonics in the marl and marl-limestone formations(Tellian nappes),formed during Cenozoic and characterized by two tectonic phases(a late Eocene and M iocene phases).These various Cenozoic tectonic phases reactivated ancient NE—SW,NW—SE,E—W oriented faults(Voute 1967;W ildi 1984).According to some authors(Durozoy 1960;Souag 1985;Djebbar 2005),all faults and fractures in carbonate formations of Mesozoic age allow the ascent of deep hot water,which constitute the main geothermal reservoirs.

3 M aterials and methods

Twelve thermal water samples were collected during the month of December 2016 in the Constantine area(Fig.1a).After f iltering,all samples were stored in polyethylene bottles in an icebox at a constant temperature＜4°C and preserved in a refrigerator(＜4°C)after acidif ication by nitric acid(5m L of 6 N HNO3).The physical parameters such as pH,temperature(T),Electric Conductivity(EC),and Total Dissolved Solids(TDS)were determ ined in-situ using a portable conductivity-meter(HANNA Hi-9813-6 Multiparameter).The major chem ical species were analyzed in the laboratory of Hydrogeology,in the Constantine Faculty of Earth Sciences.The Calcium(Ca),Magnesium(Mg),Bicarbonates(HCO3),and Chlorides(Cl)ionswere analyzed by volumetric titrations.Sulfate(SO4)ionswere analyzed using a Jenway 6051 Colorimeter.Sodium(Na),and Potassium(K)ions were analyzed by atom ic absorption spectrometer(Perkin-Elmer model AAnalyst 200).SiO2were measured by atomic absorption spectrophotometry at the water chem istry laboratory(ref ining unit)Sonatrach-Skikda.

3.1 Saturation index

SI is a usefulmethod to understand rock-water interaction in aquifers,based on equilibrium conditionsof the solution with respect to a given m ineral(Deutsch 1997).It is def ined by the follow ing equation:where IAP is the ion activity product and Ksp is the equilibrium constant.When IAP=Ksp,the SI value is 0 and the aqueous solution is at equilibrium with respect to the givenmineral.If SI＞0,the solution is supersaturated with the components of the considered mineral.If SI＜0,the solution is undersaturated w ith respect to them ineral.The hydrogeochem ical modeling code,PHREEQ-C was used to calculate the SI(Parkhurst and Appelo 1990).

3.2 M ultivariate statistical analysis

Multivariate statistical analysis is an advanced technique that has been specif ically developed for studying and analyzingdatasets(RaykovandMarcoulides2008;Wolfgang and Le′opold 2012).In groundwater research,multivariate statisticalanalysis isaquantitative approach to groundwater classif icationallow ingthegroupingof groundwater samples and making correlations between samplesand their chem icalparameters(Cu¨neytetal.2002;Feng etal.2007;Mencio and Mas-Pla 2008;Agelos etal.2010;Belkhiri et al.2010;Bencer et al.2016).In this study,twomultivariate statistical analyses(Principal Components Analysis and Q-mode Cluster Analysis)were used since they are usually applied together(Shrestha and Kazama 2007;Omo-Irabor et al.2008;Nosrati and Van-Den-Eeckhaut2012;Am inu etal.2015)in order to classify groundwater analyses and identify the majormechanisms inf luencing groundwater chem istry,on the basis of their physicochemical and chem ical characteristics.All statistical calculations were done throughM icrosoft off ice EXCEL 2016 and XLSTAT 2016(Trial version).

4 Results and discussion

4.1 Hydrogeochem ical characteristics

Before proceeding to interpreting chemical data,it is essential to check their reliability through an ionic balance calculation.The error in the ionic balance should not exceed 5%according to the follow ing equation(Appelo and Postma 1996):

Forall thewater samples,the IB(%)values varied from 0.24 to 2.99%(Table 1),and therefore,we consider these analyses reliable.

The chem ical compositions of the thermal springs are given in Table 1.The temperature of the water samples vary from 30 to 51°C;pH values vary from 6.2 to 7.8;electric conductivity(EC)varies from 853 to 5630μs/cm at 25°C(also in Table 1).TDS values range from 400 to 2540mg/L w ith an average value of 955 mg/L.Piper and ternary diagrams(Fig.2a,b)classif ied all water samples into‘M ixed Ca—Mg—HCO3Type,'Ca—Mg—Cl—SO4-Type,and Na—K—Cl Type,suggesting that these waters are mainly inf luenced by deep carbonate reservoir rocks,an interactionw ith brine f latwater and/or the dissolution of an evaporitic sequence near the upf low area of each hot spring.

The Pearson correlation coeff icients(Table 2)show the strong correlations between Na,Cl,K,and TDS(0.94,0.92,0.71 respectively)and indicate that them ineralization ismainly inf luenced by Na,Cl,and K.The concentrations of Na and Cl vary from 88.17 to 934mg/L and 40.95 to 1544.25mg/L,respectively and they have a high positive correlation(0.99)(Fig.3a),meaning that the dissolution of Halite(NaCl)is the source of Na and Cl in groundwater.

Plot(Ca+Mg)versus(HCO3)(Fig.3b)shows that the majority of samples are clustered above the 1:1 line and represents an excess(Ca+Mg)over HCO3,suggesting an additional source of Ca and Mg rather than carbonate minerals(calcite and dolom ite).The plot(Ca+Mg)versus(HCO3+SO4)(Fig.3c)shows thatmost of the samples cluster around the 1:1 line indicatingthat the dissolutions of calcite,dolom ite,and gypsumare the dom inating process in the system,asshown in the Eqs.(3),

Table 1 Values of the parameters of each sample

Fig.2 a Piper diagram,b ternary plot(anion species)

Table 2 Pearson's correlation coeff icients between physicochemical parameters for groundwater samples

(4),and(5).Thesamplesplotaboveand below the1:1 line;indicate the existence of ion exchange,which can be expressed as formulation(6).

When ion exchange is the dom inant process thataffects thehydrochem istryof groundwater,therelationship between(Ca+Mg)-(SO4+HCO3)versus(Na—Cl)should be a straight line w ith a slope of-1(Fisher and Mullican 1997).Figure 3d shows that the samples fallon a straight line(R2=0.94)w iththeslopeof-1.29 suggesting that therewas ion exchange of Ca(and/or Mg)by Na in those groundwater samples.

Saturation indices calculated from major ions(Table 3)indicate thatmost of the groundwater samples were in an oversaturation state(SI＞0)w ith respect to carbonate m inerals(calcite,dolom ite and aragonite)(Fig.4),but in an undersaturation state(SI＜0)w ith respect to the gypsum,halite,and anhydrite,indicating that groundwater chem ical composition was partly controlled by the dissolution evaporite m inerals and precipitation/dissolution of carbonate minerals.The pCO2values range between 3.24×10-1and 0.11×10-1atm(Table 3).These values are higher than thatof the Earth's atmosphere(10-3.5atm)suggesting additional sources of CO2in groundwater by root respiration and soilorganicmatter decomposition and/or the release of deep CO2(Van derWeijden and Pacheco 2003;Emblanch et al.2003).

Gibbs(1970)suggested a graphic representation to understand major processes controlling the groundwater chem istry,based on Total Dissolved Solids(TDS)versus Na/(Na+Ca)for cations and Cl/(Cl+HCO3)for anions.According to Gibbs,threemainmechanisms controlworld water chem istry:precipitation,evaporation,and rock-water interaction.Generally,Gibbs diagram(Fig.5)shows that most of the groundwater samples fall between rock-water interaction and evaporation dom inant.This diagram shows amixed controllingmechanism of groundwater chemistry.

Fig.3 a Na versus Cl,b Ca+Mg versus HCO3,c Ca+Mg versus HCO3+SO4,d(Ca+Mg)-(HCO3+SO4)versus Na—Cl

Table 3 Saturation indices of groundwater samples

Fig.4 Saturation indices of evaporitem inerals(halite,gypsum,anhydrite),and carbonatem inerals(calcite,dolom ite,Argonite)versus(Na+Cl),(Ca+SO4),(Ca+HCO3)and(Ca+HCO3+Mg),respectively

4.2 Principal com ponents analysis

Principal components analysis(PCA)is a benef icial statistical technique to reduce the number of correlated variables into a smaller set,called principal components(Farnham et al.2003).It is w idely used in groundwater quality studies to determine themain factorsand processes that control the water quality based on physicochemicaland chem ical data(Pathak et al.2008;Singh et al.2010;Trikietal.2014).

Fig.5 Gibbs plot of thermalwaters

In this study,PCAwas applied to hydrochemical parameters of the 12 analyzed thermal waters.The f irst three factors(F1,F2,and F3)represent 88.2%of the total variance(Table 4).Factor 1(55.5%)has a strong loading of EC,TD,pH,Mg,Cl,Na,and K(Table 5).It is def ined as the salinity factor since it ref lects the differentsourcesof groundwater salinization:the dissolution of evaporite minerals(mainly halite)and ion exchange of Ca(and/or Mg)by Na.The second factor explains 18.8%of the total variance.It is determined by Ca,T,and SO4.This factor indicates the dissolution of sulfate evaporites(gypsum)and the role of thermalism in groundwater.Factor 3 exhibits 13.9%of the total variance.HCO3shows a strong positive loading on this axis as explained by the dissolution/precipitation of carbonatem inerals.

Thus,from the PCA,it can be inferred that the different components PC1,PC2,and PC3 ref lect the main factors that control the geochemistry of groundwater,which is:dissolution of evaporiteminerals(halite and gypsum),ion exchange of Ca(and/or Mg)by Na,and dissolution/precipitation of carbonatem inerals.

Table 4 Distribution of f irst three principal factors

Table 5 Variables-factors correlation

4.3 Q-mode cluster analysis

Q-mode cluster analysis is a highly useful statistical tool which aims to f ind relatively homogeneous clusters based on measured characteristics(Ward 1963).Many researchers have successfully used this technique to classify water samples(Tenalem et al.2009;Belkhiri et al.2010;King et al.2014;Raf ighdoust et al.2016).In this case,Q-mode cluster analysis is used to categorize the thermal waters into different groups based on their physicochemical parameters.This analysis resulted in the grouping of thermal waters into two clusters relying on their salinity(Fig.6):

·The waters of the f irst cluster have amedium salinity(mean EC=1506.2μS/cm)(Table 6)and the watermineral solubility and/or exchange reactions(Kharaka and Mariner1989).In thisstudy,the following geothermometers were applied to estimate subsurface temperatures of water collected from hot springs(as shown in Table 7):Na—K(Truesdell1976),Na—K(Fournier 1979b),Na—K—Ca(Fournierand Truesdell1973),K/Mg(Giggenbach1988),Silicano steam loss(Fournier 1977),Silicamax steam loss(Fournier 1977),and Quartz(Verma2000).type is calcium bicarbonate and sulfate-rich(Fig.7).It hasmoderate Total Dissolved Solids(TDS)(1080mg/L)w ith a temperature up to 51°C.

Fig.6 Dendrogram of the hydrochemical samples

Table 6 Mean parameters values of the two principalwater clusters

·Thewatersof the second cluster aremore salinewaters(mean EC=4390μS/cm).The water type is sodium and potassium chloride rich w ith high concentrations sodiumandpotassium.Theirtemperaturesvary between 34 and 41°C.

5 Chem ical geothermometry

Chemical geothermometry is an important tool to estimate reservoir temperatures of hydrothermal systems.Silica and cation geothermometers are classic chem ical geothermometers used in geothermal exploration,and they are based on

Fig.7 Durov diagram of water clusters

Temperatures estimated by the Na/K and Na—K—Ca geothermometers shown in Table 7 and Fig.8a give high values ranging from 26 to 279°C.They are consistentwith changesof theNa/K and Ca/Na ratio thatmay bea resultof the dissolution of Triassic saline formations rich in Na and water—rock interaction(ion exchangeof Ca(and/orMg)by Na)during the ascentof thermalwaters toward the ground surface.

The K/Mg geothermometer(Table 7;Fig.8a),derived by Giggenbach(1988)gives low temperatures varying from 17 to 50°C(almost less than the discharge temperatures)due to the rather fastequilibrating aftermixingwith Mg-rich groundwater(Giggenbach 1988).

Another approach to evaluate reservoir temperature and recognizewaterswhich have attained equilibrium w ith the host lithologies,have also been obtained through applicationoftheK/100-Na/1000-Mg1/2ternarydiagram(Giggenbach 1988;Fig.9),where most of the thermal watersare plotted in the immaturewater f ield whichmeans that they have notattained equilibrium(close to the Mg1/2corner).Thismay indicate that they were possibly m ixed with cold water of shallow aquifers during upf low such as near-surface reaction w ith the dolomite-limestone rocks of Jurassic age that contains high Mg concentrations.However,Silica geothermometers are considered more reliable and provide good results.The silica max steam loss(Fournier 1977)geothermometer gives a sim ilar temperature to silica no steam loss geothermometer(Fournier 1977)ranging from 25 to 58°C(Table 7;Fig.8b),while the quartz geothermometer(Verma 2000)providesa lower temperature than those of silica geothermometers,which ranged from 20 to 43°C.Generally,in the study area,the estimated temperatures obtained by geothermal instruments from silica and those measured on the surface are relatively sim ilar.

Table 7 Estimated reservoir temperature(°C)of the Constantine thermalwaters using different cation and silica geothermometers

Fig.8 a Estimated reservoir temperatures based on cation geothermometers.b Estimated reservoir temperatures based on silica geothermometers

Table 8 Estimated depth(km)for Constantine thermalwaters based on silica geothermometers

Fig.9 Giggenbach ternary diagram for the Constantine geothermal waters

Considering(1)the Constantine region belongs to an area where the geothermal gradient is 35—40°C/km(Bouchareb-Haouchine etal.1994,2012;Issaadi1992)all geothermal maps wereupdatedintheBoucharebe-Haouchine thesis(2012)which relied on the oil well data(bottom-hole temperature BHT and temperature of f luids in the drillstem testDST),(2)theaverageannual temperature(about 18.5°C),and(3)the deep reservoir temperatures estimated by Silica geothermometers(20—58°C),the inferred circulation depths ranged from0.2 to 1 km(Table 8),suggesting that the corresponding reservoir should be located less than 2 km below the surface.From a geological point of view,these depths correspond to the Jurassic-Cretaceous dolom ite and limestone,which are considered to be themain geothermal reservoir.

6 Evidence ofm ixing

6.1 Silica-enthalpy m ixingmodel

Truesdell and Fournier(1977)have suggested a simplif ied graphicalmethod to estimate the temperature of the deep hot water component,which is based on dissolved silica concentration versus enthalpy of the spring water.This model relied on the assumption thatno conductive cooling has occurred after m ixing.The Silica-enthalpy m ixing diagramfor the studied thermal waters is shown in Fig.10a,where all samples plotted tend to cluster very close to and above thequartz solubility line.Ifwe take into consideration that there is no loss of heatafterm ixing,the enthalpy estimated for the thermalwaters isabout400(kJ/kg)which corresponds to the temperature of 84°C,which is considered higher than those estimated by quartz geothermometer.This indicates that the studied thermal groundwatershavemixedw ith coolerwater in the reservoir or conductive cooling occured during the upf low of the hot waters.

6.2 Enthalpy-chloridem ixing m odel

The enthalpy-chloridem ixing model is particularly useful to understand the hydrology of a geothermal f ield.It is widely used to derive information on subsurface processes such as:boiling and mixing/cooling by dilution or conductive cooling.Itwas suggested by Fournier(1979a)who assumed thatmost chloride in thermalwater is released by water interaction at high temperatures in the geothermal reservoirs.Its application involves relating analyzed chloride levels to water enthalpy,which can be derived from measured discharge temperature,geothermometry temperature,andsilica—enthalpym ixingmodel temperature(Shaoping 1997;Magana 1999).Figure 10b shows the enthalpy-chloridem ixingmodel for theConstantinethermalwaters,where,all hotwaters lie along them ixing linewhich indicates that them ixing of thermalwaterwith cold water probably occured during their ascent to the surface.

Fig.10 a Silica-enthalpy m ixing model(Truesdell and Fournier 1977)for the studied thermal waters.b Chloride-enthalpy m ixing model for the Constantine thermalwaters

7 Stable isotope geochem istry

The stable isotopes18O and2H are very useful in several hydrogeologicalstudies to investigate the origin ofwater in the aquifer systems,f low patterns,m ixing processes,and for identifying potential evaporation processes(Craig 1963;J?rgensen and Banoeng-Yakubo 2001;Gastmans etal.2010;Xu etal.2013;Al-amerietal.2014).Figure 11 shows theδ18O andδ2H for some studied waters(Souag 1985)where the global meteoric water line(GMWL,Rozanski et al.1993)and the local meteoric water line(LMWL,extrapolatedafterSouag1985)arealso presented.

The isotope contents of thermal waters range from-7.82 to-7.97δ18O‰and-56.9 to-58.5δ2H‰,respectively(Table 9).These values plot both below the GMWL,butas a group,they are approximately parallel to GMWLandLMWLw iththeequationδ2H=8 δ18O+5.6,indicating a meteoric origin for the groundwaters,while the deuterium-excess values indicate the effect of secondary evaporation processes.The high18O values probably suggest that relatively little evaporation occurred during the recharge process.(probably through Jurassic-Cretaceousdolom iteandlimestonethat are intensively f issured and karstif ied).

Table 9 Isotopic results ofδ2D(‰)andδ18 O(‰)relative to SMOW of Constantine thermalwaters(Souag 1985)

Fig.11(δ18O—δ2H)plot for thermalwaters from study area

Based on a conceptual geothermal model for the northern Algerian geothermal system(proposed by Saibi 2009),Fig.12 shows a simple conceptualmodel that has been constructed for the geothermalwaters of Constantine area where the meteoric waters recharged from higher altitude(600—1200 m;Souag 1985)and inf iltrated the reservoir rocks along faults and fracture of carbonate formations(limestone and dolom ite)of the Jurassic to Upper Cretaceous.After being heated by high conductive heat f low(80—140mW/m2;Rimi et al.2012;Barkaoui et al.2014),they rise to the surface along faults and fracturesthatactaspathways for thermalwatersand produce thehot springs;in some cases they are extracted by boreholes(Fig.12).During their ascent to the surface,they undergo chem ical changes and conductive cooling due to m ixing with cold near-surface groundwaters.

Fig.12 Conceptual evolution model for the Constantine thermalwaters

8 Conclusion

Carbonate rock reservoirs are an important geothermal resource in the Constantine area.Based on geological conditionsandhydrogeochem ical characteristics,the geothermalwaters are an intermediate type(between Ca—Mg—HCO3type,Ca—Mg—Cl—SO4-Type,and Na—K—Cl Type)w ith discharge temperatures between 30 and 51°C.Two major groups were inferred from Q-mode Cluster analysis based on salinity in this order:cluster 1(moderate)＜cluster 2(high),whereas,the results of principal components analysis highlighted three main factors that ref lect themain processes controlling the geochem istry of groundwater:dissolution of evaporiteminerals(halite and gypsum),ion exchange of Ca(and/or Mg)by Na,and dissolution/precipitation of carbonatem inerals.Saturation indices(SI)indicate that the water is supersaturated with respect to most of the carbonateminerals and undersaturated w ith respect to evaporitem inerals conf irming water—rock interactions by dissolution and precipitation processes.The temperatures of the deep geothermal reservoirs are estimated between 17 and 279°C based on Na/K(Truesdell 1976;Fournier 1979b),Na—K—Ca(Fournier and Truesdell 1973)and K/Mg(Giggenbach 1988)geothermometers,while slightly lower estimates are obtained using silica geothermometers(20—58°C)which are considered more reasonable temperature than those estimated by cation geothermometers.According to stable isotopes data(δ18O andδ2H),themeteoric water inf iltrated through the deeper faultof carbonate formation and recharged from a higheraltitude(600—1200 m).A fter they become hotdue to the high conductive heat f low(80—140 mW/m2),the heated water rises to the surface through faults and fractures that act as hydrothermal conduits.Many physical(such as cooling)and chemical changesoccursduring their ascent to the surface due to mixing with cooler Mg-richgroundwaters,which are clearly indicated in silica and chloride enthalpy m ixing diagrams.

Acknow ledgem entsThis research was supported by(Faculty of Earth Science,University of Constantine 1).We are thankful to our colleagues from the Laboratory of Hydrogeology and from the Laboratory of water chemistry(ref ining unit)Sonatrach-Skikda who provided their expertise that greatly assisted our research.