larva.indd 299 � AB STRA CT At the “Pašino vrelo” source, groundwater is exploited by means of wells which tap a carbonate aquifer of Badenian age. The natural quality of groundwater complies with the provisions of the Regulation on health and safety of drink- ing water. However, anthropogenic impacts in the form of microbiological pollution have been observed. Extensive hydrogeological investigations have been conducted in recent years in the recharge area of the groundwater source with the aim of ensuring additional quantities of drinking water and protection of the source from pollution. The analysis of hydrogeological monitoring results for groundwater levels and discharges at the groundwater source con- fi rmed the existence of a hydraulic connection between the Quaternary alluvial and Badenian carbonate aquifers. Differences in the hydrodynamic conditions between the “Pašino vrelo” spring and “Bojanića vrelo” spring have also been observed, and are related to the infl uence of pumping on the groundwater discharge regime, but also to drainage conditions at the springs. Hydrochemical indicators also suggest the mixing of groundwater from both aq- uifers in springs and exploitation wells. Specifi c hydrogeological conditions at the groundwater source determined the concept of its development. Recent hydrogeological research was carried out in order to explore the possibilities of tapping deeper parts of the carbonate aquifer in order to minimise the infl ow of groundwater with anthropogenic impacts from the shallow alluvial aquifer during groundwater abstraction. Preliminary results for groundwater qual- ity in the new well justifi ed the aforementioned investigation approach. Intensive exploitation of the well will show whether these conditions are sustainable in the long term. Keywords: carbonate/alluvial aquifer, groundwater, hydrodynamics, hydrochemistry, isotopes, groundwater quality protection, groundwater pollution Hydrodynamic and hydrochemical con- ditions at the groundwater source “Pašino vrelo”, with a focus on its development � Ozren Larva, Tamara Marković and Vinko Mraz Department of hydrogeology and engineering geology, Croatian Geological Survey, Sachsova 2, 10 000 Zagreb, Croatia; (olarva@hgi-cgs.hr) doi: 104154/gc.2010.24 Geologia Croatica 63/3 299–312 12 Figs. 1 Tab. Zagreb 2010 Geologia CroaticaGeologia Croatica 1. INTRODUCTION The “Pašino vrelo” source of drinking water is located in the Sunja river valley, between the settlements of Borojevići and Mečenčani in the Sisak-Moslavina county, in the municipal- ity of Hrvatska Kostajnica (Fig. 1). It is named after the “Pašino vrelo” spring, which is the most important hydro- geological phenomenon in the region. There is another spring at the groundwater source – “Bojanića vrelo”, as well as pumping site facilities comprising four abstraction wells (PVB-1, PVB-2, PVB-3, and SPBPv-1/06) and several ob- servation wells. The “Davidovića vrelo” and “Grabovac” Geologia Croatica 63/3Geologia Croatica 300 springs occur in the vicinity of the groundwater source. The former is periodic and the latter a perennial spring. With re- spect to surface waters, the Sunja river is the most prominent watercourse, but there are also a few smaller streams – e.g. the Bekej stream in the south-eastern part of the groundwa- ter source. The pumping site has been active since the 1970s, when two wells were drilled, based on favourable results of hy- drogeological and geophysical investigation works. Today, three active wells abstract groundwater from the carbonate aquifer of Badenian age which underlies the Quaternary al- luvial aquifer. Both aquifers form the “Pašino vrelo” aquifer system, the hydrodynamic and hydrochemical characteristics of which had not previously been extensively explored. The nominal yield of the pumping site is 38.4 l/s, however, in practice about 20 l/s are abstracted on average, due to the limited fl ow rate in the distribution network. The natural qua- lity of groundwater complies with the provisions of the Reg- ulation on health and safety of drinking water (OG 47/08). However, specifi c hydrogeological conditions in the aquifer together with the occurrence of polluters in the immediate vicinity result in occasional microbiological pollution of groundwater. Such circumstances were important consider- ations in determining the concept of groundwater source pro- tection, and in the investigative works carried out in recent years in order to ensure additional quantities of drinking wa- ter. In recent times there is a renewed need for abstraction of larger quantities of water at the pumping site, since the con- Figure 1: Location map showing the studied springs and wells. ceptual design of the water supply system anticipates, in its fi nal stage, a maximum daily water consumption of 108 l/s. The objective of this study was to increase the level of understanding of hydrogeological model of the “Pašino vrelo” groundwater source. This was accomplished by con- ducting analyses of the hydrodynamic and hydrochemical characteristics of both the Quaternary alluvial aquifer and Badenian carbonate aquifer. Research was undertaken between 2005−2007. A groun- dwater monitoring program was established in order to in- terpret the hydrodynamics of the aquifer system. It included monitoring of groundwater levels in the alluvial aquifer, wa- ter levels in the “Pašino vrelo” spring and discharge of both the “Pašino vrelo” and “Bojanića vrelo” springs. On several occasions, groundwater samples were taken from springs and wells for the purpose of defi ning the hydrochemical con- ditions in the aquifer. Besides the measurements of physical and physico-chemical parameters of groundwater in situ, analyses were subsequently conducted in laboratories regard- ing the basic chemical composition and the ratios of stable hydrogen and oxygen isotopes in the groundwater. Applica- tion of the mass-balance model facilitated estimation of the proportions of water from carbonate and alluvial aquifers in the wells and springs under different hydrological conditions. The research results provided the basis for updating the protection zones according to the Regulation on the deter- mination of sanitary protection zones of the source (OG 55/02) and directed the investigations towards deeper parts Larva et al.: Hydrodynamic and hydrochemical conditions at the groundwater source “Pašino vrelo”, with a focus on its development Geologia Croatica 301 Figure 2: Hydrogeological map. 1 – geological boundary; 2 – supposed geological boundary; 3 – supposed transgressive boundary; 4 – fault; 5 – sup- posed fault; 6 – reverse fault; 7 – Plio-Pleistocene clastic sediments, Pl, Q – poorly permeable rocks; 8 – Quaternary alluvial sediments, al-Q2 – dominant- ly very permeable rocks; 9 – Badenian lithothamnium limestone, M4 – permeable rocks; 10 – Eocene, Ottnangian, Sarmatian and Pannonian clastic sed- iments, E, M3, M5, M6 – dominantly impermeable rocks; 11 – drilled well; 12 – dug well; 13 – exploratory borehole; 14 – intake. of the carbonate aquifer with the aim of abstracting addi- tional groundwater quantities without anthropogenic conta- mination. Thus, in 2005, an exploratory borehole SPBPv -1/05, and a year later a test exploitation well SPBPv-1/06 were drilled west of the pumping site. The results of the pumping test as well as of chemical analyses of groundwater from the well are promising with respect of both groundwa- ter abstraction quantity and quality. 2. HYDROGEOLOGICAL CHARACTERISTICS OF THE GROUNDWATER SOURCE In the greater area of the groundwater source, deposits range in age from the Palaeogene to the Quaternary (Fig. 2). The Palaeogene is represented by Eocene fl ysch sediments (E), which are characterized by both vertical and lateral exchange of sandstones, shales, siltites and conglomerates. Weather- ing of these resulted in materials which formed younger sed- iments. The possibility of groundwater accumulation in them is slight, but they are therefore characterized by a developed surface hydrographic network. The springs in these deposits have low yields, of 0.1 l/s or less. Ottnangian clastites (M3) discordantly overlie the Eo- ce ne deposits. They appear as erosion remnants in Eocene fl ysch, and consist predominantly of alternations of clayey sands and clays. Badenian deposits (M4) transgressively overlie the Ot- tnangian sediments. The basal part of these mostly consists of loosely cemented conglomerates or breccias, with frag- ments of older bedrock. They are overlain by carbonate rocks represented by lithothamnium limestone. These rocks are porous, and thus enable infi ltration of surface waters and cir- culation of groundwater. Besides the porosity, the Badenian rocks are often also characterized by karst phenomena - sink- holes, smaller caves and conduits with signifi cant local in- fl uence on the hydraulic characteristics of the deposits. Sarmatian and Pannonian sediments (M5 and M6) are predominantly represented by marls, and Plio-Pleistocene clastites (Pl, Q) are mostly clays, clayey sands and some- times gravels. The youngest deposits are Quaternary alluvial sediments (al, Q2), which are found in the Sunja river valley. They are represented by gravels, sands, clays and clayey gravels with boulders of older rocks, and originate from the weathering and resedimentation of older rocks. The deeper parts mostly consist of coarser, prolluvial deposit, with subsequent sedi- mentation of river sediment. The Sunja river frequently chan- ged its course in the past, which caused vertical and lateral Geologia Croatica 63/3Geologia Croatica 302 interfi ngering of coarse-grained gravel, sand and sandy and gravelly clay. The whole Sunja alluvium is approximately 10 m thick, unconsolidated to poorly consolidated sediment, which in its coarse-grained parts, enables infi ltration and cir- culation of groundwater. From the hydrogeological standpoint, these sediments can be categorized as either impermeable or permeable rocks. Eocene, Ottnangian, Sarmatian and Pannonian clas- tites are impermeable rocks, whereas the group of permeable rocks includes the carbonate deposits of Badenian age and Plio-Pleistocene and Quaternary alluvial sediments. In the lithological cross-section in the greater area of the “Pašino vrelo“groundwater source, the Quaternary alluvial aquifer is located immediately above the Badenian carbon- ate aquifer. The appearance of the springs was enabled by favourable structural-tectonic relationships. These are char- acterized by the Dinaric strike of the main structural units and tectonic elements. Along the faults, there appears to be a gradual lowering of Neogene deposits towards the north- east. Repeated tectonic movements caused fracturing of hard carbonate rocks, thus creating conditions for infi ltration and circulation of water below ground. The appearance of the “Pašino vrelo” and “Bojanića vrelo” springs are consequen- ces of a pronounced diagonal fault in the Sunja valley, which brought into direct contact permeable lithothamnium lime- stones and impermeable Pannonian marls. Impermeable marls caused the appearance of upward springs in this part of the Sunja river valley, thus forming the “Pašino vrelo” groundwater source. 3. HYDRODYNAMIC CONDITIONS AT THE GROUNDWATER SOURCE The fi rst known data on yields at the “Pašino vrelo” ground- water source date back to 1970-1974 (MAGDALENIĆ et al., 1976). The characteristic values of daily discharges at the “Pašino vrelo” spring at that time were: Qmax = 115 l/s; Qmin = 52 l/s; Qmean = 75 l/s. The authors state that the min- imum discharge of the “Pašino vrelo” spring group, which also includes the “Bojanića vrelo” spring, exceeds 80 l/s. At the end of 2005, hydrogeological monitoring was again established in the greater area of the pumping site. Groundwater levels were monitored in dug wells, piezome- ters and an exploratory borehole (Fig. 3). The dug wells tap the Quaternary alluvial deposits of the Sunja River, whereas the piezometers, exploratory well and drilled wells tap the Badenian carbonate aquifer. Water levels were monitored at the “Pašino vrelo” spring, while spring yields were estimated at “Pašino vrelo” spring and “Bojanića vrelo” spring using weirs. Upstream and downstream of the pumping site, staff gauges were installed in the Sunja River. Figure 3: Equipotential map with groundwater fl ow direction and outer boundary of the II. protection zone. 1 – geological boundary; 2 – supposed trans- gressive boundary; 3 – fault; 4 – supposed fault; 5 – Plio-Pleistocene clastic sediments, Pl. Q – poorly per- meable rocks; 6 – Quaternary alluvial sediments, al-Q2 – dominantly very permeable rocks; 7 – Bad- enian lithothamnium limestone, M4 – permeable rocks; 8 – Eocene, Ottnangian, Sarmatian and Pan- nonian clastic sediments, E, M3, M5, M6 – imperme- able rocks; 9 – drilled well; 10 – dug well; 11 – explor- atory borehole; 12 – perennial spring; 13 - periodic spring; 14 – intake; 15 – staff gauge; 16 – equipoten- tial line; 17 – groundwater fl ow direction; 18 – piezo- metric head; 19 – outer boundary of the II. protection zone Larva et al.: Hydrodynamic and hydrochemical conditions at the groundwater source “Pašino vrelo”, with a focus on its development Geologia Croatica 303 Fi gu re 4 : Groundwater levels in dug wells and the “Pašino vrelo” spring with linear regression lines. Fi gu re 5: Depths to groundwater in dug wells and springs disc charge. Interpolation of groundwater levels in the monitored fa- cilities and the water levels between the staff gauges in the Sunja River allowed construction of a groundwater equipo- tential map for low water levels conditions on 17 November 2006 (Fig. 3). Groundwater fl ow has a general south-east di- rection. The drainage infl uence of the river is observed as well as a cone of depression caused by groundwater abstrac- tion at the pumping site. Hydrodynamic conditions in the Badenian carbonate and Quaternary alluvial aquifers were investigated by analysing a groundwater level time series at the pumping site and in its recharge area as well as through the yields of the “Pašino vrelo” and “Bojanića vrelo” springs. Oscillations of groundwater levels in the Quaternary al- luvial aquifer and the water levels in the spring “Pašino vrelo”, revealed opposite trends (Fig. 4). In the Quaternary alluvial aquifer, groundwater levels show a trend of lower- ing from the beginning of June to mid-November 2006. Con- versely, the “Pašino vrelo” spring showed a trend of rising water levels over a slightly longer monitoring period from the beginning of March to mid-November 2006. The yields of the “Pašino vrelo” and “Bojanića vrelo” springs show different dynamics from November 2005 to December 2007 (Fig. 5). The values registered at the “Pašino vrelo” spring range from 6.7 l/s in January 2006 to 25 l/s in February and April 2007. On the other hand, the “Bojanića vrelo” spring shows greater oscillations in yield, ranging from 3.5 l/s in August 2007 to 50 l/s in December 2007. When the yields of the “Pašino vrelo” groundwater source between 2005−2008 are compared to the yields of the same spring between 1970−1974, a substantial decrease is quite obvious. Geologia Croatica 63/3Geologia Croatica 304 In order to assess the hydrological conditions in the re- gion and their possible infl uence on the reduction in spring yields, the following data were analyzed: • A time series of stage data on the Sunja River from 1965−2009. The gauging station is located in Sunja (Fig. 1). The stage data were analyzed separately for the pe- riods 1965-1987 and 1987−2009 because the gauging station was moved from its original position to a new location at the beginning of 1987. • A time series of precipitation data for 1965−2009 at the main meteorological station at Sisak (Fig. 1), located approximately 20 km to the north of the investigated area. Analysis of the stage data on the Sunja River shows no meaningful trend (Figs. 6a, b). However, the precipitation data shows a defi nite increasing trend (Fig. 7). Total annual precipitation ranges from 590 to 1160 mm. The lowest and the highest amounts of precipitation were registered in 1971 and 2002 respectively. In the 44-year period, mean annual precipitation was 893 mm. These hydrological conditions could not lead to the observed reduction of spring yield. Based on performed analyses, it could be concluded that the lowered yields in recent years are mostly a consequence of groundwater exploitation over a long time. In addition to the observed differences in the ratios of maximum to minimum yields in the “Pašino vrelo” and “Bojanića vrelo” springs, occasional variation in trends were also apparent (Fig. 5). Thus, for instance, in the period from 26 April to 28 June 2006, the yield of the “Pašino vrelo” spring increased from 11.5 l/s to 22.3 l/s. At the same time, the yield of the “Bojanića vrelo” spring decreased from 38 Figure 6: Sunja river stages at the gauging station in Sunja with a linear regression line for a) 1965−1987 and for b) 1987−2009. a) b) Larva et al.: Hydrodynamic and hydrochemical conditions at the groundwater source “Pašino vrelo”, with a focus on its development Geologia Croatica 305 l/s to 22.3 l/s. These circumstances are partly related to the impact of pumping on the discharge regime at the “Pašino vrelo” spring, but above all to the different hydrodynamic characteristics of the two springs. Corresponding yield trends of the “Bojanića vrelo” spring and hydrographs of ground- water in wells K-3 and K-9 are also signifi cant. In the peri- ods marked by a decrease in groundwater levels in the wells, the yield of the “Bojanića vrelo” spring also decreased, and vice versa – an increase in the yield of the “Bojanića vrelo” spring is followed by increased water levels in the wells. Conversely, yields of the “Pašino vrelo” spring do not show such correlation with water levels in the Quaternary alluvial aquifer. In November 2006, a pumping test was conducted on the SPB-Pv-1/06 well. During the continuous pumping rate of 21.9 l/s, it was noticed that discharge from the “Pašino vrelo” spring remained relatively unaltered at 20 l/s whereas discharge from the “Bojanića vrelo” spring was only 4 l/s. Although the spring yields were not controlled immediately prior to the start of pumping test, measurement data 13 days prior to pumping may serve for orientation purposes when discharge from the “Pašino vrelo” spring was 24 l/s, and 22 l/s from “Bojanića vrelo” . About ten hours after the pump- ing test was completed, the yield of the “Pašino vrelo” spring remained at 20 l/s, whereas the yield of the spring “Bojanića vrelo” increased from 4 l/s to 10.7 l/s. Analysis of the hydrodynamic characteristics of the Bad- enian carbonate and the Quaternary alluvial aquifers in the greater area of the groundwater source, indicate a hydraulic connection between the genetically distinct aquifers and dif- ferent groundwater discharge conditions at the “Pašino vrelo” and “Bojanića vrelo” springs. The “Pašino vrelo” spring shows greater yield stability and is more strongly in- fl uenced by infl ow from the carbonate aquifer when com- pared to the spring “Bojanića vrelo”. Conversely, the “Boja- nića vrelo” spring is more directly influenced by the hy dro dynamic conditions in the alluvial aquifer, as con- fi rmed by a very good positive correlation of yields of the “Bojanića vrelo” spring and groundwater levels in the Qua- ternary alluvial aquifer. 4. HYDROCHEMICAL CHARACTERISTICS OF GROUNDWATER During June and September 2006 as well as April, Sep- tember and October 2007, groundwater samples were taken from the wells (PVB-1, PVB-2 and PVB-3) and from the springs (the “Pašino vrelo” and “Bojanića vrelo”) at the ground water source “Pašino vrelo”, as well as from the “Grabovac” spring, which drains the groundwater from the lithothamnium limestones. Water samples were also taken from the Sunja River. Prior to sampling, the following pa- rameters were measured ‘’in situ’’ by probes of the WTW company: electrolytic conductivity (EC), total dissolved sol- ids (TDS), temperature (T), pH and oxygen content in waters to be sampled. At the Hydrochemical Laboratory of the De- partment of Hydrogeology and Engineering Geology at the Croatian Geological Survey, the concentrations of basic an- ions: chlorides, sulphates and nitrates were measured by ion chromatograph (by the LabAlliance Co.), whereas the con- centrations of orthophosphates and ammonium were meas- ured by spectrophotometer DL/2010 (by the HACH Co.). Concentrations of the basic cations: calcium, manganese, sodium and potassium were measured by atomic adsorber (Perkin Elmer Co.). The content of HCO3 - was determined by titration. Ratios of the stable isotopes of hydrogen (δD, δ2H) and oxygen (δ18O) in sampled water were measured at the JOANNEUM RESEARCH Forschungsgesellschaft mbH, Institute of Water Resources Management (WRM) Hy- drogeology and Geophysics in Graz, Austria. Water temperatures ranged from 10.7 to 21.2oC (Table 1), the smallest variations being observed in springs and wells (11 to 12oC). This indicates the mean annual tempera- ture in their recharge area. However, water temperatures in Fi gu re 7: Total annual precipita- tion at the Sisak meteorological sta tion with linear regression line for 1965−2009. Geologia Croatica 63/3Geologia Croatica 306 the surface fl ow of the Sunja indicate differences in temper- atures during colder and warmer mon ths as a consequence of cooling and warming of the watercourse. It has long been observed that temperatures in surface water bodies show di- urnal and annual oscillations, rela ted to weather conditions, particularly atmospheric tem perature (KEERY et al., 2007). The pH-values range from 7.02 to 8.26, thus the waters are neutral to mildly alkaline (Table 1). The water from the springs is well saturated with oxygen, whereas the water from the wells is poorly saturated with oxygen (Table 1). Poorer oxygen saturation of well waters is a consequence of the consumption of oxygen in the alluvial aquifer for differ- ent chemical reactions, such as the oxidation of organic mat- ter, mineral wear, nitrogen transformation processes, etc. The measured EC values range from 222 to 581 µS/cm (Table 1). Higher EC values and total dissolved solids (TDS) were measured in the water of the “Grabovac” and “Bojanića vrelo” springs as well as well PVB-3. Slightly lower values were measured in the other wells (PVB-1 and PVB-2) and in the “Pašino vrelo” spring and the lowest values were measured in the Sunja watercourse itself (Table 1). The EC and TDS values of the measured waters indicate a grouping of springs and wells according to the proportion of waters from lithothamnium limestones and alluvium, depending on hydrological conditions. During wet months (maximum and medium waters), a signifi cant infl uence of waters from the alluvial aquifer was observed in the wells PVB-1 and PVB-2 and in the springs “Bojanića vrelo” and “Pašino vrelo”, whe- reas in the drier period (minimum waters) there is a stronger infl uence from the lithothamnium limestones. In the PVB-3 well, water from the lithothamnium limestones was the dom- inant infl uence observed throughout the hydrological year. According to the basic ionic composition (Piper dia- gram), the analyzed groundwater belongs to the Ca-HCO3 water type, while the water sampled from the Sunja river belongs to the CaMg-HCO3 water type (Fig. 8). Such hydro- chemical facies of water is a consequence of the dissolution of carbonate minerals in the recharge area of the springs, wells and the river. The quantity of dissolved carbonate minerals depends on the partial pressure of CO2, i.e. the solubility of CO2 in water, since its solubility increases simultaneously with the increase of partial pressure of carbon dioxide leading to in- creased quantities of carbonic acid and greater solubility of carbonate minerals (STUMM & MORGAN, 1996; HEM, 1989). If, however, for some reason, the partial pressure of CO2 is reduced, there is a reduction in its solubility, and the opposite reaction becomes possible – precipitation of car- bonate minerals (GARRELS & CHRIST, 1965; MACPHER- SON et al., 2008). The saturation index of water is a measure which serves as an indicator of changes in balance between the solid phase (carbonate minerals) and liquid phase (groundwater). Fi gu re 9: Distribution of calculat- ed saturation indices of calcite and dolomite and logpCO2 of sampled waters. Fi gu re 8: Piper diagram of sampled waters from the study area. Larva et al.: Hydrodynamic and hydrochemical conditions at the groundwater source “Pašino vrelo”, with a focus on its development Geologia Croatica 307 Based on the results obtained by chemical analyses, par- tial pressure of CO2, indices of saturation of calcite and dol- omite are calculated by application of the geochemical mass- balance model NETPATH (PLUMMER et al., 1994). It is evident that the sampled waters are mostly saturated with respect to calcite, and unsaturated in respect of dolomite (Fig. 9). A higher (positive) partial pressure of CO2 is observed during the colder months compared to warmer months (Fig. 9). A higher partial pressure of CO2 during colder months is a consequence of the transport of organic matter from the surface of the terrain into the underground, under conditions of low soil and air temperatures, so that CO2 is not released from the soil, but dissolves in the water. In addition, there is higher precipitation (snow and rain) in the colder months, dissolving CO2 and infi ltrating it below ground. Conversely, in the warmer months, with less precipitation and higher air and soil temperature, CO2 is released from the soil, thus cre- ating a lower partial pressure of CO2 in groundwater. Due to a higher partial pressure of CO2 carbonate rocks dissolve, thus the values of calcite saturation index are lower, or in some cases, water is unsaturated with regard to calcite (e.g. “Grabovac”/09/07). Lithothamnium limestones are more sol- uble than carbonate boulders which comprise the Quaternary alluvial aquifer, thus the EC values of water from the lithoth- amnium aquifer are higher in relation to waters from the al- luvial aquifer. Similarly, understanding changes in the bal- ance between the solid phases (carbonate minerals, i.e. cal cites and dolomites) and the liquid phase (groundwater) is important for the natural control of groundwater quality. Heavy metals which show an affi nity for the carbonate min- Fi gu re 10: Plot of calculated mix- ing portions of waters from ‘’Bo ja- nića vrelo’’ and ‘’Pašino vrelo’’. Fi gu re 11: Stable isotopic compo- sition of water samples collected from the spring and wells at the ‘’Pašino vrelo’’ groundwater source relative to the meteoric water line from Klagenfurt. Geologia Croatica 63/3Geologia Croatica 308 eral phase tend to be deposited together with calcite, mostly in the form of coatings, and leach out from water (HER- RERA et al., 2008; CURKOV et al., 2008; LAKSHTANOV & STIPP, 2007; ZACHARA et al., 1988, 1991). Estimation of the proportions of water from the lithoth- amnium and alluvial aquifers in the “Pašino vrelo” and “Bo- ja nića vrelo” springs, and wells PVB-1, PVB-2 and PVB-3 was carried out using the NETPATH software (PLUMMER et al., 1994). This program was used to interpret net geo- chemical mass-balance reactions between the initial waters that mix and the fi nal waters along real fl ow paths in aqui- fers or other hydrologic systems. The program uses geoche- mical modelling techniques (PLUMMER et al., 1983; PLUM MER, 1985; PARKHURST & PLUMMER, 1993; GLYNN & BROWN, 1996; NORDSTROM, 2007) to construct geo- chemical reaction models, using chemical and isotopic data for waters from the hydrochemical system. The model is a set of mixing fractions and mole transfers that precisely ac- count for the changes in concentration of elements in the waters (PLUMMER et al., 1994). Models between selected evolutionary waters are found for every possible combina- tion of the plausible phases that can account for the compo- sition of a selected set of chemical and isotopic constraints in the system (PLUMMER et al., 1994). Simultaneous water drainage was observed at springs from both aquifers (Fig 10). However, the proportions change depending on hydrological conditions. Thus, in the conditions of lower and medium waters (September 2006, April 2007 and September 2007), the prevailing groundwa- ter component is from the carbonate aquifer. Conversely, during high waters, groundwater from the alluvial aquifer prevails (June 2006 and October 2007). From the measured ratios of stable oxygen and hydro- gen isotopes, it is evident that the waters from the “Pašino vrelo”, “Bojanića vrelo” and “Grabovac” springs, and the wells PVB-1, PVB-2 and PVB-3, are directly infl uenced by renewal from recent precipitation, since the values are evenly dispersed along the local meteoric water line (LMWL) – Kla- genfurt (Fig. 11). Changes in the δ18O ratio of monitored waters are small, indicating a complex hydrogeological system in the recharge zone of the springs and wells of the “Pašino vrelo” ground- water source. The recharge area of the Sunja river is on Mt. “Zrinska gora” (Fig. 1), where altitudes range from about 178 m a.s.l to about 600 m a.s.l. Depending on the hydro- logical conditions (precipitation quantity and season), the measured δ18O values were more negative (-9.96 o/oo) when the Sunja is infl uenced by tributaries from the higher alti- tudes and colder precipitation, while they were more posi- tive (-9.36 o/oo) in reverse conditions (Fig. 12). The Sunja recharges the alluvial aquifer in its valley, and thus infl uences the isotopic composition of water in wells and springs (Fig. 12). However, at times of low waters, the isotopic composi- tion of spring waters is infl uenced by groundwater from the lithothamnium limestones. A similar scenario was observed between the waters from the “Grabovac” and “Bojanića vrelo” springs. At times of high waters, both springs have the same recharge area, i.e. the hilly parts of Mt. “Zrinska gora”. However, the δ18O values measured at times of low waters indicate the recharge of these springs from higher al- titudes. Furthermore, the measured negative δ18O values in water samples from the “Pašino vrelo” spring indicated that its recharge area is larger, encompassing not only the hilly parts of the Mt. “Zrinska gora” area but also Mt. “Zrinska gora” itself. At higher altitudes, where the average tempera- tures are lower, precipitation is isotopically depleted (CLARK & FRITZ, 1997). This altitude effect is useful in hydrogeological studies, as it distinguishes groundwater re- Fi gu re 12: Stable isotopic compo- sition of waters sampled from Gra- bovac, Sunja, ”Bojanića vrelo’’ and ‘’Pašino vrelo’’. Larva et al.: Hydrodynamic and hydrochemical conditions at the groundwater source “Pašino vrelo”, with a focus on its development Geologia Croatica 309 Table 1: Results of measured physical, physico-chemical and chemical parameters of the sampled waters. EC (µS/cm) TDS (mg/l) T (°C) pH O2 (%) Ca2+ (mg/l) Mg2+ (mg/l) Na+ (mg/l) K+ (mg/l) HCO3 – (mg/l) SO4 2– (mg/l) Cl– (mg/l) NO3 – (mg/l) NH4 + (mg/l) PO43-- P(mg/l)  PVB-1-06-06 367 256 11.4 7.47 55 63.9 4.2 4.1 1.7 203 11.5 1.9 4.9 <0.01 0.03 PVB-1-09-06 361 252 12.5 7.56 60 82 4.2 9.9 3.3 274 11.3 3.2 4.3 <0.01 0.04 PVB-1-04-07 340 238 10.8 7.74 50 69 4.8 12.4 2.1 240 12.7 3.5 4.4 <0.01 0.05 PVB-1-09-07 375 263 12.2 7.48 48 70 5.4 13.2 2.3 248 14.4 4.6 4.8 0.01 0.02 PVB-1-10-07 366 256 12.9 7.59 57 64.2 4.7 13.2 1.9 246 14.4 4.3 4.3 <0.01 0.04 PVB-2-06-06 366 256 11.3 7.47 47 66.1 5.7 6.7 1.7 203 11.8 1.7 4.5 0.07 0.05 PVB-2-09-06 360 252 12.5 7.56 59 85 5.7 9.8 3.3 274 11 2.8 3.9 0.01 0.04 PVB-2-04-07 338 237 11.7 7.5 70 69.5 4.7 12.5 2.1 240 12.7 3.5 4.3 <0.01 0.03 PVB-2-09-07 379 265 12.4 7.53 54 70 5.3 13.1 2.3 248 14.6 4.7 5.0 0.02 0.06 PVB-2-10-07 367 257 12.7 7.5 50 63.8 4.7 13.3 2.1 245 15.9 3.2 4.5 0.01 0.05 PVB-3-06-06 409 286 11.5 7.34 52 77.4 6.1 5.9 1.4 254 9.6 2.9 4.8 0.03 0.05 PVB-3-09-06 407 285 11.8 7.5 60 92 6.1 8.9 2.7 303 9.4 6.8 4.3 <0.01 0.05 PVB-3-04-07 397 277 11.7 7.28 50 64.8 5.6 14.4 1.6 238 9.7 6.2 4.3 <0.01 0.05 PVB-3-09-07 419 293 12.2 7.55 43 74.3 5.8 14.4 2.2 252 10.6 4.7 4.7 0.02 0.06 PVB-3-10-07 400 291 12 7.52 52 61.1 5.3 14.6 1.8 238 11.8 5.6 4.5 0.02 0.05 “Pašino vrelo”-06-06 357 250 11 7.2 72 67.9 5.4 4.4 1.7 214 12 2.3 4.5 <0.01 0.04 “Pašino vrelo”-09-06 355 249 11.8 7.42 57 90 5.4 10 3.3 277 10.7 4.3 3.8 <0.01 0.04 “Pašino vrelo”-04-07 330 231 11.3 7.47 60 67.2 4.8 12.8 2.1 240 12.5 5.6 4.2 <0.01 0.04 “Pašino vrelo”-09-07 375 263 11.8 7.43 50 60.3 4.8 13.4 2.3 240 15.6 2.7 5.3 0.02 0.04 “Pašino vrelo”-10-07 365 256 11.9 7.49 43 61.2 4.2 13.7 2.2 238 17.3 3.4 4.7 <0.01 0.04 “Bojanića vrelo”-06-06 381 266 13.3 7.39 53 68 5.5 7.5 1.7 240 9.6 4.2 5.1 0.08 0.03 “Bojanića vrelo”-09-06 397 278 12 7.42 78 95 5.5 11.1 2.7 283 8.1 8.4 4.1 <0.01 0.05 “Bojanića vrelo”-04-07 375 262 11.3 7.52 66 69.3 4.5 15.8 1.9 242 9.1 9.7 4.9 <0.01 0.05 “Bojanića vrelo”-09-07 402 281 11.4 7.48 55 64.7 5 14.9 2 245 9.5 4.8 4.9 0.01 0.02 “Bojanića vrelo”-10-07 406 284 10.7 7.57 80 60 4.8 15.9 1.9 238 10.1 6.6 4.9 0.02 0.07 “Grabo- vac”-06-06 581 406 11.3 7.42 84 124 1.1 1.6 0.7 391 5.3 1.3 3.1 0.03 0.01 “Grabo- vac”-09-06 570 399 11.2 7.32 96 124 0.8 2.6 0.8 396 4.9 2.4 3.3 <0.01 <0.01 “Grabo- vac”-04-07 552 386 13.2 7.62 95 75.8 1.2 6.2 0.5 250 4.5 1.5 2.7 <0.01 0.04 “Grabo- vac”-09-07 446 372 12 7.23 80 86.4 0.7 5 0.6 262 4.1 1.2 2.6 <0.01 0.01 “Grabo- vac”-10-07 550 385 11.3 7.33 78 93.8 1.7 6.5 0.5 278 4.6 1.3 2.9 <0.01 0.07 The Sunja river-06-06 291 203 21.2 7.8 103 49 8 4.6 1.5 178 13.4 1.8 3.4 <0.01 0.01 The Sunja river-09-06 311 218 15.7 8.26 121 74 8 10 2.9 178 11 4.3 3 <0.01 0.01 The Sunja river-04-07 246 172 15.4 7.9 116 49.4 7.8 12.9 1.7 170 14.3 4.3 2.8 <0.01 0.02 The Sunja river-09-07 222 155 13.3 7.86 100 31.7 3.3 12.9 2.1 100 16.2 4.9 8.8 0.03 0.02 The Sunja river-10-07 307 215 12.6 7.69 129 44.8 4.3 14.3 1.9 160 17.9 4.0 3.0 0.02 0.04 Geologia Croatica 63/3Geologia Croatica 310 charged at high altitudes from that recharged at low altitude (CLARK & FRITZ, 1997; ROZANSKI et al., 1993). Concentrations of ammonium and orthophosphates in water samples are below the MPC values (Table 1). How- ever, increased concentrations of ammonium and orthophos- phates during washing out of upper parts of alluvial deposits were observed in the wells. Increased concentrations of am- monium and orthophosphates indicate the infl uence of do- mestic wastewater (DOYLE & PARSONS, 2002; CELEN et al., 2007). A sewerage system in the settlements surround- ing the groundwater source has not yet been constructed, so the inhabitants discharge their wastewater into canals along the road and, during precipitation, wastewater from the ca- nals is infi ltrated into underground. In the “Grabovac” spring, increased concentrations of ammonium and orthophosphates are occasionally registered. However, their origin is natural, i.e. caused by the decomposition of organic matter (leaves, twigs, etc.) on the forest fl oor. Concentrations of nitrates in all water samples are well below the MPC values (Table 1). Agricultural activity is not intensive in the area (mostly meadows and pastures, less fi elds), thus there is no possibility of groundwater pollution by manure or mineral fertilizers. Similarly, concentrations of sulphates and chlorides are also well below the MPC val- ues (Table 1). The origin of sulphates and chlorides in groundwater is natural, since the aquifers are composed of rocks which contain chloride and sulphate or sulphide bear- ing minerals. It can be generally stated that the waters of the “Pašino vrelo” groundwater source are of good quality in terms of chemical indicators. However, microbiological indicators suggest groundwater pollution with domestic wastewater, as confi rmed by occasionally raised concentrations of ammo- nium and orthophosphates. 5. DEVELOPMENT OF THE GROUNDWATER SOURCE In 2005, an exploratory borehole (SPBPv-1/05) was drilled north-west of the pumping site, followed a year later by a test exploitation well SPBPv-1/06, with the aim of ensuring the availability of additional quantities of groundwater (Fig. 3). The well was 152 m deep, and the aquifer was tapped in the 50−105 m and 120−137 m intervals. During the pump- ing test with a quantity of 21.9 l/s, a drawdown of 3.95 m was achieved. An exploitation capacity of at least double this is expected, which could not be confi rmed during the pump- ing test, since a pump of adequate capacity was not avail- able. Groundwater samples were taken during testing for chemical analyses in order to determine the proportions of groundwater from different parts of the aquifer system. 83 % of groundwater was from the lithothamnium limestone with only 17% from the alluvial aquifer, which is signifi - cantly more favourable in relation to the other facilities at the pumping site. The results proved that tapping deeper parts of the carbonate aquifer was justifi ed. The sustainabil- ity of these conditions will be verifi ed in the course of long- term exploitation of the well. To date, protection of the groundwater source has been carried out through active and passive measures according to the currently unenforced Regulation from 1986 (OG 22/86). The expert maps for update, which also includes the new well SPBPv-1/06, according to the current Regulation in force on the determination of sanitary protection zones of the drinking water source (OG 55/02), is given by MRAZ et al. (2007)1. The specifi c characteristics of the hydrogeolog- ical system as previously described, were taken into consid- eration when defi ning protection zones. They played an im- portant role in the defi nition of the II. protection zone, along the external boundary of which, according to the Regulation, groundwater has a minimal travel time of 50 days before it enters groundwater capture. When determining this protec- tion zone, the hydrogeological characteristics of the carbon- ate aquifer, (which has a primary signifi cance from the water supply standpoint), were not taken into account. The focus was placed on the Quaternary alluvial aquifer, since it was assumed that the values of hydraulic conductivity, and thus also of velocity of groundwater fl ow in the alluvial aquifer were higher in relation to the carbonate aquifer. Namely, the interpretation of the pumping test results showed that the hydraulic conductivity of the lithothamnium limestone is 3.6 m/day. Based on lithological composition of the Quaternary alluvial aquifer, data from reference sources and analogy with other areas, a hydraulic conductivity of 150 m/day was estimated. According to the hydrogeological model, pollu- tion from the surface of the terrain leaches vertically to the saturated zone of the alluvial aquifer and then, under the in- fl uence of the gradient, advances by means of advection to- wards the well fi eld. Bearing in mind the aforementioned conditions, the gradient achieved during the pumping site operation and an estimated value of alluvial aquifer effective porosity of 0.2, the average fl ow velocity in horizontal fl ow was defi ned as 7.5 m/day. In the course of this, the radial fl ow of groundwater in the vicinity of the wells was ne- glected. Given the fl ow velocity and the minimum permis- sible groundwater travel time in an area included in the II. protection zone, it was possible to defi ne its external bound- ary (Fig. 3). It was especially convenient for the part of al- luvial aquifer located up-gradient from the pumping site, 1 MRAZ, V., MARKOVIĆ, T. & LARVA, O. (2007): Izvorište “Pašino vrelo”. Zaštitne zone za novi zdenac SPBPV-01/06 i vodocrpilište “Pašino vrelo” [Groundwater source “Pašino vrelo”. Protection zones for new well SPBPV-01/06 and groundwater source “Pašino vrelo” – in Croatian].– Unpubl. report, Croatian Geological Survey, Zagreb, 44 p. Larva et al.: Hydrodynamic and hydrochemical conditions at the groundwater source “Pašino vrelo”, with a focus on its development Geologia Croatica 311 where there is no natural boundary to groundwater fl ow. The external boundary of the II. protection zone was subse- quently adjusted in order to correspond to the boundaries of cadastral plots. In the north it follows the Sunja river and the Bekej watercourse in the east and south-east. In the west and south-west it follows the natural boundary of the alluvial aq- uifer and in the north-west, up-gradient from the pumping site, it is located approximately 450 m from the pumping site which is in accordance with the minimum permissible groundwater travel time. 6. CONCLUSION The groundwater accumulated in the carbonate aquifer of Badenian age presents an exceptionally valuable natural re- source in the municipality of Hrvatska Kostajnica. The focus of the recent investigations of the “Pašino vrelo” aquifer sys- tem was placed on the defi nition of its hydrodynamic and hydrochemical characteristics. The results have increased the level of understanding of the hydrogeological model, which in turn enabled better insights into the possibilities of abstraction of additional drinking water quantities of more favourable quality, and adequate protection of the ground- water source from pollution. The analysis of oscillation of groundwater levels and spring discharges at the groundwater source enabled conclu- sions to be drawn about the hydrodynamic conditions in the aquifer system. A positive correlation between oscillations of groundwater levels in the Quaternary alluvial aquifer and the discharge quantities in the “Bojanića vrelo” spring proved the hydraulic connection between the carbonate and alluvial aquifers. This had already been indicated by the bac- teriological contamination of groundwater abstracted at the pumping site. At the same time, the difference in trends of groundwater levels in the alluvial aquifer and water levels in the “Pašino vrelo” spring, as well as in trends of yields in the “Pašino vrelo” and “Bojanića vrelo” springs pointed to different discharge conditions at the groundwater source. The same conclusion is indicated by the stable discharge of the “Pašino vrelo” spring, which did not alter even during the pumping test of the SPBPv-1/06 well while, in contrast, the discharge of the “Bojanića vrelo” spring was substantially reduced. Such hydrodynamic indicators are, to a point, a consequence of the infl uence of exploitation wells (PVB-1, PVB-2 and PVB-3) in operation at the pumping site, but, above all, of a more pronounced groundwater drainage from the carbonate aquifer to the “Pašino vrelo” spring, whereas the “Bojanića vrelo” spring is more strongly infl uenced by the Quaternary alluvial aquifer. Calculation of the propor- tions of water from the lithothamnium and alluvial aquifers in the “Pašino vrelo” and “Bojanića vrelo” springs, as well as in the PVB-1, PVB-2, PVB-3 and SPBPv-1/06 wells, in- dicates larger proportion of water from the lithothamnium aquifer during low and medium water fl ow at monitored springs and wells, while during high waters there is a larger proportion of water from the alluvial aquifer. Groundwater quality in the recharge area of the ‘’Pašino vrelo’’ groundwater source is good with regards to its chem- ical indicators. However, microbiological indicators indicate groundwater pollution with domestic wastewater during high waters as confi rmed by occasionally increased concentra- tions of ammonium and orthophosphates, which are chemi- cal indicators of the impact of domestic wastewater. The mass-balance model confi rmed the hydraulic connection of the genetically different aquifers. Results of the investigations suggested exploration of the possibility of extracting drinking water of adequate qual- ity from the deeper horizons of the Badenian carbonate aq- uifer. Initial results are promising. Besides the fact that the expected exploitation capacity of the well exceeds water ab- straction quantities at the pumping site to date, a favourable ratio of the proportions of water from the carbonate and al- luvial aquifers was also determined. If such a ratio remains under conditions of long-term exploitation, a more favour- able quality of groundwater in relation to the other wells at the pumping site is guaranteed. While defi ning the protection zones for the groundwater source, the specifi c characteristics of the hydrogeological model were taken into consideration, particularly during def- inition of the II. zone of sanitary protection. Due to the hy- draulic connection between the alluvial and carbonate aqui- fers, the focus was placed on hydraulic characteristics of the alluvial aquifer when determining the size of the zone. By remedying the anthropogenic infl uences on groundwater quality in the alluvial aquifer, the existing water quality prob- lems at the pumping site will be resolved. ACKNOWLEDGEMENT The authors wish to express their gratitude to Croatian Waters for the fi nancial support of the research and State Meteorological and Hydrological Service for providing hydrologic data. The authors also wish to thank anonymous reviewers for their very useful com- ments. This paper represents part of a hydrogeological investiga- tion within the project entitled the “Basic Hydrogeological Map of the Republic of Croatia” (project no. 181-1811096-3165) which is fi nanced by the Ministry of Science, Education and Sports. REFERENCES CELEN, I., BUCHANAN, J.R., BURNS, R.T., ROBINSON, R.B. & RAMAN, D.R. (2007): Using a chemical equilibrium model to pre- dict amendments required to precipitate phosphorus as struvite in liquid swine manure.− Water Res., 41, 1689–1696. CLARK, I. & FRITZ, P. (1997): Environmental Isotopes in Hydrogeol- ogy.− Lewis Publ. CRC Press, 238 p. CURKOV, N., CMUK, P., MLAKAR, M. & OMANOVIĆ, D. (2008): Spatial distribution of trace metals in the Krka River, Croatia. An example of the self-purifi cation.− Chemosphere, 72, 1559−1566. DOYLE, J.D. & PARSONS, S.A. (2002): Struvite formation, control and recovery.− Water Res., 36, 3925–3940. GARRELS, R.M. & CHRIST, C.L. (1965): Solutions, Minerals and Equ- libria.− Harper & Row, New York. GLYNN, P.D. & BROWN, J.G. (1996): Reactive transport model ing of acidic metal-contaminated ground water at a site with sparse spatial information.− In: Reactive transport in porous media General prin- ciples and application to geochemical processes, Reviews in Min- Geologia Croatica 63/3Geologia Croatica 312 eralogy.− In: STEEFEL, C.I., LICHTNER, P. & OELKERS, E. (eds.): Mineralogical Society of America, 34 /9, 377–438. HEM, J.D. (1989): Study and Interpretation of the Chemical Character- istics of Natural Waters.− U.S.G.S Water-supply paper 2254, 263 p. HERRERA, M.L., ESPINOSA, R.J., MILAN, J.J. & HISCOCK, K.M. (2008): Integrated hydrochemical assessment of the Quaternary al- luvial aquifer of the Guadalquivir river, southern Spain.− Applied Geochemistry, 23, 2040−2054. KERRY, J., BINELY, A., CROOK, N. & SMITH, J.W.N. (2007): Tem- poral and spatial variability of groundwater – surface water fl uxes: Development and application of an analytical method using tem- perature time series.− J. Hydrology, 336, 1−16. LAKSHTANOV, L.Z. & STIPP, S.L.S. (2007): Experimental study of nickel(II) interaction with calcite: Adsorption and coprecipitation.− Geochimica et Cosmochimica Acta, 71, 3686−3697. MACPHERSON, G.L., ROBERTS, J.A., BLAIR, J.M., TOWNSEND, M.A., FOWLE, D.A. & BEISNER, K.R. (2008): Increasing shal- low groundwater CO2 and limestone weathering, Konza Prairie, USA.− Geochimica et Cosmochimica Acta, 72, 5581−5599. MAGDALENIĆ, A., JURAK, V., KOVAČEVIĆ, S. & CAPAR, A. (1976): Diffi culties with groundwater tapping in Sunja valley. 4th Yugoslavian Symposia on Hydrogeology and Engineering Geology – Book 1, Skopje, 223−234. NORDSTROM, D.K., (2007): Modeling low-temperature geochemi cal processes.− In: DREVER, J.I. (ed.): Treatise on Geochemistry, Vol- ume 5−Surface and ground water, weathering and soils: Elsevier, 1–38. OG 22/86: Regulation on protection measures and conditions for deter- mination of sanitary protection zones of the drinking water source. Narodne novine (1986): Pravilnik o zaštitnim mjerama i uvjetima za određivanje zona sanitarne zaštite izvorišta za piće. No 22, 6 p. OG 47/08: Regulation on health safety of drinking water. Narodne no- vine (2008): Pravilnik o zdravstvenoj ispravnosti vode za piće. No 47, 16 p. OG 55/02: Regulation on the determination of sanitary protection zones of the drinking water source. Narodne novine (2002): Pravilnik o utvrđivanju zona sanitarne zaštite izvorišta. No 55, 9 p. PARKHURST, D.L. & PLUMMER, L.N. (1993): Geochemical mod- els.− In: ALLEY, W.M. (ed.): Regional Ground-Water Quality: Van Nostrand Reinhold, chap. 9, 199–225. PLUMMER, L.N. (1985): Geochemical modelling − A comparison of forward and inverse methods.− In: HITCHON, B. & WAL LICK, E.I. (eds.): Canadian/American Conference on Hydroge ology, 1st, Banff, Albe rta, June 1984, Proceedings: National Water Well As- socociation, 149–177. PLUMMER, L.N., PARKHURST, D.L. & THORSTENSON, D.C. (1983): Development of reaction models for groundwater systems.− Geochimica et Cosmochimica Acta, 47, 665–686. PLUMMER, L.N., PRESTON, E.C. & PARHURST, D.L. (1994): An interactive code (NETPATH) for modelling net geochemical reac- tions along fl ow path. Version 2.0 USGS Water-Resources Investi- gation Report, 94−4169, Reston, Virginia. ROZANSKI, K., ARAGAUS-ARAGAUS, L. & GONFIANTINI, R. (1993): Isotopic patterns in modern global precipitation.− Conti- nental Isotope Indicators of Climate, Amer. Geo. Union Mono- graph. CELEN, I., BUCHANAN, J.R., BURNS, R.T., ROBINSON, R.B. & RAMAN, D.R. (2007): Using a chemical equilibrium model to pre- dict amendments required to precipitate phosphorus as struvite in liquid swine manure.− Water Res., 41, 1689–1696. STUMM, W. & MORGAN, J.J. (1996): Aquatic chemistry. Chemical Equlibria and Rates in Natural Waters.− John Wiley & Sons, INC. New York-Chichester-Brisbane-Toronto-Singapore, 1022 p. ZACHARA, J.M., COWAN, C.E. & RESCH, C.T. (1991): Sorption of divalent metals on calcite. Geochim. Cosmochim. Acta, 55, 1549– 1562. ZACHARA, J.M., KITTRICK, J.A. & HARSH, J.B. (1988): The mech- anism of Zn2+ adsorption on calcite.− Geochim. Cosmochim. Ac- ta, 52, 2281–2291. Manuscript received September 02, 2009 Revised manuscript accepted June 25, 2010 Available online November 11, 2010