Groundwater Ages and Flow Paths at a Coastal Waste Repository Site in Korea, Based on Geochemical Characteristics and Numerical Modeling

Abstract

Groundwater flow paths and groundwater ages at a radioactive waste repository located in a coastal area of South Korea were evaluated using the hydrochemical and hydrogeological characteristics of groundwater, surface water, rain water, and seawater, as well as by numerical modeling. The average groundwater travel time in the top layer of the model, evaluated by numerical modeling and groundwater age (34 years), approximately corresponds to the groundwater age obtained by chlorofluorocarbon (CFC)-12 analysis (26-34 years). The data suggest that the groundwater in wells in the study area originated up-gradient at distances of 140-230 m. Results of CFC analyses, along with seasonal variations in the δ¹⁸O and δD values of groundwater and the relationships between ²²²Rn concentrations and δ¹⁸O values and between ²²²Rn concentrations and δD values, indicate that groundwater recharge occurs in the summer rainy season and discharge occurs in the winter dry season. Additionally, a linear relationship between dissolved SiO₂ concentrations and groundwater ages indicates that natural mineralization is affected by the dilution of groundwater recharge in the rainy summer season.

Introduction

The Fukushima Nuclear Power Plant accident on 11 March 2011, a result of the Tohoku earthquake, alerted people to the risk of nuclear power plant failures and prompted several countries, including Germany and Japan, to initiate reductions in nuclear power plant capacity and operations. However, regardless of the risks involved, all nations that produce electric power from nuclear sources must address problems associated with the treatment and storage of radioactive waste released from nuclear power plants. In Korea, four nuclear power plants produce ongoing radioactive waste, and repositories for the storage of such waste are actively under development.

Both hydrochemical and hydrogeological characteristics of groundwater must be investigated carefully to ensure the safety of radioactive waste disposal sites. Cherry et al. (1974) developed hydrogeological criteria to evaluate the suitability of low-level radioactive waste burial sites (Cherry et al., 1974), and Papadopulos and Winograd (1974) reviewed the status of mathematical simulation techniques and major hydrogeological and hydrochemical factors relevant to radioactive waste storage systems and facilities. For the excavation of underground facilities for the storage of radioactive waste and petroleum, hydrochemical principles have been effectively applied to understand groundwater types, groundwater residence times, water–rock interaction processes, provenances of major groundwater bodies, and potential perturbations to hydrochemical systems (IAEA, 1999). IAEA (2001) presented a conceptual groundwater flow model for shallow radioactive waste disposal facilities. Additionally, Yi et al. (2012) evaluated a near-surface disposal site for low- and intermediate-level radioactive waste (LILW), using a geological/hydrogeological survey, geophysical explorations, in situ and laboratory tests, and numerical modeling.

In an eastern coastal area near Gyeongju City, Korea, an underground repository for LILW was completed in June 2014. The repository, which had been under construction since July 2008 (Park et al., 2008), consists of six silos at depths of 80–130 m below sea level, containing 100,000 drums of LILW. Yoon et al. (2010) analyzed 14C and tritium concentrations in the repository area, and Park et al. (2008) conducted a numerical simulation of groundwater flow using a discrete fracture network model to predict the groundwater flow system around the repository silos and in the greater Gyeongju area. Oh and Kim (2008) simulated groundwater flow, and salt and radionuclide transport in the vicinity of the Gyeongju LILW disposal site before and during construction, as well as during operation and after closure of the repository, using COFAT3D software (Kim and Yeh, 2004). Seo et al. (2009) reviewed the status of a probabilistic seismic hazard analysis of nuclear power plant sites throughout Korea.

The present study identified groundwater flow patterns and ages at the radioactive waste disposal site in the coastal area of Gyeongju City, South Korea (Fig. 1). The groundwater flow was in a quasi-steady state prior to construction of the repository, as determined by numerical modeling and chemical analyses of groundwater, surface water, rain water, and seawater. Samples of each type of water were collected and analyzed for concentrations of major ions, oxygen and hydrogen isotopes, chlorofluorocarbons (CFCs), and radon-222 (Table 1).

Fig. 1.Map of the study area on the east coast of Korea, showing collection localities and water types.

Table 1.Results of chemical analyses of water samples.

 

Hydrogeological setting

Geologically, the study area (Fig. 2) consists mainly of Cretaceous sedimentary strata (greenish-gray or light-gray sandstone and greenish-gray or dark-gray shale), Tertiary granites (granodiorite, biotite granite, and diorite), and Tertiary volcanic rocks (rhyolite and porphyritic dacite). The study area is mostly at an elevation of 200–750 m, and is located in a coastal area that slopes gently from west to east. Streams in the study area, such as Daejong Stream and Nasan Stream, follow the slope of the terrain and flow to the east into the East Sea.

Fig. 2.Geological map of the study area (modified from K.H.N.P. Corporation, 2006; Chwae et al., 1988; Hwang et al., 2007).

Hydrogeologically, the study area is divided into three domains: a hydraulic conductor domain (HCD), a hydraulic rock domain (HRD), and a hydraulic soil domain (HSD) (Bieniawski, 1979). The HCDs represent major fracture zones that act as groundwater flow conduits. Their locations in the study area, which were identified from fracture trace investigations, drill data, geophysical logs, and a geophysical survey, suggest the presence of three strike-slip fault zones (Z21, Z23, and Z31) trending approximately E–W and five inferred fault zones (Z22, Z32, and F31–F33) trending NE–SW, NW–SE, and N–S, respectively (K.H.N.P. Corporation, 2006). The HRD is defined as low-permeability rock (consisting in the study area of unweathered granite and sedimentary rocks) located between the HCDs. The HSD, which forms a shallow groundwater reservoir in the study area, includes alluvium, upper weathered and fractured bedrock zones, coastal terrace sediments and landfill.

The groundwater flows from mountainous areas in the west to coastal areas in the east. The flow was evaluated based on a spring- and groundwater-level distribution map produced in April 2008, prior to construction of the underground repository (Fig. 3). Major fault and fracture zones, generally with E–W, N–S, NW–SE, and NE–SW trends, were considered as conduits of groundwater flow, along with a small-scale fracture zone in Cretaceous sedimentary rock. A dike striking N10°W and dipping 65°NE was also thought to be a groundwater conductor.

Fig. 3.Distribution of potentiometric head in the study area during the wet season.

The groundwater flow system is linked to numerous components of the water cycle, including precipitation, infiltration, groundwater flow through geological formations, and discharge to the surface and sea. The distribution of hydraulic conductivity in bedrock was determined by a kriging technique, using an exponential semivariogram. Using this approach, the hydrogeological model was conceptualized in terms of water infiltration in the hydraulic soil domain (HSD), primary groundwater flow through pervious fractures in bedrock (HCD), slow groundwater flow through minor fractures in bedrock (HRD), and discharge to streams and to the sea (Fig. 1). At the surface, the model partitions a portion of the precipitation to flow directly into stream channels and eventually to the East Sea.

Groundwater recharge occurs throughout the study area, except in coastal areas adjacent to the shoreline, while groundwater discharge occurs mainly along the coastline and along small streams within the study area.

 

Methods

Chemical analysis of inorganic constituents

Groundwater samples were collected at seven locations (RB-1, KB-2, KB-3, DB1-2, DB1-6, SS-5, and IJ) in the high-recharge zone (RB-1), intermediate zone (KB-2, KB-3, SS-5), and low-discharge zone (DB1-2, DB1-6, IJ) of the study area, along with samples of rain water (RW), surface water (SW), soil water (RB1-S, DB1-6S), and seawater (G) (Fig. 1). Water temperature, pH, electrical conductivity (EC), and alkalinity were measured in the field. Water samples for cation and anion analyses were filtered using a 0.45 μm Durapore membrane filter (Millipore Co.); 0.05 N nitric acid was added to the filtered samples to stabilize cation species at pH < 2. The filtered samples for cation and anion analyses were then stored at 4℃ for transport to the laboratory. Water samples were also collected for determinations of CFCs and analyses of oxygen and hydrogen isotopes, and Rn-222. The collection of water samples for cations (100 mL), anions (100 mL), CFCs (1 L), and oxygen and hydrogen isotopic analyses (1 L) followed the Guideline for Groundwater Quality Conservation Works. Water temperature, pH, EC, and alkalinity were measured using a digital thermometer (Sato Co. Model SK-250WP), pH/ORP meter (Orion Co. Model 250A), EC/TDS/salinity meter (Orion Co. Model 115), and photometer (Lovibond Co. Multi Direct), respectively.

Cations (Na+, K+, Ca2+, Mg2+, SiO2, Al3+, Fe2+, Mn2+, and Zn2+) were analyzed using an inductively coupled–plasma atomic emission spectrometer (IPC-IRIS; Thermo Jarrell Ash) and anions (SO42–, Cl–, NO3–, F–) were analyzed by ion chromatography (DX-500, Dionex). Ion balance errors for the analyses were mostly within ± 5%. All Chemical analyses were performed at the Busan Branch of the Korea Basic Science Institute (KBSI), Busan, Korea.

Oxygen-18 (18O) and deuterium (2H) isotope and CFC analyses

After collecting 0.2 ml of each type of water sample, oxygen-18 (18O) was measured using an isotope ratio mass spectrometer, utilizing the H2O–CO2 equilibrium method with CO2 gas at 25.0℃ ± 0.1℃ (Epstein and Mayeda, 1953). The CO2 gas was extracted and cryogenically purified. Deuterium (2H) was analyzed using an isotope ratio mass spectrometer and an on-line pre-treatment system (Pyr-OH model, GV Instruments). Metallic Cr was used to produce H gas using an automatic on-line sample preparation system (Morrison et al., 2001).

Analyses of 18O and 2H isotopic compositions of water samples were performed by stable isotope ratio mass spectrometer (Isoprime, GV Instruments) at KBSI. The analytical reproducibility of the measurements was ± 0.1‰ for δ18O values and ± 0.5‰ for δD values.

Groundwater samples for analyses of CFCs were collected in triplicate and analyzed for CFC-11, CFC-12, and CFC-113 using a closed system purge and trap gas chromatograph with an electron capture detector, at the Korea Institute of Geoscience and Mineral Resources (Koh et al., 2007). More detailed information regarding the procedures can be found in Koh et al. (2012).

Measurement of Rn-222 concentrations

Concentrations of 222Rn in groundwater and seawater were measured using a radon detector, according to the method of Lee and Kim (2006). Water samples were collected in 1 L bottles for groundwater analysis and in 4 L bottles for seawater analyses, taking care to prevent leakage of 222Rn from the bottles and the generation of air bubbles. The sample bottle was connected to an electronic radon detector (RAD-7, Durridge) by an air tube on the cap of the sample bottle. Air was injected into the sample bottle from the RAD-7 to eject the 222Rn, which was sent to the RAD-7 as a passing desiccant so as to measure the concentration of 222Rn. The 222Rn concentration was measured 3 or 4 times until a state of equilibrium (within an error range of ± 5%) was achieved. The 222Rn concentration was measured within 4 days of sampling. The accuracy of the 222Rn concentration measurements was 8 CPM/Bq/m3 in detection mode and 16 CPM/Bq/m3 in general mode.

The ratio of the 222Rn concentration in groundwater to that in air in a closed equilibrium state is as follows (Weigel, 1978):

where T is the water temperature in degrees Celsius. Given a constant water temperature and the exact volume of water in the sample bottle, the activity of 222Rn can be calculated as the sum of the 222Rn activity in air (Ca, Bq/L) and the 222Rn activity in the water (Cw, Bq/L). The 222Rn activity was partitioned by Weigel’s eq. (2), computed as follows (Lee and Kim, 2006):

where Vw and Va are the volumes of water and air, respectively. From eq. (2), Cw can be determined from

Numerical modeling

Numerical modeling was conducted to evaluate the distribution of head, groundwater flow paths, and groundwater travel times in the study area, and to solve for the concentrations of solute in groundwater. The commonly used groundwater model MODFLOW (McDonald and Harbaugh, 1988) was incorporated into the Visual MODFLOW package or Groundwater Modeling System (GMS) software. MODFLOW can treat anisotropic and heterogeneous 3-D flows with constant density, using a finite difference method. The Visual MODFLOW package (version 4.2 by Schlumberger) was used to simulate groundwater flow paths and groundwater ages in the study area, assuming assumed piston flow (no dispersion).

 

Results and discussion

General chemical features of the waters

Inorganic constituents (Na+, K+, Ca2+, Mg2+, SiO2, Al3+, Fe2+, Mn2+, Zn2+, HOC3-, SO42–, Cl–, NO3–, and F–) were analyzed in groundwater (17 times), surface water (10 times), seawater (1 time), and rain water (8 times) from April 2009 to February 2010. Based on the distribution of data on a trilinear diagram, rain water largely belongs to the Ca-HCO3 water type, seawater to the Na-Cl type, and surface water to the Ca-SO4 type. Groundwater in the high-elevation zone (RB-1) belongs mostly to the Ca-HCO3 and Ca-SO4 types, while groundwater of the intermediate-elevation zone (SS-5, KB-2, KB-3) and groundwater of the lowland zone (DB1-2, DB1-6, IJ) belong mostly to the Na-HCO3 type. These groundwater types indicate the influence of rain water on groundwater in elevated land areas and the influence of seawater on groundwater in intermediate and lowland areas. However, the trends in chemical concentrations and EC values are indistinct in the higher-elevation lands and lowlands, possibly reflecting the variable chemical composition of groundwater obtained from wells of variable depths (100–232 m; Fig. 4).

Fig. 4.Box-and-whisker plot of major groundwater constituents in high-elevation recharge zones, intermediate-elevation zones, and lowland discharge zones.

Concentrations of Na+ in the groundwater samples were in the range of 2.54–44.67 mg/L, with an average of 22.09 mg/L; concentrations of Na+ in surface water were in the range of 5.20–21.53 mg/L, with an average of 11.32 mg/L; and concentrations of Na+ in rain water were in the range of 0.31–3.88 mg/L, with an average of 1.41 mg/L (Tables 2 and 3). Concentrations of Ca2+ in groundwater, surface water, and rain water were in the ranges of, respectively, 5.33–56.89, 10.92–47.88, and 0.71–3.81 mg/L, with average values of 16.34, 21.96, and 1.57 mg/L. Concentrations of Ca2+ in groundwater and surface water were similar, and higher than in rain water, which indicates the influence of natural mineral water interactions and/or artificial contamination. Concentrations of SO42- in groundwater, surface water, and rain water samples were in the range of, respectively, 5.84–96.51, 20.31–181.53, and 1.22–3.04 mg/L, with averages of 30.62, 65.25, and 2.21 mg/L. Concentrations of HCO3- in groundwater, surface water, and rain water samples were in the ranges of, respectively, 13.00–152.00, 33.00–73.00, and 32.00–112.00 mg/L, with average values of 77.17, 54.50, and 78.13 mg/L.

Table 2.Descriptive statistics for chemical concentrations (mg/L) of groundwater samples.

Table 3.Descriptive statistics for chemical concentrations (mg/L) of rain water, surface water, and sea water samples.

Water compositions during April 2009 to February 2010 showed no variations that indicated the influence of the early excavation phase of the LILW facility. Relatively high concentrations of inorganic constituents in surface waters were similar to those in groundwater, which is thought to result from groundwater discharge from tunneling work related to National Road 31 during the sampling period. However, the similar concentrations of HCO3- in groundwater and rain water indicate that the impact of artificial contamination of these waters is less than for surface water.

 

Isotopic compositions of water types

The δ18O and δD values of rain water were evaluated 11 times between 29 May 2009 and 20 October 2009, yielding values of –10.47‰ to –4.04‰ (average, –9.01‰) and –77.9‰ to –21.6‰ (average, –65.2‰), respectively (Fig. 5); these values are generally comparable to those reported previously for northeastern Asia, (–15.0‰ to –7.04 ‰ and –113.9‰ to –43.3‰; Lee et al., 2001). The δ18O–δD isotope regression line for rain water (δD = 8.57 δ18O + 12) is slightly lower than the local meteoric water line (LMWL) (δD = 8.05 δ18O + 12.72) (Lee and Chung, 1997). In contrast, the δ18O–δD regression line for groundwater is close to the LMWL. Seasonally, isotopic compositions for rain water are lighter in the rainy season and heavier in the dry season. In addition, δ18O and δD values for groundwater in high-elevation, intermediate-elevation, and lowland areas are similar to one another.

Fig. 5.δD vs. δ18O values of rain water, surface water, soil water, and groundwater relative to the local meteoric water line (LMWL) in the Pohang area (Lee and Chung, 1997).

The δ18O and δD values of surface water were in the range of –9.35‰ to –7.69‰ (average, –8.38‰) and –66.7‰ to –59.1‰ (average, –61.7‰), respectively. The δ18O and δD values of soil water were in the range of –9.67‰ to –6.86‰ (average, –8.37‰) and –72.0‰ to –50.3‰ (average, –60.7‰), respectively, and the δ18O and δD values of groundwater were in the range of –8.05‰ to –7.00‰ (average, –7.47‰) and –56.6‰ to –44.6‰ (average, –50.6‰), respectively. Hence, average values for groundwater are heaviest, followed by average values for soil water, surface water, and rain water. These results indicate that water becomes heavier during transit from rain water to subsurface water.

Seasonal variations in the δ18O and δD values of groundwater show an increase in the lighter isotope portion during the rainy summer season, indicating recharge from rain water (Figs 6 and 7). The variable range of δ18O and δD values in monitoring wells may reflect variable recharge and discharge effects: large variations indicate high recharge rates while small variations suggest low recharge rates or high discharge rates.

Fig. 6.Temporal variations in δ18O values of groundwater.

Fig. 7.Temporal variations in δ18O values of surface water, soil water (RB-1S, DB1-6S), and rain water.

Radon-222 composition of water types

Concentrations of 222Rn in groundwater were in the range of 1.39–107.5 Bq/L (median, 18.79 Bq/L; arithmetic mean, 42.03 Bq/L). Samples KB-2 and DB1-2 showed exceptionally high 222Rn concentrations, with average values of 74.72 Bq/L and 100.24 Bq/L, respectively, as compared with mean groundwater concentrations of 1.00–53.00 Bq/L. The lowest concentration, of 3.07 Bq/L in waters from well DB1-6, may be representative of values in shallow groundwater and may be related to direct recharge of rain water.

As distinct from groundwater, concentrations of 222Rn in seawater near the study area were in range of 2.49–11.58 Bq/L (median, 3.93 Bq/L; arithmetic mean, 7.03 Bq/L), which is approximately twice as high as normal concentrations in seawater (0.03–3 Bq/L). As compared with results of this study, the median value of 222Rn concentrations in groundwater in the eastern coastal area of Busan City was 18.36 Bq/L, that of Ilkwang Stream water was 1.41 Bq/L, and that of seawater was 0.03 Bq/L (Ok et al., 2011).

CFC composition of water types

Chemical tracers such as CFCs (Cook et al., 2003) and 3H (Cook and Solomon, 1997) were used to estimate groundwater residence time (or groundwater age), which corresponds to the time required for groundwater to travel from the water table at a location of higher land elevation to a certain point down the gradient. As a consequence, these chemicals cases are important for identifying groundwater movement and mass transport within groundwater. In this study, CFC-12, CFC-11, and CFC-113 were used to determine the residence times of groundwater.

Groundwater ages determined from five samples and using CFC-12, (more reliable than CFC-11 and CFC-113 for determining groundwater age) were 26–34 years. Surface water ages were determined to be 23–32 years (slightly younger than those of groundwater). Moreover, groundwater age showed significant temporal variations. In the case of RB1 and IJ1, groundwater residence times were significantly reduced during late August–September, but increased in October (Fig. 8), indicating that high rainfall during July–August recharged the wells with a delay of ca. 1 month, thus reducing groundwater residence times; however, this effect was limited to only a few months. Among hydrochemical parameters, concentrations of dissolved SiO2 show a linear relationship with groundwater age, except for during the month of July for RB1 and to a much lesser extent for IJ1 (Fig. 9). This linear relationship can be attributed to dilution by rapid groundwater flow, showing that natural mineralization is also affected by recharge from summer rainfall.

Fig. 8.Temporal variations in groundwater age based on CFC-12 levels at wells RB1 and IJ1.

Fig. 9.SiO2 vs. CFC-12 for well RB-1 and for IJ wells.

Numerical modeling

Model domain and input parameters

A no-flow boundary of the numerical model was defined along the hydraulic divide, which approximately coincides with the topographic divide (Fig. 10). A Dirichlet boundary condition was assigned to the main stream and the East Sea, whereas a Cauchy boundary condition was specified for small streams. A digital elevation model (DEM) was used to specify the surface topography of higher-elevation land in the west and lowlands in the east (total relief, 0–220 m). The model domain was divided into fine grids (0.5–3 m resolution) in the more important zones (e.g., the radioactive waste disposal facility and the tunnel on National Road 31) and coarse grids (5–10 m resolution) for less important outer zones.

Fig. 10.Model domain for the numerical simulation.

The model section was composed of six vertical layers, with 472 rows × 460 columns (217,120 cells per layer), yielding a total of 1,302,720 cells (Fig. 10). The vertical layers were assigned on the basis of the geology of the study area, as well as on data from 16 drill holes and core samples in the study area. The HSD was assigned to layer 1, and consists of the soil layer and the upper part of the bedrock. The HRD was assigned to layers 2–6. The HCD was assigned to all six layers (layers 1–6). The numbers of active and non-active cells were 1,083,012 and 219,708, respectively.

Initial groundwater levels were input using data obtained for April 2008 (Fig. 3). Average recharge values of groundwater, calculated using the groundwater level fluctuation method, ranged from 194 mm/a at HSD to 237 mm/a at HCD.

The hydraulic conductivity values of the six vertical layers were estimated from water pressure testing and field hydraulic testing (Table 4). The hydraulic conductivity (K) estimates of layer 1 were in the range of 9.72 × 10–9 to 1.85 × 10–6 m/s; K values of layers 2 and 3 were distributed between 1.92 × 10–6 and 2.52 × 10–5 m/s; K values of layer 4 were in the range of 1.92 × 10–6 to 4.51 × 10–6 m/s; K values of layers 5 and 6 were in the range of 2.21 × 10–7 to 4.51 × 10–6 m/s and 2.21 × 10–7 to 1.92 × 10–6 m/s, respectively. The K values of layers 2 and 3 represent the values of the main aquifer. Highly variable K values are apparent in layer 1, because of the very wide range of particles in soil. The low K values of layers 5 and 6 represent the hydraulic properties of bedrock in the region.

Table 4.Hydraulic conductivity ranges in each layer in the model area.

Model calibration

Calibration of the model was performed by matching observed groundwater levels at monitoring wells (DB1-2, DB1-6, SS-5, RB-1, IJ, and KB-2) and simulated levels in the steady state condition. During model calibration, the boundary conditions inside the model area, hydraulic conductivity, and groundwater recharge were adjusted from their initial values so as to reach the error limits of actual values, using a trial-and-error method. The calibration between observed and calculated groundwater levels resulted in a simulation that matched observations at a 95% confidence level, represented by a correlation coefficient of 0.999, a root mean square of 4.084 m, and a residual mean of 3.625 m.

Groundwater travel times and residence times

Groundwater travel times using forward particle tracking were estimated from the half depths of the six layers at four imaginary wells along a ~20-m land surface interval, with an average lateral distance between imaginary wells of ~399 m (Fig. 11). Through the forward particle tracking, average groundwater travel times from the half-depth of each layer to layer 1 were estimated at 33, 66, 199, 266, 1278, and 8520 years for layers 1–6, respectively (Fig. 12; Table 5).

Fig. 11.Particle locations for the forward and backward particle tracking.

Fig. 12.Three-dimensional oblique cross-section showing the groundwater travel times of forward particles.

Table 5.Travel time of groundwater obtained by numerical modeling.

Groundwater travel times using backward particle tracking were calculated at eight monitoring wells (DB1-2, DB1-6, IJ, IJ-1, KB-2, KB-3, RB-1, and SS-5) (Fig. 11). The particles were placed into layer 1 at wells IJ and RB-1, and into layers 1 and 2 at the other six wells. Groundwater travel times in layer 1 were estimated to be ca. 16 years at well DB1-2 and ca. 33 years at the other wells (Fig. 13; Table 5). On the other hand, groundwater travel times from the water table to layer 2 were estimated to be ca. 70 years at wells KB-2 and KB-3, ca. 266 years at well DB1-2, ca. 280 years at wells SS-5 and IJ-1, and ca. 525 years at well DB1-6. A comparison of groundwater travel times estimated by particle tracking and CFC-12 age analysis shows that for layer 1, groundwater ages determined by CFC-12 (26-34 years) approximately correspond to the average groundwater travel time determined by particle tracking (33 years), indicating the groundwater in the wells of the study area could have been recharged at distances of 140-230 m up-gradient from the wells.

Fig. 13.Three-dimensional oblique cross-section showing groundwater travel times of backward particles.

Groundwater ages (ca. 26–34 years, based on CFC–12 analyses) may represent a mixture of groundwater sources originating at different depths and along different flow paths. In fact, the age predicted by asymptotic decay techniques is invariably younger than the actual age of the mixed sample, with an extreme bias towards younger ages when the water is a mixture of one young source and one very old water source (Bethke and Johnson, 2008). Figure 14 indicates that, irrespective of the mixture of different groundwater ages in well RB-1 (located in the recharge area) and well IJ (located in the discharge area), the oldest measured age of water was in well IJ.

Fig. 14.CFC-12 ages vs. δ18O values for water in wells RB-1and IJ.

Based on the CFC data, we consider that groundwater residence times were affected by summer rainfall, after a lag period, and were a better reflection of the pre-rainy-season state of groundwater flow. Seasonal variations in the δ18O and δD values of groundwater displayed an increasing tendency towards the lighter isotope portion during the rainy summer season, indicating recharge from rain water. A reasonably good linear relationship between 222Rn concentrations and δ18O values, and between 222Rn concentrations and δD values also indicates groundwater recharge in the rainy summer season and groundwater discharge in the dry winter season. In addition, a linear relationship between dissolved SiO2 concentrations and groundwater age provides evidence that natural mineralization is affected by recharge from summer rainfall and by dilution from rapid groundwater flow.

 

Conclusions

During the early stage of construction of the underground radioactive waste facility near the coastal city of Gyeongju, South Korea, groundwater flow paths and ages were determined from the hydrochemical and hydrogeological characteristics of groundwater, surface water, rain water, and seawater, as well as by numerical modeling. The backward particle tracking of the numerical modeling indicated that the average groundwater travel time was in the range of 34 years (layer 1) to 3599 years (layer 6). Based on both groundwater travel times determined by backward particle tracking and groundwater ages determined by CFC-12 analysis, we concluded that in layer 1, groundwater ages in wells approximately matched groundwater travel times, and that groundwater in the study area was recharged from an estimated distance of 140-230 m up-gradient from the wells, given that water depths in the wells are 100-232 m.

According to the CFC-12 analysis, groundwater ages were ca. 26–34 years and surface water ages were ca. 23–32 years. The variable groundwater ages determined by the CFC-12 analysis may represent the mixing of groundwater originating along different flow paths and at different depths, as well as of variable ages. The mixing of groundwater is supported by trends in chemical concentrations and EC values of groundwater, as distinct trends from higher to lower elevations are not observed.

It is thought that tunneling in the repository area during construction of National Road 31 disturbed the natural groundwater flow pattern. Lighter δ18O and δD values of groundwater in the rainy summer season suggest recharge from rain water. Recharge in the rainy summer season and discharge in the dry winter season were also indicated by a reasonably good linear relationship between oxygen and hydrogen isotope values, and between Rn-222 concentrations and isotopic compositions of groundwater. However, the CFC-12 analysis indicated a delay of ca. 1 month for groundwater recharge by rain water.

The present results will contribute to the safety of radioactive waste disposal by providing insights into groundwater flow systems and groundwater ages affecting waste repository sites.