Performance of different assessment methods to ...

0 downloads 0 Views 1MB Size Report
of Hurley and Roscoe (1983) was used to determine an MPN/100 mL and ...... Post V, Kooi H, Simmons C (2007) Using hydraulic head measurements in ...
Performance of different assessment methods to evaluate contaminant sources and fate in a coastal aquifer Sbarbati C.1, Colombani N.1,#, Mastrocicco M.2, Aravena R.3, Petitta M.1 1

Department of Earth Sciences, “Sapienza” University, P.le A. Moro 5, 00185 Rome, Italy

2

Department of Physics and Earth Sciences, University of Ferrara, Via Saragat 1, 44122 Ferrara, Italy

3

Department of Earth and Environmental Sciences, University of Waterloo, Waterloo, Canada

#

Corresponding author: [email protected]

Phone: +39 06 49914834 Fax: +39 06 4454729

ABSTRACT

The present study deals with the application of different monitoring techniques and numerical models to characterize coastal aquifers affected by multiple sources of contamination. Specifically, equivalent freshwater heads in 243 monitoring wells were used to reconstruct the piezometric map of the studied aquifer; flowmeter tests were carried out to infer vertical groundwater fluxes at selected wells; deuterium and oxygen isotopes were used to identify the groundwater origin and tritium was analysed to estimate the residence time; compound specific isotope analyses and microbial analyses were employed to track different sources of contamination and their degradation; numerical modelling was used to estimate and verify groundwater flow direction and magnitude throughout the aquifer. The comparison of the information level for each technique allowed determining which of the applied approaches showed the best results to locate the possible sources and better understanding of the fate of the contaminants. This study reports a detailed site characterization process and outcomes for a coastal industrial site, where a comprehensive conceptual model of pollution and seawater intrusion has been built using different assessments methods. Information and results from this study encourages combining different methods for the design and implementation of the monitoring activities in real-life coastal contaminated sites in order to develop an appropriate strategy for control and remediation of the contamination.

Keywords: Polluted site, coastal environment, inorganic and organic contaminants, characterization techniques, contaminants monitoring, conceptual model. 1

1. INTRODUCTION

Throughout the world, the most densely populated regions are located along coastlines because they normally provide the best conditions for economic development and quality of life (Dillon 2005). Many large industrial sites are located on coastal environments, since the easiness to be reached by cargo boats and a sustainable water supply for plant cooling purposes. However, the presence of contaminants in the subsurface environment, especially in groundwater, poses significant challenges to contaminant delineation and quantification at coastal industrial sites. The assessment of plume location and fate can be difficult and complex and there is a wide variety of characterization approaches that can be taken (Chadalavada et al. 2012; French et al. 2014). This includes conventional approaches such as hydrogeological and geochemical characterization, as well as other less common methods such as flowmeter testing and isotope analysis. Each of these methods provides hydrodynamic and hydrogeochemical information at different scales in time and space. One common approach for characterization of groundwater contamination is the use of mathematical models that simulate the groundwater flow and transport processes that can prescribe optimal strategies for managing contamination based on simulation of the physical process (Datta et al. 2009). According to current environmental legislation (for the European Community and USA) a series of “non – to low – invasive” methods has been proposed to reduce the variability connected with the traditional monitoring techniques (French et al. 2014). Although consulting companies use a staged approach for site characterization, and these characterizations are rarely published, only a limited number of published studies described detailed site characterization processes and outcomes. Examples are rare of coastal industrial sites with conceptual models of pollution and seawater intrusion status assessed with different monitoring techniques. Most studies describe industrial sites, located far from the sea, which are characterized by multiple sources of contamination (Hunkeler et al. 2004; Chapman et al. 2007; Yang & Lee 2012; Sidhu et al. 2013 and Wycisk et al. 2013) and investigated with different monitoring techniques. Sadiq and Alam (1997) studied, instead, a petrochemical complex that covers an area of about 170 km2 on the southern Saudi coast of the Arabian Gulf. The unconfined aquifer beneath this mega site was contaminated by metals and 102 wells were monitored using an integrated depth sampling (IDS) technique to track different pollution sources. Ratios between non-reactive species, such as Cl- and metals, were used to qualitatively distinguish between different pollution sources. Nobre and Nobre (2004) studied a Brazilian site located along the Ocean shore line, where a single event of 135 m3 of DNAPL were spilled from a 1,2-DCA production facility. In this study, most of 2

the efforts were spent to quantify the natural attenuation capacity of the aquifer with specific reference to 1,2-DCA and vinyl chloride (VC) using the IDS technique. In order to understand the distribution of dissolved contaminants, many authors (Reinhard et al. 1984; Einarson & Cherry 2002; Cherry et al. 2007; Colombani et al. 2009; Netzer et al. 2011) have highlighted the importance of a detailed vertical characterization of polluted aquifers by Multilevel Sampling (MLS). The MLS technique is highly recommended for analyzing groundwater pollution due to compounds with different density such as NAPLs and/or seawater intrusion (Prommer et al. 1999; Davis et al. 2009; Mastrocicco et al. 2011). At the Kwinana site in Western Australia (Johnston et al. 1998), a shallow unconfined aquifer was contaminated by BTEX from an oil spill. Although, this site was located 900 m from the coast line and groundwater was not affected by seawater intrusion at that location, the use of high resolution Multilevel Sampling (MLS) technique generated more quantitative results than IDS. In many other coastal sites, the importance of density - dependent flow and transport was highlighted in conjunction with plume leachate from contaminated sites or diffuse pollution from intensive agricultural practices (e.g., Ataie-Ashtiani et al. 2002; Dausman et al. 2010; Mao et al. 2006; Purnalna et al. 2003). Recently, two petrochemical plants located near the shore line in Italy were studied via numerical modelling: in the first case, tidal fluctuation was considered a heavy influence on a hypothetical contaminant (La Licata et al. 2011), while in the second case the tidal fluctuation was considered negligible (>0.1 m), and the model was calibrated versus MLS concentrations (Mastrocicco et al. 2012). Besides the contaminant modelling in coastal aquifers, the use of environmental isotopes has been recognized as a key technique to distinguish different sources of pollution in complex sites (Aelion et al. 2010). The isotope approach is also a useful tool to evaluate degradation processes affecting organic contaminants in groundwater (Hunkeler et al. 1999; Chu et al. 2004). Similarly, microbial analysis is also helpful in distinguishing between different contamination sources and attenuation processes (Seurinck et al. 2005; Vogel et al. 2007; Nnane et al. 2011). This current study included the use of existing field characterization techniques and data processing tools to resolve technical issues relevant to build up a robust conceptual model for a coastal industrial site affected by multiple sources of pollution and seawater intrusion. Specifically, field work was undertaken to evaluate the Level of Informativeness (LI) of different monitoring techniques in helping understand contaminant flow and transport mechanisms in the aquifer. Modelling work was undertaken to verify the level of agreement of the various techniques employed at the study site and also to provide information for better management of the remediation system. The effective LI and costs (C) are some of the issues that must be addressed in selecting efficient, accurate and widely applicable methodologies for the identification of sources 3

and contaminant fate in aquifers. Information and results from this study encourages combining different methods for the design and implementation of the monitoring activities in real-life coastal contaminated sites, in order to develop an appropriate strategy for management of coastal aquifers impacted by industrial wastes.

2. SITE DESCRIPTION

The study site is a petrochemical plant located close to the shore line where, from the early 1960s, the aquifer has been impacted by leakage of different type of pollutants (e.g., petroleum hydrocarbons, chlorinated solvents, agricultural fertilizer, sewage leakage and heavy metals). The reconstruction of the stratigraphic and hydro-stratigraphic sequence under the petrochemical plant is based on the interpolation of more than 870 core logs distributed across the site. The multilayer aquifer system is composed of four major units: an alluvial Quaternary sandy unconfined aquifer (the only one impacted by pollution), with a variable thickness from 10 to 30 m and an average hydraulic conductivity (k) of 2x10-4 m/s; a clayey silt aquitard having an average thickness of 25 m, a mean k of 1x10-7 m/s and locally overlying a confined aquifer with a mean k of 1x10-4 m/s; a gravel aquifer, 5 m thick, with a mean k of 4x10-4 m/s; and a Pliocene clayey aquiclude at the base (more than 100 m thick). Pumping tests showed that there is no connection between the unconfined and the confined aquifers (Mastrocicco et al. 2012). The unconfined aquifer is mainly supplied by the regional groundwater flow and local recharge. The piezometric surface is monthly monitored by a network of 246 monitoring wells, having an average depth of 25 m b.g.l. and the hydraulic gradient is about 0.6%, with a maximum of 1% close to the pumping wells (Fig. 1). The natural flow is perpendicular to the coastline and it is affected by the leakage of a Seawater Canal used for plant cooling purposes (Mastrocicco et al. 2012). The unconfined aquifer hosts a Horizontal Flow Barrier (HFB) consisting of a vertical bentonite wall with an average width of 0.8 m and a mean k of 1.2x10-10 m/s. In the eastern part, it penetrates the full thickness of the unconfined coastal aquifer and is keyed into the underlying aquitard, but in the central part, the HFB is not keyed into the aquitard (Mastrocicco et al. 2012). The HFB was originally constructed to contain the migration of LNAPL towards the sea and it is actually coupled with a Hydraulic Barrier (HB) in order to enhance the efficiency of the containment and remediation system. The HB is an alignment of 67 fully screened pumping wells with an average pumping rate of 60 l/s (Arlotti et al. 2012). Due to the effect of continuous pumping by the HB, there are evident disturbances of the flow lines, in direction, magnitude and flow velocity (Mastrocicco et al. 2013), which also causes local 4

seawater intrusion (Mastrocicco et al. 2013). The density contrast between fresh and seawater have created a “stagnant zone” of fresh polluted groundwater downgradient of the HFB, limiting the groundwater flow to the HB pumping wells, located upgradient (Fig.1).

Figure 1 – Maps and section of the study site. The top panel included: the contour map of equivalent freshwater heads (blue lines), the HFB (red lines), the HB (red triangles), the shoreline (brown line), the canals and river (light blue lines) and the IDS monitoring wells (black crosses). The middle panel included: the contour map of equivalent freshwater heads (blue lines), the HFB (red lines), the MLS (red circles), the shoreline (brown line), the canals and river (cyan lines), the flowmeter logs (black crosses), the monitoring wells where isotopic analyses (purple diamonds) and the microbial analyses (green triangles) were performed. In the bottom panel, the cross section A to A’ (see the middle panel for location) is shown and the main features of the unconfined aquifer are presented.

5

As previously described, different kinds of pollutants, including nitrogen compounds, chlorinated solvents and petroleum hydrocarbons, affect the aquifer. The site contains different kinds of potential sources of nitrate and ammonium, including a past production of agricultural fertilizer and two different treatment plants receiving, actually, urban sewage. Chlorinated solvents are primarily represented by 1,2-DCA and its degradation products. The DCA sources include a former 1,2-DCA production plant and a landfill area where 1,2-DCA was stored. Petroleum hydrocarbons are stored in large tanks, mainly located in the central portion of the site.

3. MATERIALS AND METHODS

The different assessment methods (AMs) used to develop a reliable and robust conceptual model of the study area, involved the use of different field, laboratory and data processing tools. These included:

3.1 Heads survey Groundwater levels were monthly monitored in a network of 246 monitoring wells since 2002. The measurements were converted into equivalent freshwater heads to take into account the variable density effects (Post et al. 2007).

3.2 Flowmeter tests 16 vertical flowmeter tests were conducted in 2010 to verify the direction and the magnitude of vertical fluxes (Paillet 1998). These measurements are particularly useful in areas characterized by complex groundwater flow, including coastal areas (Paulsen et al. 2001; Lamontagne et al. 2002) and/or those impacted by human activities (Greswell et al. 2009). In this case study, measurements were focused where the groundwater flow is supposed to be affected by the HFB and by the pumping wells (Fig. 1). The instrument used is a HFP-2293 Heat-Pulse flow-meter of Mount Sopris®, it emits a heat-wave recorded by the two thermistors positioned at the same distance from the emitter. To ensure high accuracy in the measurement, the repeated point by point method has been applied with a vertical resolution of 1 m (Sze´kely and Galsa 2006).

3.3 Integrated Depth Sampling (IDS) The IDS performed at the study site allowed a rough screening of the groundwater pollution. The samples collected via IDS are generally representative of a flow-weighted composite of the 6

screened interval and thus integrates small-scale vertical heterogeneities of groundwater chemistry (Yeskis and Zavala 2002). From 2010 to 2013, eight sampling events (twice a year) were conducted via the IDS technique on the groundwater monitoring network composed of 246 monitoring wells (Fig. 1.) Sampling was accomplished using a MP1 Grundfos® submersible pump which was placed in the middle of the screened interval and a minimum of four volumes were purged in each well. A flow-through cell was employed to measure the water quality parameters (pH, EC, Eh, DO). Samples were collected after the stabilization of the physico-chemical parameters.

3.4 Multi-Level Sampling (MLS) In total, 30 of 246 monitoring wells were selected to be sampled using the MLS technique. The MLS technique was performed five times over three years (Fig. 1): twice in 2010, once in 2011 and twice in 2012. Most of the selected wells are located close to the shoreline where the flow and contaminant transport are strongly influenced by the presence of the containment and remediation system. Three to five groundwater samples were collected at different depths depending on the screen length from selected monitoring wells, using a Straddle Packer system (Solinst). The packer assemblage consisted of a 0.9 m long sampling port; set in the middle of two inflatable packers. The selected wells were purged pumping three volumes (approximately 15 l) until pH, DO and Eh stabilized. Field parameters were determined towards the end of purging using a multiparameter probe WTW Multi 340i, including a SentiTix 41 pH combined electrode with a built-in temperature sensor for pH measurement, a CellOx 325 galvanic oxygen sensor for DO measurement, a combined AgCl-Pt electrode for Eh measurement and a Tetracon 325 4-electrode conductivity cell for EC measurements. The major cations and anions were determined by Ion Chromatography with ICS-2000 Dionex. Petroleum hydrocarbons and chlorinated hydrocarbons, as well as methane and carbon dioxide were analysed via GC-MS (Gas Chromatography-Mass Spectrometry) Thermo-Electron MAT95 XP.

3.5 Isotope Analysis During 2011 and 2012, three sample campaigns were performed to collect samples for isotopes analyses. Oxygen and hydrogen isotopes analyses in water were performed in the Isotope Laboratory of the University of Parma, Italy. The isotopic measurements were carried out using a Finnigan Delta S, Fisons Optima and Finnigan Delta Plus mass spectrometers. The isotopic results are reported in 7

terms of δ units (‰ deviation of the isotope ratio from the international standard VSMOW); the analytical error was ±0.1‰ and ±1‰ for δ 18O and δ2H, respectively (Longinelli & Selmo 2003). The tritium analysis was conducted at the Environmental Isotope Laboratory of the University of Waterloo, Ontario, Canada using electrolytic enrichment and a liquid scintillation counting (LSC) technique. The liquid scintillation counter is a LKB Wallac 1220 Quantulus. 3H results were reported in the tritium units (TU), with standard deviation (± 1σ) of ± 0.8 TU. Nitrogen isotopes were analysed in the Environmental Isotope Laboratory of the University of Waterloo (for low salinity samples) and in the Isotope Science Laboratory of the University of Calgary, Alberta, Canada (for high salinity samples). Samples with low salinity were analyzed using the silver nitrate (AgNO3) method (Silva et al. 2000) and they were analysed for 15N and 18O in NO3 - using a VG Micromass spectrometer and a VG Prism Micromass spectrometer, respectively. Samples having high salinity were processed accordingly to the “denitrifier method” (Rock and Ellert 2007) and analysed using a HP 6890 gas chromatogram PreCon® device interfaced to a Finnigan Mat Delta+XL mass spectrometer that enables the simultaneous determination of δ 15N and δ18O in NO3-. Accuracy and precision of δ15N and δ18O determinations based on a long term record of analyses of ISL in-house reference (ISLKNO3) is 0.3‰ and 0.7‰, respectively. δ15N in NH4+ was quantified via a modified version of the acidified disk diffusion method (Sørensen and Jensen 1991), using a Carlo Erba 1108 Elemental Analyser interfaced with a Thermo Instruments Delta plus – IRMS (Thermo Fisher Scientific, Milan Italy). NH4+-N isotope ratios are reported as δ relative to atmospheric N2. The uncertainty associated with the 15N-NH4+ analysis was generally better than ±0.3‰. Carbon and chlorine isotopes in 1,2-DCA, which is the most abundant chlorinated organic compound at the site, were analysed in the Environmental Isotope Laboratory of the University of Waterloo, using Compound Specific Isotope Analyses (CSIA). δ 13C analysis was carried out using a Thermo Trace gas chromatograph (GC) coupled to a Thermo Finnigian Delta plus XP isotope ratio mass spectrometer (IRMS) via a GC-CIII combustion interface set at 940 0C (Thermo Scientific, Waltham, MA, USA). δ 37Cl analysis was performed by direct injection of the GC separated 1,2-DCA into the ion source of a Micromass Isoprime IRMS (Isoprime, Cheadle Hume, UK) as described by Shouakar-Stash et al. (2006). All isotope ratios are reported relative to an international standard (VPDB for carbon, SMOC for chlorine) with an analytical error of ±0.5 for 13

C and ±0.2 for 37Cl.

3.6 Microbiological Analyses 8

To verify the occurrence of leakage from sewage pipeline, groundwater samples were collected in duplicate using sterile 100 mL bottles containing sodium thiosulfate and stored at 4°C in the dark until incubation for measuring total coliforms. Samples were analysed within eight hours using a most probably number (MPN)-based Colilert assay (IDEXX Laboratories, Inc), according to the manufacturer's instructions. The MPN solution of Hurley and Roscoe (1983) was used to determine an MPN/100 mL and respective 95% confidence intervals. The detection limit based on duplicate measurements was 0.5 MPN/100 mL.

3.7 Groundwater Flow and Transport models To verify the efficiency of the remediation system to contain the contaminant plumes within the site’s boundaries, a fully three-dimensional groundwater flow and contaminant transport model was developed for the whole site. Steady-state groundwater flow simulations were performed using MODFLOW-2005 (Harbaugh 2005), whereas density-dependent transport simulations were performed using SEAWAT 4.0 (Langevin et al. 2007), which is a coupled version of MODFLOW and MT3DMS that allows to simulate three-dimensional density-dependent groundwater flow and solute transport. Total dissolved solids (TDS) were used as the only parameter influencing the density-dependent groundwater flow and solute transport. Heat transport was not simulated and its influence on density-dependent flow and transport was neglected, because very little variations were detected in monitoring wells located in the unconfined aquifer (18-22°C). Density-dependent transport was previously demonstrated (Mastrocicco et al. 2012) to be fundamental in order to accurately assess seawater intrusion and salinity variations at the site. The site wide densitydependent flow and transport model was calibrated versus 246 observation wells for heads and 15 MLS for chloride (Cl-) concentrations (Mastrocicco et al. 2012). The tidal fluctuations showed minor relevance to this part of the Mediterranean Sea, in the order of 0.2 m (Sammari et al. 2006), so their effect on groundwater flow can be considered minimal. Consequently, tidal fluctuations were neglected in the numerical modelling process.

4. RESULTS

4.1 Heads Survey The analysis of the equivalent heads data showed the large capture zone of the HB is mainly located parallel to the coast line (Fig. 1). The general groundwater flow direction is from inland towards the sea, although several distortions are present due to heterogeneities in the hydraulic conductivity distribution and leakages from pipelines and a canal. Between the HB and the coastline, the 9

groundwater flow direction is reverted (from sea towards inland) and a saltwater wedge can develop where the HFB is not keyed into the aquiclude (Fig. 1 bottom panel).

4.2 Flowmeter Tests The 16 vertical profiles performed in proximity to the HFB indicated that the groundwater flow is converging towards the pumping wells of the HB preventing the spread of contaminants toward the sea (Mastrocicco et al. 2013). This result confirmed the

interpretation made based of the

equivalent freshwater heads distribution in the nearby monitoring wells.

4.3 IDS Characterization Figures 2 and 3 show the distribution of selected parameters measured with IDS technique. EC values (Fig. 2) generally range between 10000 and 15000 µS/cm (up to 30000 µS/cm in some monitoring wells), along the coast line. Within the site, EC values are heterogeneously distributed across fresh, brackish and saline water, as expected. Consequently, it is complicated to determine a background EC value. This is due to the coexistence of many factors governing EC distribution: local recharge, leakage from the Seawater Canal (used for cooling purposes) and from other pipelines (fire control, deionized water, sewage, etc.), and the presence of an old saline lagoon located outside the north-eastern corner of the Plant (Mastrocicco et al. 2012).

Figure 2 - On site electrical conductivity measured via IDS in µS/cm (March 2011).

Dissolved Total Petroleum Hydrocarbons (TPH) is mainly present in the central part of the Plant (Fig. 3), where a LNAPL-free phase has been detected and its concentrations are high, exceeding in some cases the theoretical aqueous concentration of 70 mg/L calculated with the Raoult law for multi-component solubility. The concentration distribution, obtained via IDS sampling, provided 10

preliminary information about the source location; however, it was not reliable to understand the distribution of the TPH contamination along the aquifer vertical profile.

Figure 3 - Total Petroleum Hydrocarbons distribution, measured via IDS in µg/L (March 2011).

4.4 MLS Characterization Among physico-chemical parameters measured in a monitoring well located in the coastal area (Fig. 4), the pH data shows circumneutral values along the whole profile because the carbonate-rich sandy sediments act as a buffer. The dissolved oxygen (DO) profile presents an abrupt decrease in concentration within the first 2 meters of the aquifer, switching from hypoxic to anoxic conditions; the dissolved oxygen is reaching values relatively close to 0 mg/L. The Eh is gradually more negative along the depth profile, reaching the values of -100 mV, at ~ 29 m confirming that the aquifer is in reducing condition as suggested by the DO profile.

Figure 4 – Dissolved oxygen (DO) - pH profiles (left panel) and electrical conductivity - redox potential profiles (right panel) of a representative monitoring well located near the shoreline.

11

The EC profile clearly represents the stratification of variable density waters with depth. At the top of the aquifer, a lens of recharged freshwater is present, underlined by brackish water (around 20.000-25.000 µS/cm) from 5 to 14 m a.s.l., followed by a sharp increase in the EC values (up to 47.300 µS/cm), between 23 and 29 m a.s.l., suggesting seawater intrusion. The trends of the physico-chemical parameters in this representative monitoring well highlighted the importance of a vertical characterization within the aquifer, especially in cases of seawater intrusion.

4.5 Isotopes Data Groundwater samples for

18

O and D analysis were collected using the IDS and MLS approaches.

The isotope data of the 32 samples collected from 26 monitoring wells via IDS (Fig. 5 and Table 1) showed a wide range between -7.1‰ and +1.6‰ for δ18O and between -45.4‰ and +5.1‰ for δD, suggesting a variable mixing of seawater and freshwater, which is clearly observed in the wells (MW11, MW12 and MW13) located near the coast. For some monitoring wells isotopic results are available for two monitoring surveys. Table 1 - δ18O and δD results for groundwater sampled via IDS in 26 monitoring wells.

δD Monitoring well (‰ V-SMOW) MW1 -26.1 MW2 -20.9 MW3 -45.4 MW4 -17.2 -28.4 MW5 -30.6 MW6 -23.3 MW7 -28.2 MW8 -18.7 -16.3 MW9 -16.2 MW10 -26.1 -6.0 MW11 -5.7 5.1 MW12 1.7

The most depleted

18

δ18O δD (‰ V-SMOW) Monitoring well (‰ V-SMOW) -4.5 MW13 -1.1 -3.2 MW14 -27.8 -7.1 MW15 -25.5 -2.8 MW16 -29.3 -4.6 -28.1 MW17 -4.9 -28.6 MW18 -3.7 -28.7 -5.0 MW19 -26.9 -3.3 MW20 -30.5 -2.6 MW21 -43.4 -3.0 MW22 -5.8 MW23 -4.5 -31.4 -2.6 MW24 -26.7 -1.1 -28.7 MW25 1.6 -33.3 0.0 MW26 -24.3

δ18O (‰ V-SMOW) -0.2 -4.9 -4.4 -4.7 -4.6 -4.8 -4.9 -4.4 -5.1 -6.6 -1.2 -5.3 -4.3 -4.8 -5.4 -4.0

O value of -7.1 ‰ is associated with the regional groundwater flow which is

fed by rainfall up-gradient to the Plant, and local recharge is characterized by a

18

O value of

approximately -5.0 ‰. The first isotopic screening of the groundwater in the main aquifer allowed a preliminary evaluation of the areas along the coastline where affected by seawater intrusion

12

(Schiavo et al. 2009). However, a more detailed investigation would be required to infer the magnitude and the exact location where the seawater enters the main aquifer.

Figure 5 – Map of the monitoring wells selected for groundwater isotopes analyses.

Vertical profiles of δD and δ18O for a representative monitoring well located down-gradient of the HFB (Fig. 6) showed that both δD and δ18O shift with depth, starting from depleted values in the shallower part of the aquifer associated with local recharge moving to more enriched values in the deeper part, associated with seawater intrusion at the bottom of the main aquifer. The data collected in the middle point start to show the impact of sea water at this depth. In case of isotope data generated by IDS, the values of -19 and -4‰ for δD and δ18O, respectively, is only showed the effect of a mixing of local recharge and sea water.

Figure 6. δD and δ18O depth profiles of a representative monitoring well located near the shoreline.

13

Tritium was also measured by MLS to evaluate the groundwater residence time within the aquifer. Most of the shallow groundwater samples (data not shown) are relatively young (2.5 to 3.3 TU), with a clear influence of meteoric recharge (about 3.1 TU), except for piezometers located upgradient of the HFB where the older groundwater with TU values