Assessment of seawater intrusion using multivariate statistical , hydrochemical and 1 geophysical techniques in coastal aquifer , Cha-am district , Thailand 2

Assessment of seawater intrusion using multivariate statistical, hydrochemical and 1 geophysical techniques in coastal aquifer, Cha-am district, Thailand 2 Jiraporn Sae-Ju1, Srilert Chotpantarat1, 2, 3, 4, and Thanop Thitimakorn1,3 3 1 Department of Geology, Faculty of Science, Chulalongkorn University, Bangkok 10330, Thailand 4 2 Center of Excellence on Hazardous Substance Management (HSM), Chulalongkorn University, Bangkok, Thailand. 3 Research Program on Controls of Hazardous Contaminants in Raw Water Resources for Water Scarcity Resilience, Center of Excellence on Hazardous Substance Management (HSM), Chulalongkorn University, Bangkok, Thailand. 4 Research Unit of Green Mining (GMM), Chulalongkorn University, Bangkok, Thailand. 5


INTRODUCTION
Many coastal areas in the world contain densely populations as it is an area that has food integrity and important economic activities such as urban development, trade, and touristic activities.These are factors that have attracted people to settle in these areas, as a result, the water demand for consumption, agriculture, and industry has increased.Groundwater resources are an alternative water source.Compared with surface water, the groundwater is of high quality, is barely affected by seasonal effects ( i.e., constant temperature), with large available quantities.For the reasons above-mentioned, groundwater has long been pumped with a large quantity of water.As a result, the common phenomenon, so-called seawater intrusion, has occurred in many coastal areas worldwide (Mas-Pla et al., 2014;Shi and Jiao 2014;Morgan and Werner, 2015).
In the coastal aquifer, seawater lies under fresh water since fresh water is less dense than seawater; consequently, the zone of contact between fresh water and seawater is brackish water (Bear, 1999).Fresh water is commonly over the top of the heavier seawater and serves to push the seawater interface seaward.
In contrast, when pumping fresh groundwater from coastal aquifers with a large quantity, the pressure of fresh water is reduced, which in turn causes the seawater to migrate further landward.The seawater intrusion problem is one of the most important environmental issues that negatively affects groundwater resources significantly since groundwater salinity can lead to a reduction in fresh water availability and the degradation of groundwater quality (Essink, 2001;Werner et al. 2013;Kang and Jackson 2016;Ros and Zuurbier 2017) .Therefore, the study of seawater intrusion into coastal aquifers is needed to identify the affected zones where it should be able to prevent problems or remediate such areas efficiently.
The coastal study area is located at Amphoe Cha-am, Changwat Phetchabur, and was selected as it is considered a densely populated area and is one of the most famous tourist areas in Thailand.Therefore, the groundwater resource may become a primary water resource in the near future and is consequently drawn out over the yield in the aquifer.Under this current situation, the natural equilibrium of the seawater interface is directly changed, and the sea water laterally moves landward.Geophysical and hydrochemical techniques have been integrated to investigate areas disturbed by seawater intrusion ( McInnis et al. , 2013; Hydrol.Earth Syst.Sci.Discuss., https://doi.org/10.5194/hess-2018-137Manuscript under review for journal Hydrol.Earth Syst.Sci. Discussion started: 3 May 2018 c Author(s) 2018.CC BY 4.0 License.Agoubi et al., 2013;Kouzana et al., 2010;Cimino et al., 2008;Song et al., 2006;Kazakis et al., 2016;Fadili et al., 2017;Najib et al., 2017) .Geophysical techniques such as the 1-D electrical resistivity survey or vertical electrical sounding ( VES) have been used since the electrical resistivity between fresh water and seawater saturated zones show large differences (Van Dam & Meulenkamp, 1967;Sabet, 1975), therefore, it is capable of identifying the contrast in terms of resistivity values between seawater and freshwater in coastal aquifers.In addition, the VES technique can be enabled in large areas as it is less time-consuming and has an economical cost when compared with drilling exploration methods.However, geophysical survey data are not capable of identifying the clearly lateral penetration of seawater in various lithologic units in a hydrogeological formation and groundwater facies or chemical constituents as the low resistivity depends on various factors such as the formation materials and groundwater chemistry (Zohdy et al., 1974).
As a result, to fulfill the limitations of geophysics for delineating seawater intrusion areas, the integration of a hydrogeological investigation (with the help of lithologic information from drill wells), chemical analysis of groundwater samples, and multivariate statistical analysis were carried out.Therefore, the objectives of this study were to integrate multiple techniques including multivariate statistical, hydrochemical, and geophysical approaches to delineate the impact of seawater intrusion in coastal aquifers in the study area and further explain the geo-hydrochemical process in the coastal aquifers.

STUDY AREA
The study area is located in Amphoe Cha-am, Changwat Phetchaburi, which is a part of the central part of Thailand.The area covers approximately 360 square kilometers and lies between the latitudes 12°37.6'-12°53.845'N and longitudes 99°50.827'-99°729'E (Figure 1).The area is bounded by the northern and western borders of Tha-Yang District, by the southern border of Prachuap-Khiri-Khan Province, and by the eastern border of the Gulf of Thailand (Figure 1a).The area can be classified by topography into two major landforms consisting of a plain interleaved mountain covers (~20% of the area) and low-plain or coastal plain covers (~80% of the area).The plain interleaved mountain landform is located to the west of the area, covered mainly by forest areas, while the eastern side is a low-plain or coastal plain Hydrol.Earth Syst.Sci.Discuss., https://doi.org/10.5194/hess-2018-137Manuscript under review for journal Hydrol.Earth Syst.Sci. Discussion started: 3 May 2018 c Author(s) 2018.CC BY 4.0 License.
to the Gulf of Thailand, and consists of various land use types including agriculture, facilities, forests, and water bodies, respectively (Land Developement Department, 2011) (Figure 1b).

Geological Setting
The study area is located on the Shan-Thai subcontinent consisting of Carboniferous rocks to Quaternary sediments as shown in Figure 2. The Permo-Carboniferous sedimentary rocks and Permian limestone are the basement rocks in the area, and are found distributed to the east of the study area.In addition, basement rocks intruded by Cretaceous granite that appear as a mountain range and isolate hills are found in the west of the study area.The basement rocks are filled with Quaternary sediments.
Quaternary sediments consist of marine sediments that are coastal tide-dominate deposits, colluvial sediments that accumulate around the hill foot (Department of Mineral Resources, 2014).

Hydrogeology
The study area is underlain with unconsolidated and consolidated aquifers (see Figure 3).More than 60% are underlain with the Quaternary unconsolidated aquifer that consists of a beach sand aquifer (Qbs).These are common sediments in the east coastal plain and are distributed from north to south.
Therefore, the groundwater has accumulated in the pore space of the sand deposited in the old ridge.The average depth of the aquifer is 5-8 meters with groundwater levels of approximately 1-2 meters deep.Well yield is less than 2 m 3 /hour.The total dissolved solid content (TDS) ranges from 500-1000 mg/L.The flood plain aquifer (Qfd) is dispersed over in the upper part and is lined between Qbs and granitic aquifers (Gr).
The groundwater has accumulated in the pore space between the gravel and sand grains.The average depth of the aquifer is 25-45 meters with groundwater levels of approximately 3-8 meters deep.Well yield is between 2-10 m 3 /hour, but in some areas, its range from 10-20 m 3 /hour.The TDS ranged from 600-2000 mg/L.The colluvial sediments aquifer (Qcl) deposits near the foothill of the granite mountain are in the north and south of the area.Thus, the sediments are poorly sorted, are angular to sub-angular, and the groundwater has been stored in the pore spaces of the gravel, sand, quartzite fragments, granite fragments, and clay deposits.The average depth of the aquifer is 30-50 meters.Well yield is relatively high at approximately >50 m 3 /hour.The TDS ranged widely from 300 to >1200 mg/L.The consolidated aquifers are composed of 40% of the study area as described follows.The Silurian-Devonian metamorphic aquifer (SDmm) was found in a small area in the lower west.The average depth of the SDmm aquifer is 30-35 meters with a groundwater level of approximately 3-5 meters deep.Well yield ranged from <2 m 3 /hour.
The quality of the groundwater is generally good with a TDS <800 mg/L.The Permian-Carboniferous metasedimentary aquifer (PCms) can be classified in two groups.First, in the northern part, groundwater has accumulated in the fracture of mudstone and shale inter bedded with sandstone.The average depth of the aquifer ranged from 50-200 meters.Well yield ranged from <2 m 3 /h, but some areas can develop groundwater of 5-10 m 3 /h in the fracture zone.Second, in the upper eastern part of the study area, groundwater has accumulated in the fracture of quartz rich sandstone interbedded with thin shale.The average depth of the aquifer is 50-100 meters.Well yield ranged from 2-10 m 3 /h, but some areas develop groundwater of >20 m 3 /h.The Permian limestone aquifer (Pc) has been found dispersing a small monadnock in the north of the area.The average depth of the aquifer is 10-50 meters with a groundwater level of approximately 3-5 meters.Well yield ranged between 10-20 m 3 /h.The quality of groundwater is good with a TDS of <500 mg/L.Finally, the Cretaceous granite aquifer (Gr) is dispersed along the large mountain ranges in a north-south direction to the west of the area.The average depth of the aquifer ranged from 50-150 meters with groundwater levels of 6-9 meters deep.Well yield is <2 m 3 /h, but some areas can develop groundwater of more than 20 m 3 /h (in fracture and/or fault zones).The TDS ranged from 200-1000 mg/L (Department of Groundwater Resources, 2015).

METHODOLOGY
Seawater intrusion into coastal aquifers can be detected through several methods.In most of the case studies, the electrical method has been universally recognized (George et al., 2015;Cimino et al., 2008;Samouëlian et al., 2005) as the electrical conductivity (EC) of a saturated subsurface depends primarily on three major factors: porosity, the connectivity of pores, and the specific conductivity of water in the pore (Telford et al., 1990).The difference between the EC of the seawater saturated subsurface and freshwater saturated subsurface is significant; thus, the electrical resistivity survey was well suited for evaluating the relationship between freshwater and seawater in coastal areas (Sherif et al., 2006).The electrical method analyzed with hydrogeological measurements can map interfaces of fresh water and seawater in coastal aquifers more precisely ( Zarroca et al., 2 0 1 1 ) .In this study, we used both the direct method by sampling groundwater samples for hydrochemical analysis, and the indirect method by Vertical Electrical Sounding ( VES) .Both the sampling of groundwater samples or VES investigation collected data in the dry season (April to May) since seawater intrusion may occur evidently.

Vertical Electrical Sounding Method
One-dimensional resistivity survey, known as VES, was applied in this research.The principle of VES is currently released into the ground through current electrodes A and B, then the voltage is measured by potential electrodes M and N. The resistivity meter converts the voltage to the resistance value, then is plotted on log-log paper in the field to check for anomalies.Eighty VES points in this research used the resistivity meter (Syscal R1 Plus model, Iris) by the Schlumberger configuration (Figure 4).
The results of the VES survey method were the resistivity of the soil and rock layers vertically.The shape and slope of the VES data on the graph represents changes of the layer that had different resistivity (Telford et al., 1990).Therefore, the correct interpretation of VES required the use of information regarding the local geology and drilling information of the groundwater wells.

Chemical Analysis of Groundwater
A total of 57 groundwater samples were collected in the study area.Prior to sampling, the groundwater must be pumped out for 5-10 minutes to represent the groundwater in the aquifer.Samples were necessary to measure physical and chemical parameters such as pH, electrical conductivity (EC), groundwater, and total dissolved solids (TDS) in the field by using portable meters.Samples for the analysis of cations and anions were collected in two bottles of 500 mL Poly-Ethylene (1 bottle with nitric acid (HNO3) added to maintain the acidity of the water and the other with no nitric acid to be used for anion analysis.Samples were kept at 4 °C to reduce the process of microorganisms and reduce the speed of physical and chemical processes.The groundwater analysis was divided into two parts: anion analysis and cation analysis.Chemical analysis was carried out for a group of cations (Ca 2+ , Mg 2+ , Na + , K + , and Fe 2+ ) analyzed by the Absorption Spectrometry Method (AAS), whereas the group of anions (Cl -, Br -, NO3 2-, and SO4 2 ) were analyzed using the Ion Chromatographic Method (IC) and CO3 2-, HCO3 -were analyzed by the volumetric titration method.The results of the chemical analysis can be analyzed by the piper diagram and hydrochemical facies evolution (HFE) diagrams to classify the types of water that seawater intrusion can be indicated by an increase in TDS and the major ions of seawater.Furthermore, correlation analysis and principal component analysis (PCA) were used to clearly acquire the relationship among the hydrogeochemical parameters measured in the groundwater samples, which enabled the identification of the hydro-geochemical processes occurring in the groundwater system.
In this study, the multivariate statistical technique were carried out using the SPSS software, version 22.
Ten hydrochemical parameters (K, Na, Fe, Ca, Mg, F, Cl, Br, SO4, and HCO3) as well as pH, and electrical conductivity (EC) in the groundwater samples were carried out using correlation analysis and PCA.These two techniques are a statistical technique to group and establish the relationship between a group of groundwater samples based on hydrochemical characteristics (Abderamane et al., 2012).

Vertical Electrical Sounding
Figure 1a shows 80 VES points.All 80 of the VES data showed the H type curve (Telford et al., 1990) (Figure 5a).This implied that there were three layers in the subsurface that consisted of resistivity with ρ1 > ρ2 < ρ3 (ρ1 = resistivity of upper layer, ρ2 = resistivity intermediate layer ,and ρ3 = resistivity bottom layer) and were consistent with the characteristics of geology and hydrogeology of study area.The top layers represented the unsaturated sediments, the middle layer was the saturated sediments, and the bottom layer was interpreted as bed rock (Figure 5b).

[Insert Figure 5]
Figure 5b shows the results of the interpretation of VES St25 where apparent resistivity is the black line and the master curve is the red line.It was found that at the St 25 location, it could be divided into four layers (blue line).The top layer had a resistivity of 9.38 Ωm with a thickness of 3.11 m and a depth of 3.11 m.The second layer had a resistivity of 2.02 Ωm, thickness of 11.3 m, and depth of 14.4 m.The third layer had a resistivity of 1.49 Ωm, thickness of 60.1 m, and a 74.5 m depth.The bottom layer was the bedrock layer with a resistivity of 757 Ωm.The RMS error was 2.46%.After the layers were interpreted for all points, then pseudo cross-sections for the four lines were generated.The pseudo cross-sections ( A-A', B-B', C-C', and D-D') were generated from 1D VES data that selected 47 VES points from this study and 9 VES points from the Department of Groundwater Resources (DGR) database.All cross-section lines were created from west to east that was perpendicular to the coast line (Figure 1a).
The pseudo cross-section A-A' (Figure 6a) had 14 VES that consisted of St1,St2,St3,St4,F5,St5,St6,F6,St7,St8,St9,and St10.This line was located at the lower part of the study area and the total length was approximately 6750 m.It was found that the surface layer (0-10 m) had resistivity ranging from 10-1000 Ωm.It was interpreted as an unsaturated zone of the Qfd aquifer.At the western side of the profile, very high resistivity (400-1000 Ωm) was observed.This is due to the granite batholith that intrudes in the base rock, as a granite outcrop was found near the St8.In the eastern side, at a depth of 130-200 m, the layer that had a resistivity of 50-100 Ωm was the bedrock layer.The top of this side showed resistivity values between 3-40 Ωm, and represents sediment saturated with water, especially near the coastal line where there was a very low resistivity zone (<5 Ωm).This may represent the influence of seawater intrusion that intrudes inland about one kilometer from the coast line.
The pseudo cross-section B-B' (Figure 6b) consisted of 16 VES points (St11,St12,St13,St14,St15,St16,St17,St18,St19,St71,St20,St21,St22,St23, and started from east to west.This line was close to the A-A' profile and total distance was approximately 8450 m.It was found that the surface layer (0-10 m) had resistivity anomalies between 50-400 Ωm.This represents the unsaturated zone of the Qfd aquifer.The next layer showed resistivity values ranging from 2-20 Ωm.This was interpreted as a sediment layer saturated with water.Near the coast line, there was a low resistivity zone (<5 Ωm) that extended approximately two kilometers inland, which may represent the influence of seawater intrusion.At the west side of the profile at a depth 20-200 m, there was a zone of resistivity ranging from 60-250 Ωm that represented the bedrock layer.
The pseudo cross-section C-C' (Figure 6c) consisted of 12 VES as follows: St24,St25,St26,St27,St28,St29,St30,St31,St32,and  The pseudo cross-section D-D' (Figure 6d) consisted of 14 VES as follows: St34,St35,St36,St37,St39,St41,St42,St43,St44,St45,St46,and St47 from an east to west direction.This profile was on the northern part of the study area and the total length was approximately 14,300 m.It was found that the surface (0-10 m) layer had resistivity in a range from 1.5-400 Ωm, representing the unsaturated zone of the Qfd aquifer.In this profile, there were low resistivity anomalies (<5 Ωm) near the surface at St7, possibly caused by the waste water from community areas.The next layer showed resistivity values in a range from 1.5-20 Ωm, representing the sediment layer saturated with water.Near the coast line, the low resistivity zone (<5 Ωm) extended far from the coast inland approximately four kilometers, representing the influence of seawater intrusion.In the west side of the profile, the bedrock was found with resistivity values of 40-100 Ωm at a depth between 90-200 meters.
Since soil or rock layers with low resistivity may be caused by many reasons including the clay content or salt water in hydrogeologic formations, this research compared the lithologic data with the VES data survey and found that the low resistivity at the area near the coast line may have been caused by the seawater.Figure 7 shows the resistivity values compared with the litho-log of lines D-D'.The left side shows the lithological data of well no.Q168 versus the VES of St47 that represents a position located far from the coast, and the right side shows the lithological data of well no.PW7962 versus the VES of St34 that represents a position located near the coast.It was found that the low resistivity (<5 Ωm) at VES St34 corresponded to a gravelly clay layer, but at VES St47, the layer with components of clay or clayey gravel showed a resistivity of only 10-20 Ωm.From the comparison, it can be concluded that although the layer had a clay component, the resistivity value was not very low (<5 Ωm).As a result, the area that showed very low resistivity may have been influenced by seawater intrusion.This is consistent with previous studies (Ravindran, 2013;Kaya et al., 2015;Gopinath and Srinivasamoorthy, 2015) who concluded that the areas with low resistivity (<5 Ωm.) were influenced by seawater intrusion.

Geoelectrical Section
When the pseudo cross-sections of the four profiles were analyzed with borehole log and electric log data, the comparison of these two data was established as shown in Figure 8.The four geo-electrical cross-sections clearly showed the boundaries of each aquifer.The top layer was the Qfd aquifer consisting of well sorted sand with high sphericity of approximately 0-20 m thick.The next layer was the Qcl aquifer, which consisted of clayey gravel (poorly sorted and angular to sub-angular) interbedded with sand in some areas.The average thickness was approximately 50-60 m, but it gradually increased by 100 meters in the area near the coast line.The next layer underneath the Qcl aquifer was the PCms aquifer, consisting of greenish gray sandstone and shale with an average depth between 50-200 meters.In the A-A' and B-B' profiles, which were located in the lower part of the study area, the PCms aquifer was not found, whereas granite (Gr) aquifers could be found in all profiles.In the D-D' profile, the Pr aquifer was located in the central part of profile.It lay on top of the PCms aquifer at an approximate depth of 50-60 m.Since this area had a limited number of borehole log data, the interpretation to create a cross-section in some areas was required as shown in the dashed line in the cross-section.Figure 9 shows the apparent resistivity map established from the apparent resistivity of VES for comparison of the resistivity of various depths.In this study, the apparent resistivity values for AB/2 = 5, 10, 30, 50, 70, 100, 150, and 200 meters were selected to create an apparent resistivity map by using the ArcGIS9.3 program.The dark blue color on the map represents low resistivity and the red color represents high resistivity, respectively.Figure 9 shows that all depths showed the location of low apparent resistivity (<5 Ωm) in the same place.This was located on the east side of the study area which covered Tumbon Tha-Yang, Nong-Sa-La, Bang-Kao, and some parts of Tambon Cha-Am.The maps of AB/2 = 5 and 10 meters (Figures 9a and 9b) showed low apparent resistivity distributed in a sporadic pattern.This was found in a wide area in the maps of AB/2 = 30, 50, and 70 m and was decreased in maps of AB/2 = 100, 150, and 200 m (Figure 9).The area with the highest apparent resistivity (>1000 Ωm) was located on the western side of the study area covering most areas of Tumbon Khao Yai and Sam Pra Ya, corresponding to the bedrock in the area.The map of AB/2 = 70 meters (Figure 9c) had the widest dark blue (<5 Ωm) area, which corresponded to the depth of 50 m from the ground surface when compared to the cross-section; consequently, it was found that this depth was located in the Qcl aquifer.As a result, it can be concluded that the Qcl aquifer has been highly influenced by seawater intrusion.

Groundwater Chemistry
The hydrochemical analysis of 58 groundwater samples is shown in Table 1.The total dissolved solids (TDS) ranged from 195-3580 mg/L.The electrical conductivity (EC) of the groundwater samples ranged from 292-5360 mS/cm.Figure 8 shows the hydrochemical facies classification of the groundwater samples with the chemical result plotted on a piper diagram (Galloway & Kaiser, 1980)   ( % Δ <10) and could be further used for discussion and interpretation while the unreliable results of 16 samples may have resulted from the process of groundwater collection and preservation, as well as the dilution of the concentration of samples.
Figure 8 presents the piper diagram that shows the hydrochemical facies classification of 42 samples.The piper diagram can be divided into five facies as follows: the Ca-Na-HCO3, Ca-HCO3-Cl, Ca-Na-HCO3-Cl, Ca-Na-Cl, and Na-Cl facies represent fresh water and were found in W5, which was opened as a well screen to the Gr aquifer.This groundwater facie is younger and is weakly acidic, which is generally found in high terrain or recharge areas (Appelo and Postma, 2005).The Ca-HCO3-Cl facie was found in W19 and W20 that were opened as a well screen to the Gr aquifer.This facie was in water-bearing permeable rocks given that when groundwater moves through the rock formation, the ion exchange process takes place (Appelo & Postma, 2005).The quality of the water in this facie is fresh water.The Ca-Na-HCO3-Cl facie was found in 21 wells, which opened the well screen in four aquifers as follows: Gr aquifer (eight wells), PCms aquifer (nine wells), Pr aquifer (one well, namely, W52), and Qcl (three wells).This facie presents a complex chemical pattern since the water was influenced by many factors.This may have resulted from the mixing of fresh water with seawater (Appelo and Postma, 2005).The Ca-Na-Cl facie was found in W53 that was opened as a well screen to the Pr aquifer, while three wells (W24, W25, and W34) were opened as a well screen to the Qcl aquifer, and two wells (W22 and W23) were opened as a well screen to both the Qcl and PCms aquifers.The groundwater facies may have been changed due to the influence of seawater intrusion by the ion exchange process (Appelo and Postma, 2005).The Na-Cl facie was found in 10 wells that were opened as a well screen to the Qcl aquifer.The groundwater facie indicated that these wells were influenced by seawater intrusion, and the quality of the water was saline water.

[Insert Table 1]
According to the piper diagram by Galloway and Kaiser (1980), the groundwater facie can be classified into five facies, which correspond to the influence of seawater intrusion, and not only depend on the distance from the coast, but also on the depth of the well or the screen level of the well.Most of the groundwater samples in the Qcl aquifer were relatively influenced by seawater intrusion.In general, an average depth of the Qcl aquifer ranged from 20-50 m and the thickness was typically higher than 100 meters approaching the coast line.The Na-Cl facies (approximately ~48%) and the Ca-Na-Cl facies (approximately ~19%) were mainly found in the Qcl aquifer.The remaining approximate 20% of the groundwater wells represented Ca-Na-HCO3-Cl.According to the geochemical facies, the average composition of cations (in meq/L) in the Qcl aquifer was in the following order: Na >> Ca > Mg~K.
Furthermore, the composition of anions (in meq/L) in the aquifer was in the descending order: Cl >> HCO3 > CO3 > SO4.Therefore, their hydrochemical composition could be addressed as seawater intrusion in this aquifer (Ahmed et al., 2017).The groundwater facies in the PCms and Gr aquifers, which opened a well screen in the high weathering zone (underneath Qcl aquifer) at a depth of 40-60 meters was composed of the Ca-Na-HCO3-Cl facies, suggesting that the groundwater was moderately affected by seawater intrusion.
Moreover, the groundwater facies in the PCms and Gr aquifers (where a well screen opened at a depth of 60-80 meters) were the Ca-HCO3-Cl and Ca-Na-HCO3 facies, suggesting that the groundwater was slightly affected by seawater intrusion due to the mixing of recharge rainwater.
The relationship between the depth variation with the Cl and Na concentrations in each aquifer is shown in the Supplementary Information (Figure SI.1).The chloride concentration was mostly found in relatively high concentrations in the Qcl aquifer, especially in shallow to moderate depths (approx.15 to 70 m depth).However, this relationship of salinity with depth was not well correlated in the Pcms and Gr aquifers, implying that the groundwater in these aquifers were not highly impacted by seawater intrusion.
The red line on the piper diagram (Figure 10) is the line that shows the hydrochemical evolution of the groundwater facies by the cation exchange reaction (Appelo & Postma, 2005).This line resulted from two points plotted between the composition of seawater (blue point) and fresh water (red point) on the piper diagram.By using this line, the hydrochemical results of the groundwater samples fall in the zone between the blue and red points, representing mixed water occurring between the fresh water and seawater.The groundwater facies close to the blue point were Na-Cl facies and Ca-Na-Cl facies, while the groundwater facies close to the red point were Ca-HCO3-Cl and Ca-Na-HCO3.The groundwater facies found between the blue and red points were the Ca-Na-HCO3-Cl facies.As mentioned earlier, these were similar to the study of Zghibi et al. (2014), who studied contamination in the Korba unconfined aquifer, which was influenced by seawater intrusion.They showed a chemical analysis of water on the piper diagram and then interpreted the results by creating the Theoretical Mixing Line (TML) of seawater and fresh water.They found that groundwater showed paths of hydrochemical evolution along the TML line.The groundwater facies can be changed from a Ca-SO4 type to a Ca-Cl type to an Na-Cl type, and vice versa, from a Ca-SO4 type directly to an Na-Cl type, indicating that the chemical composition of groundwater is changed by a cation exchange reaction.

[Insert Figure 10]
Figure 11 shows the relationship between the Na and Cl ions.The dominant ions in seawater are Na and Cl, while the dominant ions in fresh water are Ca and HCO3 (Appelo & Postma, 2005).Therefore, the study of seawater intrusion has to focus on the dominant ions of seawater, which are Na and Cl ions.
The plotted graph comparing the Na and Cl ions found that it exhibited a strong correlation where the groundwater samples fell on a 1:1 line with an R 2 of 0.941.This relationship suggests that both ions have the same origin from seawater.The concentrations of Na and Cl depend on the degree of seawater intrusion into the aquifer.From Figure 11, the groundwater can be divided into three groups.The first group, located in the red circle, mainly consisted of samples in the Qcl aquifer that are severely influenced by seawater intrusion.The groundwater types in this group showed Na-Cl and Ca-Na-Cl.The second group appeared in the green circle, and consisted of groundwater samples from the PCms and Gr aquifers, which are moderately influenced by seawater intrusion.The groundwater types in the second group showed Ca-Na-HCO3-Cl facies.The last group appeared in the blue circle and consisted of groundwater samples from the deeper zone (when compared to the second group) of the PCms and Gr aquifers, which are slightly influenced by seawater intrusion.The groundwater types in the last group showed Ca-HCO3-Cl facies and  2014) that revealed the characteristics of groundwater chemistry influenced by seawater intrusion.They found that groundwater influenced by seawater intrusion had dominant ions of Na and Cl.Therefore, when plotting Na and Cl, both ions are well defined in the correlation where the groundwater sample falls on the 1:1 line, indicating that both ions come from the same source (seawater).Similarly, according to the ionic ratio estimated from Cl/(HCO3 + CO3) and recommendation of Raghunath (1990), who suggested the ratio of 2.8 as the threshold for indicating the saltwater intrusion, we found that more than 85% of groundwater samples in the Qcl aquifer were moderately to highly contaminated due to seawater intrusion (Ebrahimi et al., 2016).

[Insert Figure 11]
Furthermore, the study of Al-Agha and El-Nakhal (2004) found that by plotting the results of groundwater samples on the piper diagram, phases of freshening and intrusion could be interpreted, but it is difficult to recognize the sequence of facies in detail.Therefore, this research used the hydrochemical facies evolution diagram (HFED) developed by Giménez-Forcada (2010) to describe the dynamic of seawater intrusion.The percentage of major ions in the hydrochemical process associated with the dynamic of seawater intrusion interface, consisting of Ca 2+ , Na + , HCO3 -, SO4 2-, Cl -, was considered in the HFED.
Figure 12 shows that the red block (no.4) is the Na-Cl facies (seawater), and the blue block (no.13) is the Ca-HCO3 facies (fresh water).After plotting the percentage of ions on the diagram, it will generate a mixing line between the fresh water and seawater to divide the phases of seawater intrusion.When the groundwater samples appear above the mixing line, it represents the freshening phase, and if it falls below the mixing line, this implies that it is during the intrusion phase.The results found that most of the samples fell close to the mixing line (facies path 4-7-10-13), demonstrating mixing between seawater and fresh water.Some samples fell in the intrusion phase (below the mixing line).In this initial phase, water gradually increases the salinity along the facies path 13-14-15-16, which causes a reverse exchange, showing a Ca-Cl facies (no.16), which was not found in a groundwater sample in this facies.Finally, in this phase, water evolves toward facies that are closer to seawater (Na-Cl facies) along facies path 16-12-8-4, and most samples in this study were found in these groundwater facies.However, in this research, the groundwater sample fell in Na-mixCl (no.3), Na-mixHCO3 (no.2), Na-HCO3 (no.1), and MixNa-HCO3 (no.5), which are characteristic of the freshening process, and was not found in all groundwater samples.
Most samples that fell in the Na-Cl (no.4) facies were groundwater samples collected from the Qcl aquifer, and corresponded to the Na-Cl facies, which appeared in the piper diagram.This indicated that the groundwater samples in the Qcl were severely influenced by seawater intrusion.Moreover, the groundwater samples from the weathering PCms and Gr aquifers fell in the MixNa-Cl facie (no.8) and MixCa-Cl (no. 12), which corresponded to the Ca-Na-HCO3-Cl facies in the piper diagram, indicated a moderate influence of seawater intrusion.In addition, the groundwater samples from the PCms and Gr aquifers at deeper levels (depth of 60-80 meters) showed the Ca-HCO3-Cl facies in the piper diagram and HFED fell in the Mix Na-Cl facie (no.8) and MixCa-Cl (no.12).Even though it fell in the facies of fresh water in the HFED, it was close to the mixing line; therefore, it was slightly influenced by seawater intrusion.These were consistent with the study of Ghiglieri et al. (2012), who used HFED for depicting the salinization processes in the coastal aquifers in Italy.They found that the sample plotted on HFED followed the succession of facies along the mixing line, indicating that seawater and fresh water were slightly mixed or that the ionic exchange process occurred.Moreover, they used the HFED results in comparison with the EC contour lines, showing that seawater had intruded quite far inland.Najib et al. (2017) highlighted the succession of different water facies developed between the intrusion and freshening phases by analysis on the HFED.The formation of Na-HCO3 facies, which characterizes the last facies of the freshening phase, followed the succession of Na-Cl, MixNa-MixCl, MixCa-MixCl, MixCa-MixHCO3, and Na-HCO3.Moreover, the obtained HFED results allowed us to extend the intrusion process in the Holocene groundwater and accept the fresh water recharge such as meteoric water and lateral recharge from rivers (Liu, 2017).

Principal components analysis (PCA)
The varimax method used to rotate the parameters in the principal component analysis uses the Kaiser criterion rotating with the varimax method.Therefore, all of parameters were classified into four components (see Table 3).Each component included the values (bold) that represent a good correlation.
Table 4 shows the Eigen value of all parameters, and the four groups had Eigen values more than 1 and cumulative variance was more than 80.35%.A total dissolved solid (TDS), EC, Mg 2+ , Na + , Cl -, Br -, SO4 2- were included in the first component that had the highest factor loading (6.13) and accounted for 47. 17% of the total variance.This factor, with a high positive loading, ranged from 0.757 to 0.965, probably indicating the consequence of seawater intrusion in the study area.Therefore, factor 1 can be defined as the "seawater intrusion factor".Similarly, the study of Ahmed et al. (2017) found that factor 1, which was defined as the seawater intrusion impact, accounted for 66% of the total variance and consisted of these following elements: EC, Cl, Na, SO4, K, Mg, Br, Sr, B, Cr, Co, Arsenic, and Selenium.The second component had a factor loading of 1. 60, accounting for 12. 28% of the total variance.The component showed the positive relationship between Ca 2+ (0.62), F -(0.63), and HCO3 (0.77).This factor can be expressed as the natural process when recharge water infiltrates into the groundwater system and waterrock interaction occurs, which eventually releases Ca 2+ and HCO3 -in groundwater (Jiang et al., 2009).The values of HCO3 widely ranged from 19.52 mg/L to 189.5 mg/L, depending upon various geological formations in this area.Fluoride is naturally released from the dissolution of fluorapatite and flurite, which occurs in sedimentary and igneous rocks.According to Rama Rao (1982) and Heinrich (1948), they revealed that fluorite was detected in granite, gneiss, and pegmatites.Moreover, due to the similarity charge and radius, fluoride can substitute the hydroxide ions via water-rock interaction.Similar to Ca and HCO3, fluoride (F) in groundwater can be inferred from the weathering of rocks in this area.The third component consisting of K + and Fe showed positive loadings in the range of 0. 7--0.81 and the factor loading was 1.448 or 11.14% of the total variance.Through the weathering of igneous rocks, K can be mainly released from potassium feldspar into the groundwater (Kim et al., 2004) and Fe represents the natural dissolution of rocks and minerals via water-rock interaction.The last component consists of only pH showing the low factor loading (1.27) or 9.78% of the total variance.These parameters might be the result of various dynamic hydro-geochemical processes in the area such as seawater intrusion, recharge, water-rock interaction, etc.
In addition, all variables were plotted in rotated space in Figure 13 to clearly demonstrate the separation of the four components.

CONCLUSIONS
In this research, 80 VES surveys were conducted using a Schlumberger configuration integrated with the hydrochemical analysis of 58 groundwater samples to indicate seawater intrusion into coastal aquifers.Four pseudo cross-sections were generated from the 80 VES data.These were in a good agreement with those obtained from the hydrogeological data and lithologic data in the study area.The resistivity map at different depths, generated from the VES data, successfully revealed the interaction of seawater and freshwater along the coast line.The geophysical results found that seawater mainly intruded in the Qcl aquifer.The resistivity values of <5 Ωm were found at a depth of approximately 50 m.However, the VES is limited when evaluating the seawater intrusion in highly contaminated aquifer located close to the coast line.Therefore, the evaluation of seawater intrusion in coastal areas with VES data needs the assistance of hydrochemical and hydrogeological data to describe the seawater intrusion more accurately.According to the hydrochemical analysis of 58 groundwater samples, five types of groundwater facies (Ca-Na-HCO3, Ca-HCO3-Cl, Ca-Na-HCO3-Cl, Ca-Na-Cl and Na-Cl) were noticed, which were dependent upon aquifer types and depths.Na-Cl facies were typically found in the Qcl aquifer, corresponding to the resistivity values of <5 Ωm.As the geophysical and hydrochemical results showed, the levels of seawater intrusion could be classified into three zones depending on the degree of seawater intrusion.As a suggestion of this research, groundwater samples should be periodically collected from at least two periods to analyze the dynamic and evolution of seawater intrusion.The VES investigation should be concerned with the distance between the VES survey point and the VES point near the coast line as the depth of survey cannot penetrate through the highly seawater contaminated groundwater.Moreover, the lithologic data for aquifers that have Hydrol.Earth Syst.Sci.Discuss., https://doi.org/10.5194/hess-2018-137Manuscript under review for journal Hydrol.Earth Syst.Sci. Discussion started: 3 May 2018 c Author(s) 2018.CC BY 4.0 License.
St33 from the east to west direction.This profile was between the B-B' and D-D' profiles.The total distance of this line was approximately 7975 m.The surface Hydrol.Earth Syst.Sci.Discuss., https://doi.org/10.5194/hess-2018-137Manuscript under review for journal Hydrol.Earth Syst.Sci. Discussion started: 3 May 2018 c Author(s) 2018.CC BY 4.0 License.(0-10 m) layer had a resistivity range from 6-700 Ωm, which represented the unsaturated zone of Qfd aquifer.The next layer showed resistivity values between 2-20 Ωm and represents the sediment layer saturated with water, and there was a low resistivity zone (<5 Ωm) near the coast line where it extended inland approximately 3.2 kilometers, representing the influence of seawater intrusion.Furthermore, the western side of the profile at a depth 40-200 meters showed resistivity values between 60-200 Ωm, representing the bedrock layer.
Hydrol.Earth Syst.Sci.Discuss., https://doi.org/10.5194/hess-2018-137Manuscript under review for journal Hydrol.Earth Syst.Sci. Discussion started: 3 May 2018 c Author(s) 2018.CC BY 4.0 License.a lot of clay components are required to interpret the results more accurately.In the future, when this area needs to use the groundwater resources, people should use groundwater in the rock aquifers (PCms and Gr aquifers) at depths of higher than 70 meters.The proper criteria for selecting the study area of seawater intrusion should consider the following: groundwater demand in the area, ecological and hydrogeological characteristics, and the amount of groundwater recharge needed to prevent problems that may occur from a large amount of groundwater pumping in the future.

Figure captions Figure 1a .
Figure captions Figure 1a.Study area map including VES points, sample collected well and location of four crosssection lines A-A', B-B', C-C', and D-D'.

Figure 1b .
Figure 1b.Land use map of the study area (Land Development Department, 2011).

Figure 2 .
Figure 2. Geological map in the study area (adapted from the Department Mineral Resources 2007).

Figure 3 .
Figure 3. Hydrological map in the study area (adapted from the Department of Groundwater Resources (DGR), 2014).

Figure 7 .
Figure 7. Resistivity value versus lithologic-log from the pseudo cross-section lines D-D' (well Q168 versus VES St 47 and well PW 7962 versus VES St 34).

Figure 10 .
Figure 10.Hydrochemical analysis of groundwater sample plotted in the piper diagram.

Figure 11 .
Figure 11.The relationship plotting between Na and Cl concentration (in meq/L) in the groundwater samples collected from different aquifers.

Figure 12 .
Figure 12.Hydrochemical Facies Evolution Diagram (HFED) for depicting the salinization process in this area.

Figure 13 .
Figure 13.A component plot in rotated space.

Figure 14 .
Figure 14.The boundary of seawater intrusion in the Qcl aquifer, based on the EC contour map superimposed on the resistivity map.
by the Groundwater Chart Program from United States Geological Survey (USGS).The reliability of the results of the hydrochemical analysis determined the charge balance of cation and anion ( % Δ, which in this study was acceptable at % Δ <10 ( ALS Environmental) .From Table1, it was found that 42 samples were acceptable

Table 5 .
The levels of seawater intrusion with resistivity and EC values Level