• For Contributors +
• Journal Search +
Journal Search Engine
ISSN : 1225-6692(Print)
ISSN : 2287-4518(Online)
Journal of the Korean earth science society Vol.43 No.4 pp.539-556
DOI : https://doi.org/10.5467/JKESS.2022.43.4.539

# Spatio-temporal Variation of Groundwater Level and Electrical Conductivity in Coastal Areas of Jeju Island

Woo-Ri Lim1, Won-Bae Park2, Chang-Han Lee3, Se-Yeong Hamm4*
1Research Institutes of Applied Science, Catholic University of Pusan, Busan 46252, Republic of Korea
2Jeju Groundwater Research Center, Jeju Research Institute, Jeju 63147, Republic of Korea
3Department of Environmental Administration, Catholic University of Pusan, Busan 46252, Republic of Korea
4The Institute of Environmental Studies, Pusan National University, Busan 46241, Republic of Korea
*Corresponding author: hsy@pusan.ac.kr Tel: +82-51-510-1874
August 2, 2022 August 26, 2022 August 26, 2022

## Abstract

In the coastal areas of Jeju Island, composed of volcanic rocks, saltwater intrusion occurs due to excessive pumping and geological characteristics. Groundwater level and electrical conductivity (EC) in multi-depth monitoring wells in coastal areas were characterized from 2005 to 2019. During the period of the lowest monthly precipitation, from November 2017 until February 2018, groundwater level decreased by 0.32-0.91 m. During the period of the highest monthly precipitation, from September 2019 until October 2019, groundwater level increased by 0.46-2.95 m. Groundwater level fluctuation between the dry and wet seasons ranged from 0.79 to 3.73 m (average 1.82 m) in the eastern area, from 0.47 to 6.57 m (average 2.55 m) in the western area, from 0.77 to 8.59 m (average 3.53 m) in the southern area, and from 1.06 to 12.36 m (average 5.92 m) in the northern area. In 2013, when the area experienced decreased annual precipitation, at some monitoring wells in the western area, the groundwater level decreased due to excessive groundwater pumping and saltwater intrusion. Based on EC values of 10,000 μS/cm or more, saltwater intrusion from the coastline was 10.2 km in the eastern area, 4.1 km in the western area, 5.8 km in the southern area, and 5.7 km in the northern area. Autocorrelation analysis of groundwater level revealed that the arithmetic mean of delay time was 0.43 months in the eastern area, 0.87 months in the northern area, 10.93 months in the southern area, and 17.02 months in the western area. Although a few monitoring wells were strongly influenced by nearby pumping wells, the cross-correlation function of the groundwater level was the highest with precipitation in most wells. The seasonal autoregressive integrated moving average model indicated that the groundwater level will decrease in most wells in the western area and decrease or increase in different wells in the eastern area.

## Introduction

Many studies have been conducted on saltwater intrusion in coastal aquifers composed of porous media (e.g., Glover, 1959;Henry, 1959;Strack, 1976;Todd, 1980;Song et al., 2007; Lu et al., 2013; Lee et al., 2016). Saltwater intrusion into coastal aquifers is associated with groundwater-level changes (Goswami and Clement, 2007). Several Korean researchers, such as Kim et al. (2009) and Shim and Chung (2004), simulated the freshwater-seawater interface of coastal aquifers using groundwater modeling.

Unlike other regions of South Korea, groundwater comprises most of the water resources in Jeju Island. On Jeju Island, groundwater is vulnerable to water level drop, pollution, and saltwater intrusion due to various factors, such as urbanization, land use change, and climate change. Studies on high salinity in the groundwater of Jeju Island, particularly on saltwater intrusion in the eastern part, have been conducted since 1980 (Koh et al., 1992, 1998, 2003; Choi and Kim, 1989; Choi et al., 1991;Choi, 1992;Park et al., 2003;Park et al., 2002;Jejudo, 2003;Han et al., 2000). Choi and Kim (1989), Choi et al. (1991), and Choi (1992) attributed the intrusion of saltwater in the eastern part to excessive groundwater pumping. However, Koh (1997), Yoon and Koh (1994), and KWater and Ministry of Construction and Transportation (1993) interpreted high salinity in the eastern region as natural phenomenon due to the distribution of geological formations. Koh (1991) reported that the groundwater level is governed by primary topographical factors and secondary geological factors (underground distribution of the Seogwipo Formation) based on the observation of rock specimens and drill core data. They also reported that the Seogwipo Formation acts as a barrier to fresh groundwater reserves. Ahn et al. (2013) analyzed the groundwater flow and capture zone based on the amount of groundwater pumped on Jeju Island using numerical modeling. Kim et al. (2008, 2009, 2013, 2015) performed an integrated water balance study by applying surface and groundwater modeling and analysis of long-term surface runoff. Song et al. (2014) confirmed that summer precipitation directly affects the groundwater system on Jeju Island.

To preserve the most important water resource on Jeju Island, the Saltwater Intrusion Monitoring Network (SIMN), which monitors coastal groundwater flow associated with geology and groundwater use, has been operating in Jeju Special Self-Governing Province. Kim et al. (2005) revealed that groundwater flow direction and flow velocity in the three monitoring wells of the SIMN in the eastern coastal area of Jeju Island varied depending on strata and tidal effect, even at the same depth. Hwang et al. (2006) reported that flow, groundwater distribution, and characteristics in Jeju Island are controlled by geological characteristics based on SIMN geophysical data. Existing studies cannot clearly elucidate hydrogeological characteristics and causes and spatial distribution of saltwater intrusion in different regions, especially because of the lack of an extensive study on saltwater intrusion in the western, southern, and northern areas of Jeju Island.

This study aims to examine the groundwater level change and saltwater intrusion characteristics of the coastal areas of Jeju Island using groundwater level and electrical conductivity (EC) data from the SIMN. We analyzed the relationship between the precipitation amount, sea level, and pumping quantity from nearby wells, as well as the spatio-temporal fluctuation of the groundwater level and EC.

## Study Area

### Geographic and geological setting

Jeju Island, with a total area of 1,849 km 2 , is the largest among the islands in Korea, with an elliptical shape that is 73-km length in the east-west direction and 31-km length in the north-south direction. Volcanic activity for approximately 1.8 million years since the early Pleistocene, has produced Halla Mountain (shield volcano), which is 1,950 m in height, on Jeju Island. It has also produced approximately 370 large and small volcanic cones (cinder cone, tuff cone, tuff ring, lava dome, maar, etc.) and 130 lava tubes. Volcanic activity has extended lava plateaus in the eastern and western coastal lowlands (Woo et al., 2013). Geologically, Jeju Island is composed of basalt, trachybasalt, basaltic trachyandesite, trachyandesite, and trachyte due to volcanic eruptions that occurred 200,000-300,000 years ago (Fig. 1(a)). During the dormant period of the volcanic eruption, a paleosol was formed between the lava layers. The Seogwipo Formation is a sedimentary layer deposited in the early stage (0.5-1.8 million years ago) of the volcanic eruptions on Jeju Island and is distributed in certain surface areas of the southern coast of Jeju Island and underground (Sohn and Park, 2004;Koh et al., 2013). It consists of pyroclastic materials from shallow submarine eruptions and marine sediments during the dormant period of the volcanic eruption. The unconsolidated U layer under the Seogwipo Formation is a marine deposit mostly composed of clay and finegrained sand. It accumulated primarily in the shallow continental shelf before the volcanic eruption on Jeju Island (Koh, 1997). The bedrock of Jeju Island lies under the U layer and is composed of volcaniclastics and granite from the Mesozoic and early Cenozoic. The volcaniclastics are correlated with the Cretaceous Yucheon Group (Koh, 1997). The granite, which is judged to have intruded the volcaniclastics in the early Tertiary, was formed at a similar time and environment as micrographic granite (Masanite) distributed along the southern coast of Gyeongsangnam-do Province (Ahn et al., 1995).

The permeable structures of Jeju Island are mainly clinker, scoria, hyaloclastite, and lava tube, while the low-permeable sedimentary layers are tuff, volcanic breccia, Seogwipo Formation, Seongsan Formation, and paleosol (Jeju Special Self-Governing Province, 2013, Fig. 1(b)). The low permeable layers are distributed primarily in the central part of Jeju Island and are smaller in size in the eastern and northern regions (Kim et al., 2015). The rivers from the summit of Mt. Halla have a steep slope in the northsouth direction and length approximately 15 km, which is much shorter than the river lengths in the mainland. These rivers run dry and flow only during short periods of heavy rainfall.

### Hydrogeological setting

Unlike other regions of South Korea, Jeju Island uses groundwater for most of its water requirements. According to Jejudo (2003), Jeju Island is divided into the eastern (Gujwa, Seongsan, and Pyoseon watersheds), western (Hanlim, Hangyeong, Daejeong, and Andeok watersheds), southern (West-seogwi, Middle-seogwi, East-seogwi, and Namwon watersheds), and northern (Aewol, West-jeju, Middle-jeju, East-jeju, and Jocheon watersheds) areas (Fig. 2).

For the monitoring of saltwater intrusion, 53 monitoring wells of the SIMN equipped with multidepth sensors are located within the range of 0.1-7.4 km from the coastline: 20 wells in the eastern, 15 in the western, 8 wells in the southern, and 10 wells in the northern areas. Groundwater at the 53 wells along the coast is mostly composed of basal and parabasal groundwaters (Fig. 2). Basal groundwater is predominantly distributed in basalt and scoria, which exhibit good permeability unlike the Seogwipo Formation. Parabasal groundwater is not directly underlain by seawater because of the low permeability of the Seogwipo Formation; thus, the Ghyben-Herzberg principle is not applied to parabasal groundwater areas (Koh, 1991). The depths and elevations of the wells range from 60 to 330 m and from 7.5 to 167.2 m above mean sea level (amsl), respectively. The well casing was installed at a depth of 4-127 m, and the screen was installed from the bottom casing to the well depth.

Precipitation on Jeju Island was measured at Gosan (GS), Jeju (JJ), Seogwipo (SGP), and Seongsan (SS) stations using an automated synoptic observing system (ASOS) and at 19 automatic weather stations (AW, DH, DJ, GJ, GJw, HL, JM, NW, OD, PS, SBOR, SCD, SD, SG, SH, SR, TPC, WJ, and YSA) (Fig. 3). From 2005 to 2019, the average precipitations were 1,494 mm (JJ station), 1,194 mm (GS station), 2,083 mm (SS station), and 2,017 mm (SGP station) (Fig. 3). The large differences in precipitations between (JJ and GS stations) and (SS and SGW stations) is thought due to the Fohn phenomenon caused by Mt. Halla and the drought characteristics (Park et al., 2011). The highest average total annual precipitation was 2,206 mm in 2012, and the lowest was 1,002 mm in 2013. The highest monthly precipitation occurred in August 2004 and the lowest monthly precipitation occurred in November 2007.

At Jeju Island, as of the end of 2019, 4,615 wells are being used for agriculture, fishery, housing, industry, bottled water, etc. (Ministry of Environment and K-Water, 2020). Agriculture and fishery wells accounted for 66.2% of the total wells. Based on monthly groundwater usage data from January 2013 reported by the water policy division of Jeju Special Self-Governing Province (Jeju Research Institute, 2017), 30 monitoring wells located within a radius of 7-1,720 m from pumping wells were considered for groundwater level analysis and their influence on the quality of pumping wells in this study (Table 1, Fig. 4).

## Methods

### Auto-correlation and cross-correlation analyses

The time series characteristics of groundwater data was analyzed using autocorrelation and crosscorrelation in the time domain (Larocque et al., 1998;Lee and Lee, 2000). The autocorrelation function represents the memory effect persistence of time series data, such as the groundwater level data, by the degree and duration of the effect due to events such as precipitation. In the 95% confidence interval, the period when the autocorrelation function decreases from 1 to 0 is called delay time (Cryper and Chan, 2008). The longer the delay time, the higher the longterm autocorrelation owing to the strong linearity and memory effect of time series data (Choi et al., 2011). The autocorrelation function r(k) was calculated as follows:

$r ( k ) = c ( k ) c ( 0 )$
(1)

$C ( k ) = 1 n ∑ t = 1 n − k ( x t − X ¯ ) ( x t + k − X ¯ )$
(2)

where k is the lag, n is the length of the time series xt, and X is the arithmetic mean of the time series.

The cross-correlation function represents the interdependence of two time-series data: the input timeseries of precipitation, sea level, atmospheric temperature, and nearby pumping amount, and the output timeseries of groundwater level, EC, and groundwater temperature. The delay time was obtained using the maximum cross-correlation function. The cross-correlation function rxy(k) was calculated as follows:

$r x y ( k ) = c x y ( k ) σ x σ y$
(3)

$C x y ( k ) = 1 n ∑ n = 1 n − k ( x t − X ¯ ) ( y t + k − Y ¯ )$
(4)

where σx and σy are the standard deviations of input xt and output yt time-series data, respectively, and Y is the average value of the yt time series. The delay time is the time difference, k, when the cross-correlation function is the largest. The shorter the delay time, the faster is the response time of the output time series with respect to the input time series. The crosscorrelation function lies between −1 and 1, and the larger the value, the greater the influence of the input time series on the output time series.

### Seasonal auto-regressive integrated moving average model

Stationary time series models include autoregressive (AR), moving average (MA), and autoregressive moving average (ARMA) models. The autoregressive integrated moving average (ARIMA) model is a stationary time series that removes the trend of a nonstationary time series through differencing (Shumway and Stoffer, 2011). The ARIMA model is used to characterize various properties of time-series data and to predict future values (Yi and Lee, 2004). Future groundwater-level fluctuations can be estimated by constructing an ARIMA model suitable for groundwaterlevel time series. The pattern of a groundwater time series is governed by geology, topography, hydrogeological factors, and anthropogenic causes, such as abstraction from the nearby pumping well.

When time-series data show abnormalities during a specific period or fluctuate periodically due to seasonal factors. This is called the seasonal ARIMA (SARIMA) model, which is calculated as follows (Shumway and Stoffer, 2011):

$Φ ( B s ) = 1 − Φ 1 B s − Φ 2 B 2 s − … − Φ p B P s$
(5)

$Θ ( B s ) = 1 − Θ 1 B s − Θ 2 B 2 s − … − Θ q B Q s$
(6)

$ϕ ( B ) = 1 − ϕ 1 B − … − ϕ p B p$
(7)

$θ ( B ) = 1 − θ 1 B − … − θ q B q$
(8)

(9)

where D is the seasonal difference, s is the period, Φ(Bs) is the operator of the seasonal AR model, Θ(Bs) is the operator of the seasonal MA model, φt is the weighting factor at time t. B is the postfix operator, and BjZt = Zt-j. P and Q represent the order of the seasonal AR and MA models, respectively.

### Analysis of tidal variation

Tides arise due to the relative motion of the sun and moon, which occur in various combinations depending on their positions relative to earth. The amplitudes of tides change over time according to different tidal cycles. Tidal oscillations are decomposed into four components (M2, O1, S2, and K1) (Lee et al., 2015). In general, in coastal areas, it is often difficult to estimate the influence of precipitation and groundwater pumping on groundwater level fluctuations owing to the influence of tides (Jha et al., 2003). Harmonic analysis was used to calculate the net groundwater level by removing tidal effects. The tide levels predicted over time through harmonic analysis are as follows (Jahromi et al., 2010):

$y ( t ) = y 0 + ∑ j = 1 m f j H j cos ( w j t + θ j − δ j ) = y 0 + ∑ j = 1 m R j cos ( w j − ϕ j )$
(10)

where y0 is the average sea level height; fj, Hj, wj, and θj are the node factor, amplitude, frequency, and phase of the jth subtidal wave, respectively and m is the number of subtidal waves used in the analysis. The effect of tides during a period, even at large amplitudes, can effectively be removed using the T_Tide program, which applies standard highfrequency and band-filtering techniques (Pawlowicz et al., 2002).

## Results

### Spatial and temporal distribution of groundwater level

The monitoring wells of the SIMN equipped with multi-depth sensors were used to assess the spatial and temporal distribution of the groundwater level and EC. Among the 53 monitoring wells, a total of 30 monitoring wells (13 wells (JD-GN2, JR-GJ1, JDHD2, 3, JW-PD, JD-JD3, 4, JD-SS2, JD-SD2, JDHC1, 2, JD-SeHw1, 2) in the eastern; 8 (JD-HL1, JDHJ, JD-YS1, JW-HJD, JD-MR1, JD-YR1, JD-SM2, and JD-HM2) in the western; 3 (JD-WP, JM-NW, and JR-TH1) in the southern; and 6 (JR-SG1, JM-SU, JRHG1, JD-SY1, JD-HmD1, 2) in the northern areas) were located within a radius of 7-1,720 m from the pumping wells.

The spatial distribution of groundwater levels was determined using data from the 132 single-depth and multi-depth monitoring wells and by applying the kriging method (Fig. 5). The average groundwater levels in the eastern area ranged from 0.21 to 2.61 m amsl (Fig. 5(a)). In the Gujwa watershed, groundwater levels with and without tidal effect were 0.48-6.38 m amsl and 0.41-6.23 m amsl, respectively. In the Seongsan watershed, groundwater levels with and without tidal effect were 0.26-5.90 and 0.43-5.64 m amsl, respectively. In the Pyoseon watershed, groundwater levels before and after removing tidal effect were −0.21-3.01 and −0.10-2.96 m amsl, respectively.

The average groundwater levels in the western area was −0.59-6.77 m amsl in the Hanlim watershed. Groundwater levels before and after removing tidal effect were 0.55-8.22 and 0.62-8.22m amsl, respectively. In the Hangyeong watershed, groundwater levels before and after removing tidal effect were −2.56-8.70 and −2.32-9.42 m amsl, respectively. In the Daejeong watershed, the groundwater levels before and after removing the tidal effect were −2.57-9.82 and −2.83- 9.45 m amsl.

The average groundwater level in the southern area was 0.39-38.58 m amsl. The slope in groundwater level in the southern area was steep, and in the Andeok and Namwon watershed, the 10 m amsl groundwater contour was located close to the coast. The steep slope in groundwater level in the southern area can be attributed not only to the steep topographic slope but also to the distribution of upper groundwater. In the West- and Middle-Seogwi watersheds, groundwater levels before and after removing tidal effect were 8.16-42.25 and 8.07-41.97 m amsl, respectively. In the East-Seogwi watershed, groundwater levels before and after removing tidal effect were 1.17-12.89 and 0.63-11.44 m amsl, respectively. In the Namwon watershed, groundwater levels before and after removing tidal effect were 0.01-2.52 and 0.06-2.63 m amsl, respectively.

The average groundwater levels in the northern area ranged from 0.86 to 8.41 m amsl. In the Aeweol watershed, groundwater levels before and after removing tidal effect were 0.35-17.87 and 0.48-18.17 m amsl, respectively. In the Middle-Jeju and East-Jeju watersheds, groundwater levels before and after removing tidal effect were 0.79-8.87 and 0.42-8.74 m amsl. In east Jocheon watershed, groundwater levels before and after removing tidal effect were 0.20-9.69 and 0.25-9.42 m amsl, respectively.

From 2017 to 2019, the lowest precipitation of 1- 41.5mm (average 17.8mm) was recorded in November 2017 and the highest precipitation of 342-1,631.5 mm (average 837.5 mm) was recorded in September 2019. The groundwater level continued to drop for up to four months from November 2017 to February 2018. Groundwater levels in the eastern, western, southern, and northern areas in February 2018 were −0.21-1.91, −2.57-5.24, 0.01-33.73, and 0.39-3.81m amsl, respectively, showing the largest decline of 0.32-0.91 m amsl in groundwater level overall (Fig. 5(b)). However, decreased precipitation in September 2019, led to a rise in the groundwater level that persisted until October 2019. In October 2019, groundwater levels in the eastern, western, southern, and northern areas were 0.58-5.52, −1.36-8.61, 0.80-38.45, and 1.81-16.17 m amsl, respectively. The overall maximum rise in groundwater level was 0.46-2.95 m amsl (Fig. 5(c)).

The groundwater level varied between dry and wet seasons by 0.79-3.73 m (average 1.82 m) in the eastern, 0.47-6.57 m (average 2.55 m) in the western, 0.77-8.59 m (average 3.53 m) in the southern, and 1.06-12.36 m (average 5.92 m) in the northern areas, with a higher variation in the northern and southern areas than in the eastern and western areas.

### Distribution of groundwater electrical conductivity

According to the EC classification by Mondal et al. (2010), 23 monitoring wells (1 well in the eastern, 9 in the western, 6 in the southern, and 7 in the northern areas) contained freshwater (EC values less than 1,500 μS/cm), and three monitoring wells in each of eastern, western, and northern areas contained brackish water (EC values of 1,500-3,000 μS/cm). Twenty-seven monitoring wells (18 in the eastern, 5 in the western, 2 in the southern, and 2 in the northern areas) contained saline water (EC values over 3,000 μS/cm). Thus, 52.6% of monitoring wells were affected by seawater (Fig. 6).

The EC values of the Gujwa watershed in the eastern area ranged from 119 to 62,093 μS/cm and tended to increase with depth, with different EC values at different depths based on the shape of the salt-freshwater mixing zone. The EC values of the Seongsan watershed ranged from 109 to 55,592 μS/ cm. It gradually decreased as well depth increased. The EC values of the Pyoseon watershed ranged from 113 to 57,521 μS/cm, showing a tendency to increase with depth.

The EC values of the Hanlim watershed in the western area ranged between 249 and 19,167 μS/cm. The EC values of the Hangyeong and Daejeong watersheds were 185-27,504 and 84-55,848 μS/cm, respectively. The EC values of the West Seogwi-and Middle-Seogwi watersheds and the East-Seogwi watershed in the southern area were 80-251and 72- 357 μS/cm, respectively, indicating no influence of saltwater intrusion. In contrast, the EC values of the Namwon watershed ranged from 78 to 43,371 μS/cm. The EC values of the Aeweol watershed in the northern area ranged from 113 to 1,134 μS/cm, indicating a weak influence of saltwater intrusion. The EC values of the Middle-and East-Jeju and Jocheon watersheds ranged from 40 to 454 μS/cm and from 105 to 40,087 μS/cm, respectively.

The temporal fluctuation of EC values varied depending on the depth of the wells. The spatial distribution of EC values by depth, calculated using the kriging method, is displayed in Fig. 7. Based on the average EC values at −20 and −80 m amsl for the entire coastal area of Jeju Island, the overall EC increased in February 2018 when the groundwater level was lowest and the EC values decreased in October 2019 when the groundwater level was highest. Unlike the eastern area, which exhibited saltwater intrusion at most depths, the western, southern, and northern areas exhibited only partial saltwater intrusions.

### Auto-correlation analysis

The results of the autocorrelation analysis of the groundwater level after removing tidal effect (ARTE) showed that the eastern area displayed a delay time of 3.22-29.23 months (Fig. 8, Table 2). Based on the autocorrelation function, groundwater levels in all monitoring wells except in JD-JD2 in the Gujwa, JDSD1 and JD-SD2 in the Seongsan, and JD-SeHw1 in the Pyoseon watersheds, showed a 12-month periodicity with groundwater levels both before and after removal of the tidal effect. In the western area, the delay time of groundwater-level ARTE was 4.16-43.37 months. The groundwater level, excluding the tidal effect, was affected for a long time by precipitation and other factors, such as geological and topographical characteristics.

In the southern area, the delay time of groundwater level ARTE was 3.47-16.79 months, with values exhibiting a periodicity of 12 months with reference to groundwater level fluctuations. In the northern area, the delay time of groundwater level ARTE was 3.70- 13.84 months, exhibiting a 12-month periodicity with reference to groundwater level fluctuations at most monitoring wells.

### Cross-correlation analysis

Cross-correlation between groundwater level ARTE and precipitation showed a cross-correlation function of 0.244-0.708 at a delay time of 0-2 months in the eastern, 0.103-0.570 at a delay time of 0-3 months in the western, 0.186-0.602 at a delay time of 1-2 months in the southern, and 0.428-0.640 at a delay time of 1-3 months in the northern areas (Fig. 9).

The cross-correlation functions of the groundwater level ARTE and sea level were 0.253-0.563 at a delay time of 0-1 month in the eastern, −0.103-0.394 at a delay time of 0-3 months in the western, −0.129-0.351 at a delay time of 0-3 months in the southern, and 0.176-0.365 at a delay time of 0-4 months in the northern areas.

The cross-correlation functions of the groundwater level ARTE and nearby pumping amount (at a distance of 7-1,720 m from the monitoring well) were −0.35-0.602 at a delay time of 0-2 months in the eastern, −0.555-0.347 at a delay time of 0-5 months in the western, −0.489-0.203 at a delay time of 0-3 months in the southern, and −0.375-0.679 at a delay time of 0-2 months in the northern areas. Hence, the cross-correlation functions between the groundwater ARTE and precipitation were higher than other input factors (Fig. 9).

Some of the wells revealed positive cross-correlation functions between groundwater level and pumping quantity at nearby pumping wells. The groundwater level rise near pumping wells may be caused by multiple factors, such as recharge from streams and reservoirs during pumping or saltwater intrusion in accordance with sea level rise (Kim et al., 2012).

Cross-correlation analysis between EC and raw groundwater level largely revealed negative crosscorrelation, but some of the monitoring wells exhibited a positive correlation function with sea level rise or groundwater pumping from nearby wells (Fig. 10).

### SARIMA model of groundwater level

The variation in groundwater level over time was analyzed by applying the SARIMA model. The SARIMA model was defined using the Akaike’s information criterion (AIC; Akaike, 1973) and Schwartz’s Bayesian criterion (SBC; Schwartz, 1978). The order of differentiation for converting unsteady time series into stationary time series was 1-3, and the autocorrelation function exhibited a periodicity of 12 or 24 months. For 15-year data, the validation and forecasting periods were set to 10 years. In the case of 10-year data, the validating and forecasting periods were set to five years, and in the case of 5-year data, the validating and forecasting periods were set to two years.

The SARIMA model for the groundwater level was determined using the model judgment coefficients AIC and SBC. The model with the minimal AIC and SBC was selected (Fig. 11, Table 3). The trend of the linear regression line was determined for the predicted groundwater levels. In the eastern area, the groundwater level of the monitoring wells was expected to increase, as seen in the JR-GJ1 well (with AIC value of −4.08 and SBC value of 2.14) or decrease. However, in the western area, the groundwater level of the monitoring wells was expected to decrease, similar to that of a typical JW-HJD well (with AIC value of 14.43 and SBC value of 16.70).

## Conclusions

From 2005 to 2019, groundwater level changes in between dry and wet seasons in Jeju Island showed a higher variation in the northern and southern areas than in the eastern and western areas. Based on the average EC values at −20 and −80 m amsl for the entire coastal area of Jeju Island, the overall EC values increased in February 2018, when groundwater level was the lowest. EC values decreased in October 2019, when groundwater level was the highest. Unlike the eastern area, which exhibited saltwater intrusion in the deepest levels, some of the monitoring wells in the western, southern, and northern areas exhibited saltwater intrusion at different depths depending on the monitoring wells.

The autocorrelation analysis of groundwater level revealed a delay time of 0.43 months in the eastern, 0.87 months in the northern, 10.93 months in the southern, and 17.02 months in the western areas, indicating a longer delay time that may be affected by precipitation and other external factors over a long period. The cross-correlation function between groundwater ARTE and precipitation mostly exhibited higher values than those of the other cross-correlations. Some of the monitoring wells exhibited a positive crosscorrelation function with the amount of pumped water from the nearby pumping well, which is expected to be influenced by external factors such as recharge from streams and reservoirs during pumping and saltwater intrusion in accordance with the sea-level rise.

According to the prediction of the SARIMA model, the groundwater level will decrease in most of the wells in the western area, whereas the groundwater level in the eastern area will decrease or increase depending on the wells.

Groundwater level fluctuations and saltwater intrusion in the coastal aquifer of Jeju Island are governed by various hydrogeological factors such as geological structures, aquifer depth, and hydraulic characteristics of the aquifer, the depth of Seogwipo Formation, and groundwater use. In particular, the western area is vulnerable to saltwater intrusion due to a deeper location of the Seogwipo Formation than common fresh water-saltwater boundary. Hence, the results of this study can help establish an appropriate groundwater management plan for Jeju Island, with a better understanding of the quantitative geological and hydrogeological characteristics of the monitoring wells and groundwater level, using quality data over a longer time period.

## Acknowledgments

This research was funded by the National Research Foundation of Korea (NRF) under the Ministry of Science and ICT (no. NRF-2020R1A2B5B02002198).

## Figure

Distribution of (a) hydrogeological units by rock formations and (b) permeable terrains in Jeju Island (modified after Park et al., 2000;Lim, 2021).

Groundwater occurrence types (Koh, 1997) and location of multi-depth monitoring wells in Jeju Island (Lim, 2021).

Annual precipitation (mm) of the four stations in Jeju Island.

Location of meteorological stations and wells for groundwater use in Jeju Island.

Groundwater level distribution for (a) average of entire period, (b) February 2018, and (c) October 2019.

Classification of groundwater based on EC in Jeju Island.

Spatial distribution of EC at depth of -20m (left) and -80m (right) amsl throughout the study period, (a, d) average, (b, e) February 2018, and (c, f) October 2019.

Auto-correlation function and delay time of groundwater level ARTE in the (a) eastern, (b) western, (c) southern, and (d) northern areas.

Classification of cross-correlation function of groundwater level ARTE for the dominant input factors. (a) precipitation, (b) sea level, and (c) amount of groundwater usage.

Classification of cross-correlation function of groundwater EC for the dominant input factors. (a) raw groundwater level, (b) precipitation, (c) sea level, and (d) amount of groundwater usage.

Prediction of groundwater level using SARIMA models for (a) JW-HJD in western area and (b) JR-GJ1 in eastern area.

## Table

Specifications of groundwater use wells near the monitoring wells.

Calculated delay times (month) from the auto-correlation function of groundwater level ARTE

Parameters of SARIMA model and groundwater level trend test results

## Reference

1. Ahn, K. S. , Lee, H. K. , and Lim, H. C. ,1995, A petrological and geochemical study of granites in the Cheju Island. Economic and Environmental Geology, 28, 513-518.
2. Ahn, S. S. , Park, D. I. , and Oh, Y. H. ,2013, Characteristics of ground water capture zone according to pumping rate. Journal of Environmental Science International, 22, 895-903.
3. Choi, H. M. , Lee J. Y. , Ha, K. , and Kim, G. P. ,2011, The study on time series analysis of groundwater data and groundwater recharge in Jeju Island, The Journal of Engineering Geology, 21, 337-348.
4. Choi, S. H. , Kim, Y. K. , and Lee, D. Y. ,1991, Sea water intrusion in the coastal area of Cheju volcanic island, Korea. Economic and Environmental Geology, 24, 319- 327.
5. Choi, S. H. ,1992, Geological structure and groundwater resources of Cheju and Oahu Island, The Journal of Engineering Geology, 2, 70-91.
6. Glover, R. E. ,1959, The pattern of freshwater flow in a coastal aquifer. Journal of geophysical research, 64, 457-459.
7. Goswami, R. R. , and Clement, T. P. ,2007, Laboratory scale investigation of saltwater intrusion dynamics. Water Resources Research, 43.
8. Han, K. E. , Yi, S. S. , and Jeong, C. Y. ,2000, Protection for sea-water intrusion by geophysical prospecting & GIS, Korean Society of Earth and Exploration Geophysicists Symposium, 54-69.
9. Henry, H. R. ,1959, Salt intrusion into freshwater aquifers. Journal of Geophysical Research, 64, 1911-1919.
10. Hwang, S. , Shin, J. , Park, K. H. , Park, I. , Koh, G. W. ,2006, Geophysical well logs in basaltic volcanic area, Jeju Island, Mulli-Tamsa, 9, 231-240.
11. Jahromi, M. J. , Maswood, A. I. , and Tseng, K. J. ,2010, Long term prediction of tidal currents. IEEE Systems Journal, 5, 146-155.
12. Jeju Special Self-Governing Province, 2013, Water Resources Management Master Plan, 47-87.
13. Jejudo,2003, Comprehensive Survey of Hydrogeology and Groundwater Resources in Jeju Island (III), 173-179.
14. Jha, M. K. , Kamii, Y. , and Chikamori, K. ,2003, On the estimation of phreatic aquifer parameters by the tidal response technique. Water Resources Management, 17, 69-88.
15. K-Water and Ministry of Construction and Transportation,1993, Jeju Island Water Resources Comprehensive Development Plan Establishment Report, VIII3-24.
16. Kim, K. Y. , Park, Y. S. , Kim, G. P. , and Park, K. H. ,2009, Dynamic freshwater-saline water interaction in the coastal zone of Jeju Island, South Korea. Hydrogeology Journal, 17, 617-629.
17. Kim, N. W. , Chung, I. M. , and Na, H. ,2015, An integrated water budget analysis of Oedocheon watershed in Jeju Island. Journal of Environmental Science International, 24, 471-480.
18. Kim, N. W. , Chung, I. M. , Won, Y. S. , and Arnold, J. G. ,2008, Development and application of the integrated SWAT-MODFLOW model. Journal of Hydrology, 356, 1-16.
19. Kim, N. W. , Chung, I. M. , Yoo, S. Y. , Lee, J. W. , and Yang, S. K. ,2009, Integrated surface-groundwater analysis in Jeju Island. Journal of Environmental Science International, 18, 1017-1026.
20. Kim, N. W. , Lee, J. W. , Chung, I. M. , and Kim, C. H. ,2012, Change of groundwater-streamflow interaction according to groundwater ion in a green house land. Journal of Korea Water Resources Association, 45, 1051-1067.
21. Kim, N. W. , Na, H. , and Chung, I. M. ,2013, Simulation of groundwater variation characteristics of Hancheon watershed in Jeju Island using integrated hydrologic modeling. Journal of Environmental Science International, 22, 515-522.
22. Koh, G. W. ,1991, The relationship between the underground distribution of the Seogwipo Formation in Jeju Island and groundwater, Journal of the Geological Society of Korea, 27, 528.
23. Koh, G. W. , Park, W. B. , Kim, H. W. , and Chae, J. I. ,1992, The relationship between the subsurface geological structure of Jeju Island and groundwater level fluctuations (I): groundwater level fluctuations due to precipitation. Proceeding of the Geological Society of Korea Conference, 48-48.
24. Koh, G. W. ,1997, Characteristics of the groundwater and hydrogeologic implications of the Sequipo formation in Cheju Island, Dissertation, Pusan National University, Busan.
25. Koh, G. W. , Park, Y. S. , and Kim, T. Y. ,1998, A study on the deep freshwater aquifer developed in the coastal area of Gwideok-ri in the northwestern part of Jeju Island. Jeju Development Research, 223-233.
26. Koh, G. W. , Park, Y. S. , Park, W. B. , and Moon, D. C. ,2003, A study on hydrogeology in the eastern part of Jeju Island (II). Proceedings of the Korean Society of Soil and Groundwater Environment Conference, 67-72.
27. Koh, G. W. , Park, J. B. , Kang, B. R. , Kim, G. P. , and Moon, D. C. ,2013, Volcanism in Jeju Island. Journal of the Geological Society of Korea, 49, 209-230.
28. Larocque, M. , Mangin, A. , Razack, M. , and Banton, O. ,1998, Contribution of correlation and spectral analyses to the regional study of a large karst aquifer (Charente, France). Journal of Hydrology, 205, 217-231.
29. Lee, J. Y. and Lee, K. K. ,2000, Use of hydrologic time series data for identification of recharge mechanism in a fractured bedrock aquifer system. Journal of Hydrology, 229, 190-201.
30. Lee, S. H. , Kwak, H. J. , Kim, J. D. , Lee, E. C. , Kim, J. C. , and Jeon, K. S. ,2015, The study on Estimate of Tidal Power Using Matlab T-tide Based on Tide Prediction, Proceedings of the KIEE Conference, 15-17.
31. Lee, S. , Park, D. , Lee, Y. D. , and Won, K. S. ,2016, Preliminary evaluation of groundwater solute behavior in a coastal confined aquifer of the Nakdong River area. Journal of the Geological Society of Korea, 52, 237-245.
32. Lim, W. R. ,2021, Spatio-Temporal Characteristics of Groundwater Level, Electrical Conductivity, and Temperature in Coastal Aquifers of Jeju Island, Korea (Ph.D. dissertation). Pusan National University, Pusan, Korea.
33. Mondal, N. C. , Singh, V. P. , Singh, V. S. , and Saxena, V. K. ,2010, Determining the interaction between groundwater and saline water through groundwater major ions chemistry. Journal of Hydrology, 388, 100-111.
34. Park, K. H. , Lee, B. J. , Kim, J. C. , Cho, D. L. , Lee, S. R. , Choi, H. I. , Park, D. W. , Lee, S. R. , Choi, Y. S. , Yang, D. Y. , Kim, J. Y. , Seo, J. Y. , and Shin, H. M. ,2000, Explanatory note of the Jeju (Baekado, Jinnampo) sheet (1:250,000). Korea Inst. Geosci. Min. Resources, Taejon, p.59.
35. Park, W. B. , Kim, G. P. , Lee, J. H. , Moon, D. C. , Kim, S. J. , Koh, G. W. , Pang, S. J. , and Pang, I. C. ,2011, Variation of groundwater level and recharge volume in Jeju Island. Journal of Environmental Science International, 20, 857-872.
36. Park, W. B. , Park, Y. S. , Koh, G. W. , and Moon, D. C. ,2003, Distribution and change of freshwater-saltwater boundary in the eastern region of Jeju Island, Proceedings of the Korean Society of Soil and Groundwater Environment Conference, 55-58.
37. Park, Y. S. , Koh, G. W. , Kang, B. R. , and Hamm, S. Y. ,2002, A study on hydrogeology in the eastern part of Jeju Island (I), Proceedings of the Korean Society of Soil and Groundwater Environment Conference, 81-84.
38. Pawlowicz, R. , Beardsley, B. , and Lentz, S. ,2002, Classical tidal harmonic analysis including error estimates in MATLAB using T_TIDE. Computers & Geosciences, 28, 929-937.
39. Shim, B. O. , and Chung, S. Y. ,2004, Hydrogeologic parameter estimation by using tidal method in a fractured rock aquifer. Journal of Soil and Groundwater Environment, 9, 27-32.
40. Shumway, R. H. and Stoffer, D. S. ,2011, Time Series Analysis and Its Applications. Springer, New York, 596 p.
41. Sohn, Y. K. , and Park, K. H. ,2004, Early-stage volcanism and sedimentation of Jeju Island revealed by the Sagye borehole, SW Jeju Island, Korea. Geosciences Journal, 8, 73-84.
42. Sohn, Y. K. , and Yoon, S. H. ,2010, Shallow-marine records of pyroclastic surges and fallouts over water in Jeju Island, Korea, and their stratigraphic implications. Geology, 38, 763-766.
43. Sohn, Y. K. , Cronin, S. J. , Brenna, M. , Smith, I. E. M. , Nemeth, K. , White, J. D. L. , Murtagh, R. M. , Jeon, Y. M. , and Kwon, C. W. ,2012, Ilchulbong tuff cone, Jeju Island, Korea, revisited: a compound monogenetic volcano involving multiple magma pulses, shifting vents, and discrete eruptive phases. GSA Bulletin, 124, 259-274.
44. Sohn, Y. K. , Park, K. H. , and Yoon, S. H. ,2008, Primary versus secondary and subaerial versus submarine hydrovolcanic deposits in the subsurface of Jeju Island, Korea. Sedimentology, 55, 899-924.
45. Song, S. H. , Lee, J. Y. , and Yi, M. J. ,2007, Evaluation of long-term data obtained from seawater intrusion monitoring network using variation type analysis. Journal of the Korean Earth Science Society, 28, 478- 490.
46. Strack, O. D. L. ,1976, A single-potential solution for regional interface problems in coastal aquifers. Water Resources Research, 12, 1165-1174.
47. Todd D. K. ,1980, Groundwater Hydrology. Chapter 14, Saline Water Intrusion in Aquifers, 2nd ed. New York: John Wiley and Sons, pp 496-498.
48. Werner, A. D. , and Simmons, C. T. ,2009, Impact of sealevel rise on sea water intrusion in coastal aquifers. Groundwater, 47, 197-204.
49. Wood, C. , and Harrington, G. A. ,2015, Influence of seasonal variations in sea level on the salinity regime of a coastal groundwater-fed wetland. Groundwater, 53, 90-98.
50. Yi, M. J. , and Lee, K. K. ,2004, Transfer function-noise modelling of irregularly observed groundwater heads using precipitation data. Journal of Hydrology, 288, 272-287.
51. Yoon, S. , and Koh, G. W. ,1994, Structural history of Jeju Island. Proceeding of the Geological Society of Korea Conference, 42-43.