Evaluation of Seasonal Groundwater Quality Changes Associated with Groundwater Pumping and Level Fluctuations in an Agricultural Area, Korea

Abstract

This study was conducted to evaluate seasonal groundwater quality due to groundwater pumping and hydrochemical characteristics with groundwater level fluctuations in an agricultural area in Korea. Groundwater levels were observed for about one year using automatic monitoring sensors, and groundwater uses were estimated based on the monitoring data. Groundwater use in the area is closely related to irrigation for rice farming, and rising groundwater levels occur during the pumping, which may be caused by the irrigation water of rice paddies. Hydrochemical analysis results for two separate times (17 July and 1 October 2019) show that the dissolved components in groundwater decreased overall due to dilution, especially at wells in the alluvial aquifer and shallow depth. More than 50% of the samples were classified as CaHCO₃ water type, and changes in water type occurred depending on the well location. Water quality changes were small at most wells, but changes at some wells were evident. In addition, the groundwater quality was confirmed to have the effect of saltwater supplied during the 2018 drought by comparison with seawater. According to principal component analysis (PCA), the water quality from July to October was confirmed to have changed due to dilution, and the effect was strong at shallow wells. In the study areas where rice paddy farming is active in summer, irrigation water may be one of the important factors changing the groundwater quality. These results provide a qualitative and quantitative basis for groundwater quality change in agricultural areas, particularly rice paddies areas, along with groundwater level and usage.

1. Introduction

In Korea, precipitation is concentrated in summer and varies greatly from season to season. In areas where there is a lot of rice paddy farming, irrigation water in summer is in high demand. If it rains less during that time, irrigation water should be drawn from other areas or secured from groundwater by developing wells in the area. When groundwater is pumped, its quality before and after pumping may change with the seasons. As groundwater pumping continues, the quality may change, which is caused by mixing from different aquifers, inflow of surface water from rivers or reservoirs [1,2,3,4], seawater intrusion [5,6,7], and so on. Also, groundwater quality may change temporally due to hydrologic processes, including seasonally varying rainfall and irrigation patterns [8,9,10,11].

Hydrochemical facies in unconfined aquifers vary between young and old groundwater along the groundwater flow path, and the concentration of dissolved components, total dissolved solids (TDS), and electrical conductivity (EC) are related to the groundwater quality and residence time [1,12]. In an agricultural region, nitrate in groundwater is decreased during dry/growing seasons of spring and fall due to the mixing of groundwater with deeper groundwater by heavy pumping, whereas it increases due to infiltration of water through the soil zone with highly concentrated nitrate [3]. In addition, the chemical composition and groundwater flow patterns are changed by land use and groundwater use [2]. In agricultural riparian zones, where intense groundwater pumping frequently occurs for the warming of greenhouses during the winter season, groundwater drawdown results in losing stream conditions and groundwater temperature decreases with considerable spatial variability, followed by hydrochemical and microbial community changes in groundwater [4,8].

In coastal aquifers, seawater intrusion by groundwater pumping has been reported in many studies [5], and the studies have been done to find the processes and factors associated with seawater intrusion [6]. However, it is sometimes tough to distinguish the effect of pumping in these coastal aquifers. The aquifer in Fukuoka, Japan, was originally below the current seafloor, and deeper saline groundwater was isolated and not mixed with the shallower fresh groundwater, then the increased salinity by pumping did not result in seawater intrusion, but high-saline old groundwater was introduced [13].

Many cases in the literature show that groundwater quality changes with the seasons [14]. Although many papers have been published on the change in groundwater quality, there are not many that have been dealt with the change in groundwater quality based on level fluctuations caused by groundwater pumping. This study was conducted to observe seasonal groundwater quality due to pumping and to evaluate hydrochemical characteristics with groundwater level variations in an agricultural area where groundwater is used intensively at specific times. Particularly, rice paddies in Korea require a lot of water in summer, and if it rains less in that time, groundwater must be used more for irrigation. Therefore, we wanted to evaluate the groundwater quality in different seasons based on the long-term monitoring for groundwater levels and hydrochemical analysis.

2. Study Area

The study area is located about 4 km from the west coast in the mid-western region of the Korean Peninsula at longitude 126°29′30″–126°31′10″ E and latitude 36°32′43″–36°33′45″ N; it is 2.83 km² (283.3 ha) of land that includes Yanggokri and a part of Sinri in Hongseong County, Chungnam Province. A regional water supply system has been established from the water source several kilometers away, so there is little problem with the supply for domestic use. However, agricultural water is not enough, particularly under drought conditions. Although a small reservoir is nearby, it is hard to get irrigation water during severe drought and groundwater has to be supplied or drawn from other regions.

The area consists of hilly mountains with an altitude of fewer than 100 m and relatively flat farming areas. The geology consists of Precambrian schist, gneiss, and Quaternary sediments [15,16]. The upper layer is an alluvial layer mainly composed of porous media, and the lower layer is composed of bedrock. Therefore, the upper porous layer constitutes an alluvial aquifer, and the lower fractured bedrock constitutes a bedrock aquifer.

For agricultural land use, rice farming areas (paddy fields) account for 17.5% with 49.5 ha, and upland farming areas (upland fields) account for 13.3% with 37.7 ha. The area classified as forest or grassland is about 189.0 ha, accounting for 66.7% of the total area. In upland farming areas, various crops such as red pepper, sesame, soybean, pumpkin, and green onion have been cultivated (Figure 1).

As of December 2018, the number of people living in the area was estimated to be about 120 based on the Annual Statistics of Hongseong County (http://www.hongseong.go.kr/). The total water supply rate of the whole county is 92.5%, the daily water supply is 255 L/day/person on average, and the amount of water supply of the area is a little more than the average. There are 107 wells in the area, 71 of which are for agricultural use and 36 for domestic use, based on data provided by the county administration. Annual groundwater use is reported to be 1846 m³/day for agricultural use and 274 m³/day for domestic use, indicating that most of the water is used for agriculture. However, it is difficult to trust these figures of groundwater use, because it is not actually monitored. Moreover, since seasonal use is unknown, the groundwater quality associated with pumping cannot be evaluated.

In 2018, due to the prolonged summer drought after the rainy season, irrigation water became scarce for rice farming. The local government urgently decided to supply water for irrigation from Hongseongho Lake. However, it is an artificial lake in the coastal area, storing water from a stream that flows into the sea, which is located about 4 km from the study area. At that time, Hongseongho Lake water contained a lot of salt, so the supplied water quality had to be checked regularly to make sure it did not exceed the salinity management criterion (<1280 ppm). From 8 August, Hongseongho Lake was supplied through a waterway. On 13 August, the salinity of the supplied water at the site was checked as 1320 ppm, and the water supply was stopped. This situation was repeated. Fortunately, at the end of August, about 200 mm of precipitation fell, and the drought situation was over.

3. Materials and Methods

3.1. Monitoring and Estimation of Seasonal Groundwater Use

To evaluate seasonal changes in groundwater quality, 55 wells were identified on-site in the study area, and 25 wells were finally selected for investigation.

Table 1. Investigation wells.

Well ID	X	Y	Elevation (m.a.s.l.)	* Usage	** Main Aquifer	Well Depth (m)	Pump Permit (m3/day)	Automatic Monitoring	Hydrochemical Analysis
B-20	911,634	1,839,600	29.8	A	B	30	50	×	○
B-11	910,662	1,839,400	29.5	A	B	75	50	×	○
C-01	911,567	1,839,850	20.0	M	Al	13	-	○	○
C-02	911,458	1,839,710	15.4	M	Al	23	-	○	○
C-03	911,436	1,839,610	15.0	M	Al	15	-	○	○
C-04	911,399	1,839,510	11.6	M	Al	16	-	○	○
C-05	911,472	1,839,500	15.0	M	Al	18	-	○	○
C-06	911,345	1,839,370	10.6	M	B + Al	52	-	○	○
C-07	911,144	1,839,330	10.0	M	B + Al	52	-	○	○
C-08	910,503	1,839,750	25.3	M	Al	20	-	○	○
C-09	910,809	1,839,540	19.2	M	Al	18	-	○	○
D-01	910,919	1,839,530	14.2	M	B	151	-	○	○
D-02	910,807	1,839,480	20.0	M	B	151	-	○	○
D-03	911,739	1,840,220	35.6	A	B	335	200	○	○
D-04	911,468	1,840,030	31.3	A	B	40	30	×	○
D-05	911,522	1,839,830	20.0	A	B	90	90	○	○
D-06	911,664	1,839,800	25.5	A	B	90	140	×	○
D-07	911,831	1,839,950	36.8	A	B	175	90	×	○
D-08	910,855	1,839,579	15.0	A	B	103	220	○	○
D-09	911,029	1,839,440	10.0	A	B	50	60	○	○
D-10	911,080	1,839,380	10.3	A	B	100	135	○	○
D-11	910,833	1,839,429	22.1	M	B	>10	-	○	×
D-12	910,997	1,839,442	10.3	A	B	>10	-	○	×
D-13	911,635	1,840,104	25.0	M	Al	>10	-	○	×
D-14	911,555	1,840,439	34.3	M	Al	>10	-	○	×

* A, agricultural pumping well; M, monitoring well. ** Al, alluvial aquifer; B, Bedrock aquifer. ○ and × represent the monitoring or chemical analysis made and not, respectively.

shows the coordinates, elevation, usage, depth, and pump permit of the investigation wells, and automatic monitoring and hydrochemical analysis wells are marked. It is difficult to determine the exact structure of each well, but it is assumed that the screen of each well was installed according to main aquifer as shown in Table 1.

Groundwater level fluctuations are mainly caused by factors such as precipitation, groundwater pumping, tides, and so on. In the study area, long-term observations of groundwater levels are necessary because pumping for irrigation frequently happens. Automatic water level and quality monitoring sensors (TD-Diver, Van Essen Instruments, Delft, The Netherlands) were installed at 20 monitoring wells to assess the groundwater level fluctuations and pumping influence around the wells. Groundwater level monitoring was done automatically at 10 min intervals for approximately one year from September 2018 to September 2019. The automatic monitoring data were corrected by manually measured groundwater levels once a month.

The amount of groundwater use was estimated from the groundwater level data recorded by the automatic water level sensors. During a single pumping event, the pumping rate was checked, and it was multiplied by the pumping duration based on the groundwater level data. Estimated groundwater use is not exact because of the varying pumping rate during pumping and the monitoring interval, which is 10 min in this study. However, these errors could be ignored because groundwater uses in different seasons, and a long-term context must be evaluated rather than individual pumping events.

3.2. Hydrochemical Analysis

In order to evaluate the hydrochemical characteristics and seasonal changes of the groundwater in the study area, groundwater sampling from 21 wells in two different seasons, summer (17 July 2019) and fall (1 October 2019) was conducted, and measurement of groundwater levels was conducted at the same time to understand their spatial distribution in each time. The sampled groundwater was transported to the laboratory in a box with ice. Field parameters for groundwater samples, including temperature (T), pH, dissolved oxygen (DO), electrical conductivity (EC), and redox potential (Eh), were measured in the field using a digital meter for Intelligent Digital Sensors (IDS) (Multi3620 IDS, Xylem Analytics GmbH, Weilheim, Germany). The collected samples were filtered through a 0.2 μm membrane filter for instrumental analysis.

Major cations for Ca²⁺, Mg²⁺, Na⁺, K⁺, Fe, Mn, and SiO₂ (aq) were measured using inductively coupled plasma-optical emission spectrometry (Optima 7300 DV ICP-OES spectrometer; Perkin Elmer, Shelton, CT, USA). Anions for F⁻, Cl⁻, Br⁻, NO₂⁻, NO₃⁻, PO₄³⁻, and SO₄²⁻ were measured using ion-exchange chromatography with conductivity detection (ICS-1500; Dionex, Sunnyvale, CA, USA). Alkalinity was evaluated using the Gran titration method to determine the HCO₃⁻ concentration. The charge balance errors of all samples were less than 5–6%. Water-stable isotopes were analyzed by wavelength-scanned cavity ring-down spectroscopy (L1102-i; Picarro, Sunnyvale, CA, USA) with typical precision of 0.1‰ and 0.5‰ for δ¹⁸O and δ²H, respectively [17].

3.3. Multivariate Statistical Analysis

Principal component analysis (PCA) was used to identify major geochemical principal components (PCs) controlling groundwater chemistry. The PCs were extracted from 14 hydrochemical variables, including field parameters and dissolved components in the groundwater sample (T, pH, DO, EC, Ca²⁺, Mg²⁺, Na⁺, K⁺, SiO₂ (aq), HCO₃⁻, Cl⁻, NO₃⁻, SO₄²⁻, Br⁻) for 21 water samples in two different seasons. Some parameters (Fe, Mn, Sr²⁺, F⁻, NO₂⁻, PO₄³⁻) were excluded from PCA due to low concentration, and TDS and ORP (Oxidation Redox Potential) were excluded because EC and DO, respectively, can act as surrogates for them. Statistical analysis was conducted using the XLSTAT program by Addinsoft (version 2020.5, Addinsoft, Long Island, NY, USA).

4. Results and Discussion

4.1. Groundwater Level Monitoring and Groundwater Use

4.1.1. Groundwater Level Fluctuations

About one year of groundwater level fluctuations at the investigation wells in the study area, from September 2018 to September 2019, are shown in Figure 2. The graphs in Figure 2 show groundwater level fluctuations of the main pumping wells and their surrounding observation wells by regions. Figure 2a shows the pumping wells are D-08 and D-11, and D-09 and D-11 in Figure 2b. In Figure 2c, there is no pumping well, and in Figure 2d, D-03 and D-05 are pumping wells. Except for the pumping wells, the wells shown in Figure 2 are observation wells. Precipitation is also shown in Figure 2c. Although the recovery of groundwater levels by precipitation was observed locally at some wells, fluctuations of levels showed a significant effect due to pumping.

Figure 2a shows fluctuations in the groundwater levels of D-08 and D-11 pumping wells and D-01, D-02, and C-09 observation wells. D-08, D-11, D-01, and D-02 wells are installed in the bedrock aquifer, and C-09 is installed in the alluvial aquifer. D-08 is the main supply well for irrigation in the area. Based on the groundwater level fluctuation data, pumping increased from mid-April and continued for about two months until mid-July. From 22 May to 16 July, the groundwater level drawdown was lower than the depth of the sensor installed in the well, and groundwater levels were recorded at the sensor location, about –6.0 m. Therefore, it is believed that the water levels actually dropped to a lower level than this. The effect of pumping from D-08 also led to a drop in groundwater level at nearby observation wells D-01 and D-02. At D-01, the level continuously lowered from 12.62 m on 12 May to 0.15 m on 7 June, showing a 12.47 m drop. At D-02, the level went from 12.69 m to 4.20 m at the same time as D-01, showing an 8.49 m drop. Based on the groundwater level data, D-01 and D-02 were affected by pumping from D-08, and they seemed to be closely related to each other. Groundwater levels from another pumping well, D-11, showed rapid drop and recovery due to pump and stop. Here, groundwater level data below the sensor installation depth, about −5.0 m, were missing, but a correlation of levels at D-11, D-01, and D-02 were not evident. C-09 is an observation well installed in the alluvial aquifer, unlike the other wells in the bedrock aquifer. The groundwater levels at C-09 fluctuated between 13.68 and 15.60 m in mean sea level elevation, showing a 1.2 m drop from 14.9 m on 12 May to 13.7 m on 7 June. C-09 is the closest observation well to D-08, the main pumping well in Figure 3a, but showed a different fluctuation pattern compared to D-01 and D-02, indicating poor connectivity between bedrock aquifer and alluvial aquifer.

It is observed that groundwater levels rose as of 8 June despite continuous pumping from mid-April to early August, as shown in Figure 2a. Even though there was only a little rain during that time, as seen by the precipitation data shown in Figure 2c, the rise in groundwater levels must be attributed to other factors. The most likely factor is the recharge of the irrigation water in rice paddies. Water supplied to rice paddies from the pumping goes back into the ground and recharges the groundwater in the aquifer with a time delay. The groundwater level increase as of mid-June can be seen in Figure 2a and the other graphs in Figure 2. Moreover, the groundwater level rise is observed in wells installed in both the alluvial aquifer and bedrock aquifer. The effect of recharge on paddy fields into aquifers can be found in various studies [18,19,20].

Figure 2b shows the groundwater level fluctuations at D-09 and D-10 pumping wells and D-12 and C-07 observation wells. Rapid groundwater drawdown and recovery at D-09 and D-10 are shown, which was caused by the pump and stop. Groundwater levels below the sensor installation depth of −21.75 m at D-09 and −12.33 m at D-10 were not observed. A groundwater level drop of about 33.2 m occurred at D-09, and 22.64 m at D-10. Groundwater pumping frequently occurred from mid-April to late June, and the levels at D-10 were lowered to about −22.0 m from 9 May to 19 June. Due to pumping from D-09 and D-10, the groundwater levels at D-12 went down to about 1.5 m from May to mid-June. D-well is installed in bedrock and D-12 in the alluvial aquifer, but the groundwater level fluctuations between the two are correlated. Similarly, the levels at C-07 in the alluvial aquifer were from 5.64 to 7.98 m, and the effect of pumping from D-10 was evident at the well. Groundwater levels at D-10 were lowered by about 1.68 m from early May to mid-June.

Figure 2c shows the groundwater level fluctuations at C-02, C-03, C-04, C-05, C-06, and C-08 observation wells. The effects of both precipitation and pumping from nearby wells on these wells are illustrated. C-08 among the wells in Figure 2c is located at the highest altitude and the groundwater levels are the highest. The levels at C-05 reflected the effect of nearby groundwater pumping. The distance between C-05 and the nearby pumping well is about 25 m, and they are thought to be connected to each other. The groundwater level fluctuations at C-04 were similar to those at C-05, reflecting the pumping effect.

Figure 2d shows the groundwater level fluctuations at D-03, D-05, and D-13 pumping wells and D-14 and C-01 observation wells. Pumping from D-03 for irrigation of paddy fields was very often from May to July, and groundwater was also pumped from November to March, which was mainly for upland field irrigation. Groundwater levels were not recorded below the sensor installation depth, and the maximum drawdown of levels at D-03 is estimated to be over 54.2 m. The groundwater level at D-05 was affected by pumping from 22 May to 16 June, and was constant at about 21.3 m. At C-01, the groundwater level dropped by about 2 m due to pumping from D-05, and was at about 18.3 m. The level at D-14 was the highest at about 36.2 to 36.5 m with little fluctuation. The levels at D-13 fluctuated due to the individual pumping.

4.1.2. Spatial Distribution and Factors for Groundwater Level Changes

Figure 3 shows the spatial distribution of groundwater levels in two seasons, the first survey in July 2019 and the second in October 2019. The levels were only described in plain areas because there are no observation wells in forest areas. The depths of groundwater levels were 0.69 to 16.27 m (average 5.07 m) below the land surface in July, and 0.10 to 7.91 m (average 2.90 m) in October. The groundwater levels in mean sea level elevation were 2.38 to 32.75 m (average 15.18 m) in July and 7.02 to 32.80 m (average 15.71 m) in October. The groundwater levels in October were about 53 cm higher, on average, than in July. It is certain that the pumping influence in the eastern part of the area was reduced in October. In the southern part, the groundwater levels in October were higher than in July. Although the levels around D-09 and D-10 were still down in October, there was a slight rise by about 1.2 m compared with those in July.

Groundwater use in the area is closely related to irrigation for rice farming. Due to the large supply of irrigation water for rice paddies in summer, groundwater levels decrease by pumping in July. However, irrigation water in October, when rice is harvested, is rarely required. Precipitation is another factor that affects changes in groundwater levels in the area. Precipitation, which was 37.5 mm in July and 112 mm in September (the month prior to the second survey on 1 October 2019), may also have caused groundwater levels to increase. Additionally, irrigation water for rice paddies can flow back into the ground, raising the groundwater levels. Monitoring data also suggest that the irrigation water in rice paddies entered the aquifer to increase groundwater levels.

4.2. Monthly Groundwater Use

Monthly groundwater use was estimated based on the fluctuation data at D-03, D-05, D-08, D-09, and D-10 wells, which are major agricultural wells in the area. The groundwater use for a certain period was calculated by multiplying the pumping rate by pumping duration. The pumping rate of D-03, D-05, D-08, D-09, and D-10 wells for a single pumping event is 242, 90, 248, 95, and 174 m³/day, respectively.

Table 2. Monthly groundwater use of main pumping wells in the study area. (unit: m³).

Date	D-03	D-05	D-08	D-09	D-10
18 October	54	-	-	-	-
18 November	118	-	-	-	23
18 December	77	-	3	1	364
19 January	469	-	-	2	172
19 February	924	-	-	1	40
19 March	597	-	2	11	28
19 April	324	-	854	395	41
19 May	8648	689	5322	460	5351
19 June	1294	629	7440	614	580
19 July	704	603	5042	141	225
19 August	2035	-	5998	486	1613
19 September	328	-	1755	114	53
19 October	2	-	-	-	-
Sum	15574	1921	26417	2223	8490

- indicates no pumping recorded.

shows the estimation results of monthly groundwater use. The total amount of groundwater of these five wells was 54625 m³/year. D-08 pumped the most groundwater for one year, 26417 m³. The maximum monthly groundwater use was estimated at D-03, 8648 m³ in May 2019. In summer 2019 (June to August), the amount of precipitation was 172 mm, which was only about 27% of the normal amount of the season. Therefore, groundwater use during that time could have been large compared with other years.

There are a total of 25 wells for agricultural use in the area and the pump permit of those wells is 1918 m³/day, which is 700,070 m³/year. However, the actual amount used for one year is reported as 28,414 m³, so this amount is only about 4% of the pump permit. In this study, 54,625 m³ from five wells (D-03, D-05, D-08, D-09, and D-10) was used over one year, from October 2018 to September 2019, equivalent to 192% (almost twice as much) of the total reported groundwater use. The estimated amount of groundwater use of the five wells is around 21.2% of the pump permit, or 705 m³/day (257,325 m³/year). It can be deduced that actual use was quite different from the reported amount, and only a small portion of the pump permit was used.

4.3. Hydrochemical Analysis and Groundwater Quality Evaluation

4.3.1. Groundwater Quality Changes and Water Types

The main causes of water quality change in the study area are various such as precipitation, groundwater pumping, and pollution from agricultural activities. In particular, when the drought in the summer of 2018 was severe, the temporary supply of salty water from Hongseongho Lake as irrigation water is also thought to affect water quality changes.

The hydrochemical analysis was properly conducted with a charge balance error (CBE) less than 5.5%, and the results are shown in

Table 3. Hydrochemical analysis results of groundwater.

Survey	Well	Temp	pH	DO	Eh	EC	TDS	Ca2+	Mg2+	Na+	K+	Fe	Mn	SiO2 (aq)	Sr2+	HCO3−	F−	Cl−	Br−	NO2−	NO3−	PO43−	SO42−	CBE
		°C		mg/L	mV	μS/cm	mg/L	mg/L	mg/L	mg/L	mg/L	mg/L	mg/L	mg/L	mg/L	mg/L	mg/L	mg/L	mg/L	mg/L	mg/L	mg/L	mg/L	(%)
17 July 2019	B-11	16.1	6.48	5.05	179	309	201	26.9	12.4	11.1	2.5	ND	ND	22.1	0.17	99	0.1	22.2	0.1	ND	41.1	ND	13.9	(4.8)
	B-20	16.1	7.70	6.89	210	367	234	38.1	12.5	14.0	1.9	ND	ND	19.7	0.29	141	0.1	20.2	0.1	ND	48.1	0.1	9.7	(3.6)
	C-01	20.5	6.74	3.73	94	595	319	65.2	13.9	24.2	2.6	0.1	1.5	14.2	0.65	194	0.1	57.1	0.2	ND	0.2	ND	45.4	(2.0)
	C-02	19.9	6.57	1.90	−17	728	381	70.8	17.8	26.1	7.4	8.6	4.6	17.7	0.44	271	0.1	60.7	0.2	ND	0.1	ND	38.4	(2.5)
	C-03	21.7	6.79	4.92	151	2630	1409	103.9	47.1	320.6	27.6	ND	0.3	6.2	0.72	175	1.4	735.5	2.1	ND	2.6	ND	77.7	(3.3)
	C-04	19.1	7.30	1.70	−83	797	440	92.3	19.6	31.2	5.5	2.5	1.8	15.9	1.30	399	0.1	70.7	0.3	ND	0.0	ND	5.0	(5.1)
	C-05	20.1	7.27	5.48	197	450	324	55.1	15.6	23.9	5.0	ND	ND	20.6	0.53	185	0.1	56.0	0.1	ND	28.0	ND	29.0	(4.2)
	C-06	19.2	7.10	4.70	−45	654	430	86.2	19.0	37.1	5.7	1.3	2.1	15.5	1.33	268	0.1	106.4	0.3	ND	0.2	ND	26.4	(1.8)
	C-07	18.20	6.97	0.66	102	1008	600	127.8	21.8	54.5	7.2	ND	ND	10.0	0.72	291	ND	179.6	ND	ND	5.2	ND	50.6	(1.1)
	C-08	17.6	6.33	6.19	213	569	324	58.7	16.3	15.5	3.4	ND	ND	18.6	0.28	66	0.1	97.0	0.3	ND	60.5	ND	21.0	(2.0)
	C-09	16.6	6.17	1.65	177	1081	549	89.0	26.3	60.7	6.8	ND	ND	10.7	ND	107	ND	279.5	0.8	ND	2.8	ND	20.4	(3.4)
	D-01	16.3	9.34	0.11	84	160	81	6.9	2.9	17.4	4.4	ND	ND	0.1	ND	50	0.2	21.9	0.1	ND	ND	ND	2.2	(1.2)
	D-02	15.0	8.11	2.60	−32	442	195	33.3	11.7	17.6	2.6	ND	0.2	3.5	0.20	132	0.1	51.6	0.2	ND	0.1	ND	9.6	(5.1)
	D-03	16.1	7.65	7.34	266	219	141	26.3	7.7	7.0	1.3	ND	ND	15.7	0.22	125	0.1	9.9	0.0	ND	5.2	ND	6.7	(5.5)
	D-04	14.0	6.68	9.81	327	290	188	29.0	7.3	15.4	2.5	ND	ND	15.1	0.16	97	0.1	21.1	0.0	ND	30.6	ND	19.5	(5.0)
	D-05	15.7	7.93	0.04	148	353	226	47.7	4.9	20.0	1.3	ND	0.2	25.3	1.30	166	0.0	28.9	0.1	ND	ND	ND	16.2	(2.4)
	D-06	16.5	7.34	7.60	196	164	87	12.9	5.3	6.2	0.8	ND	ND	14.8	0.15	59	0.1	8.9	0.1	ND	6.2	ND	3.2	(0.3)
	D-07	16.0	8.24	8.74	237	227	139	19.1	7.6	17.8	0.9	ND	ND	16.9	2.29	117	0.3	7.3	0.0	0.026	0.3	ND	11.2	0.1
	D-08	15.2	6.65	1.82	157	260	162	29.3	8.6	9.3	1.4	ND	ND	19.5	0.15	122	0.1	23.5	0.1	ND	6.1	ND	4.4	(4.4)
	D-09	15.7	6.82	0.08	111	590	329	66.3	19.1	15.4	2.5	ND	0.7	23.3	0.43	175	0.0	102.0	0.4	ND	ND	ND	14.5	(3.7)
	D-10	15.9	7.07	0.05	211	659	363	69.8	17.7	30.1	3.0	ND	0.3	7.6	0.98	196	0.1	107.0	0.4	ND	ND	ND	22.8	(3.0)
1 October 2019	B-11	16.1	6.28	5.26	224	335	214	29.7	12.0	12.8	6.1	ND	ND	20.0	0.15	75	0.1	21.9	0.1	ND	55.2	0.1	20.0	0.5
	B-20	15.7	7.58	6.40	229	388	245	43.0	13.5	14.4	2.0	ND	ND	19.2	0.39	128	0.0	19.8	0.1	ND	55.3	0.1	15.2	0.9
	C-01	20.9	6.63	3.93	109	409	265	56.0	11.1	23.8	3.1	ND	0.9	12.4	0.50	180	0.1	48.3	0.2	ND	0.6	ND	20.7	0.7
	C-02	21.2	6.39	1.50	−17	601	357	69.3	16.4	24.6	5.3	3.2	4.9	20.5	0.40	231	0.1	58.6	0.2	ND	0.1	ND	44.8	(2.0)
	C-03	20.1	6.78	2.33	−113	1361	709	73.8	25.6	136.6	14.0	3.3	2.9	11.0	0.50	266	1.0	303.3	0.9	ND	0.2	ND	9.8	(3.8)
	C-04	19.2	7.30	3.33	−54	648	420	91.6	19.0	34.6	6.1	0.1	1.7	16.4	1.23	326	0.1	85.2	0.3	ND	0.2	ND	6.4	(0.6)
	C-05	20.5	7.06	5.79	174	281	186	35.5	9.5	9.5	5.0	ND	ND	15.7	0.27	129	0.3	17.3	0.1	ND	17.3	ND	12.7	(1.2)
	C-06	20.0	7.30	3.63	223	597	390	75.3	17.5	37.5	4.6	0.7	2.1	16.0	1.58	248	0.1	102.5	0.3	ND	0.4	ND	13.2	(1.9)
	C-07	21.2	6.67	2.60	189	639	395	78.1	13.8	41.5	5.4	ND	1.7	9.2	0.70	224	0.1	87.3	0.2	ND	0.7	ND	48.6	(1.3)
	C-08	17.5	6.19	7.10	220	243	180	27.6	8.9	11.3	2.0	ND	ND	17.9	0.14	33	0.1	50.8	0.2	0.0	40.7	0.1	4.9	(1.5)
	C-09	20.1	6.29	4.05	203	898	529	83.1	25.6	60.8	6.0	ND	0.1	13.9	0.40	91	0.0	266.5	0.8	0.2	3.8	ND	24.8	(2.9)
	D-01	21.3	9.26	3.59	94	153	78	6.0	1.6	17.6	4.1	ND	ND	ND	0.11	42	0.2	19.1	0.1	ND	0.1	ND	8.5	(4.0)
	D-02	15.8	7.82	2.22	−320	336	182	31.5	11.6	17.3	2.6	ND	0.2	2.7	0.27	109	0.1	52.7	0.2	ND	0.1	ND	9.9	(2.0)
	D-03	15.3	7.59	4.40	228	228	139	26.7	8.0	7.1	1.3	ND	ND	15.4	0.22	110	0.0	10.0	0.1	ND	7.6	ND	9.1	(1.5)
	D-04	16.5	6.64	8.07	247	313	197	32.2	7.7	16.1	2.6	ND	ND	15.2	0.17	80	0.1	24.9	0.1	ND	36.5	ND	22.8	(1.2)
	D-05	18.9	7.85	3.50	225	352	216	46.4	4.7	19.7	1.2	ND	ND	24.4	1.28	148	0.1	29.6	0.1	ND	0.2	ND	16.9	(0.3)
	D-06	14.9	6.90	7.35	229	160	100	12.9	6.3	7.3	1.1	ND	ND	17.3	0.12	59	0.1	11.4	0.1	ND	10.3	ND	4.6	(1.7)
	D-07	15.5	8.08	3.21	217	222.5	138	17.9	7.8	17.1	0.9	ND	0.1	16.6	2.41	117	0.4	7.3	0.0	ND	ND	ND	12.3	(2.1)
	D-08	15.2	6.54	1.25	95	238	145	25.9	8.3	9.2	1.2	ND	0.1	18.8	0.12	85	0.2	29.8	ND	ND	4.6	ND	5.0	(0.3)
	D-09	15.7	7.68	2.60	201	289	168	33.1	10.6	8.2	1.6	ND	ND	15.9	0.38	121	0.1	27.6	0.1	ND	4.2	ND	7.0	(0.9)
	D-10	16.2	7.64	3.29	180	332	188	28.0	6.5	24.4	2.2	ND	ND	12.0	0.82	119	0.3	29.5	0.1	ND	5.6	ND	20.7	(4.1)

ND, not detected; CBE, charge balance error; (·) in CBE, negative value.

for the first survey on 17 July 2019 and the second on 1 October 2019. Total dissolved solids (TDS) amounted to 81–1409 mg/L (average 339 mg/L) in July and 78–709 mg/L (average 259 mg/L) in October, showing slightly lower TDS in groundwater in October than in July. Concentrations of major ions, both cations and anions, were also lower in October than in July. In addition, electrical conductivity (EC) was measured between 160–2630 μS/cm (average 598 μS/cm) in July and 153–1361 μS/cm (average 430 μS/cm) in October.

Table 4. Change of water types for groundwater samples in the study area.

Change of Water Type		Groundwater Wells
19 July	19 October
CaCl (I)	CaCl (I)	C-08, C-09
CaHCO3 (II)	CaHCO3 (II)	B-20, C-01, C-02, C-04, C-05, C-06, D-02, D-03, D-05, D-06, D-07, D-08
CaHCO3 (II)	Mixed CaMgCl (III)	B-11, D-04
Mixed CaMgCl (III)	CaHCO3 (II)	C-07, D-09, D-10
Mixed CaNaHCO3 (IV)	NaCl (VI)	D-01
NaCl (VI)	NaCl (VI)	C-03

and Figure 4 show groundwater quality changes in the different seasons based on piper diagrams. More than 50% of the samples were classified as CaHCO₃ (II) water type, and water type changes occurred depending on the well location. Water quality changes were small at most wells, but changes at some wells were evident. Comparing the groundwater quality in July and October, the water type at B-11 and D-04 was changed from CaHCO₃ (II) to mixed CaMgCl (III); at C-07, D-09, and D-10 from mixed CaMgCl (III) to CaHCO₃ (II); and at D-01 from mixed CaNaHCO₃ (IV) to NaCl (IV).

It seems certain that the groundwater in October was more diluted than in July. Such changes in groundwater quality may have been caused by precipitation and recharge of surface water, such as irrigation water of rice paddies. When the groundwater level is lowered by pumping, irrigation water may flow more into aquifers. Finally, intensive pumping in the area may have induced the irrigation water into aquifers, changing the groundwater quality. The flow of irrigation water into aquifers was reflected in the groundwater level rise despite continuous pumping from mid-June to early August.

4.3.2. Evaluation of Characteristics of Groundwater Quality

In order to see whether there were any seasonal changes in groundwater quality, the wells were categorized into two groups, group C and group B/D, according to the first letter of the well label, and box plots were constructed, shown in Figure 5. Group C wells are all installed at shallow depths in the alluvial aquifer, except for C-06 and C-07, and group B/D wells are installed deep in the bedrock. Although C-06 and C-07 are installed at about 50 m depth, it is reasonable to assume that they are affected by the alluvial layer because well screens penetrate both the upper alluvial aquifer and the lower bedrock.

Most of the hydrochemical parameter values of group C are higher (temperature, TDS, EC, Ca²⁺, Mg²⁺, Na⁺, K⁺, HCO₃⁻, Cl⁻, Br⁻, and SO₄²⁻) than those of the relatively deep wells of group D/B. pH, Eh, and SiO₂ (aq) values are relatively higher in group D/B than group C. DO and NO₃⁻ in group B/D show large variations. NO₃⁻ concentration, as the indicator of contamination from agricultural activity or livestock waste, is lower than the drinking water quality standards of Korea (NO₃–N: 10 mg/L; NO₃⁻: 44.3 mg/L), but the groundwater samples of C-08, B-20, and B-11 in both the first and second survey exceeded the standards. The NO₃⁻ concentration at B-20 and B-11 in the first survey was 48.1 mg/L and 41.1 mg/L, and in the second survey was 55.3 mg/L and 55.2 mg/L, respectively. They showed a higher concentration in October than in July. On the other hand, the groundwater sample at C-08 showed 60.5 mg/L in the first survey and 40.7 mg/L in the second survey, and the NO₃⁻ concentration was reduced from July to October.

The seasonal variation of groundwater quality in group C was very clearly compared with group B/D. In particular, the parameters indicating dissolved components in groundwater (TDS, EC, Ca²⁺, Mg²⁺, Na⁺, K⁺, HCO₃⁻, Cl⁻, Br⁻, and SO₄²⁻) were lower in October than in July. It may have been caused by precipitation or surface water infiltration, and the irrigated water used for rice paddies may be an important factor in reducing dissolved components in groundwater in the area. However, group B/D does not have large seasonal variation compared with group C. The main difference between the groups is whether the main aquifer for well development is in the upper alluvium (soil) or the lower bedrock. Group C wells are installed in the alluvial layer and are highly affected by agricultural activity, and it seems that a lot of matter in the soil might be dissolved in the groundwater. In addition, group C wells are developed at the upper layer, where the inflow of rainfall and surface water can occur more easily than at group B/D wells, and then the dilution effect can be more clearly seen.

Figure 6 shows the relationships between Cl⁻ and Na⁺, and Cl⁻ and Br⁻ in each well to evaluate the characteristics and sources of groundwater. The seawater data were extracted from the literature [21]. The groundwater samples are clearly lined up with groups C and B/D to the seawater as the endpoint in the graph of Cl⁻ versus Na⁺ and Br⁻ concentration. Group C wells are lined up between group B/D and seawater. This suggests that the groundwater quality is affected by saltwater more in group C than in group B/D. As such, the possible seawater effects on groundwater occur due to the fact that the saline water was supplied from Hongseongho Lake during the summer drought in 2018. The effects are shown particularly in alluvial groundwater at shallow depths. The group B/D samples were farther from the seawater than group C samples, and the effect of seawater in deep wells seems to be relatively small. Comparing the groundwater quality in July and October, the groundwaters tend to move farther away from seawater, especially in Cl⁻ and Na⁺ in October. Over time, the influence of seawater seems to have decreased.

Figure 7 shows the analysis results of oxygen and hydrogen isotopes to identify the source of groundwater. The water samples were from the first survey, and no analysis was conducted for the second survey samples. The groundwater samples were close to the local meteoric water line (LMWL) [22] during the summer, the rainy season, and the groundwater seems to have been mostly recharged in summer. The samples of group C and some group B/D wells reflect the evaporation effect away from the LMWL. Particularly, the evaporation effect was the largest at C-03. Samples showing the evaporation effect were from shallow depths, 0.74–4.93 m from the surface, and had relatively high temperatures (15.9–21.7 °C). The presence of evaporation effects at shallow depths means that external factors may easily cause groundwater quality to change. The wells in the elliptical region shown in Figure 7 have a large seasonal change in groundwater quality.

Table 5. Correlation matrix of hydrochemical parameters of groundwater samples.

Variables	T	pH	DO	EC	Ca2+	Mg2+	Na+	K+	SiO2 (aq)	HCO3−	Cl−	NO3−	SO42−	Br−
T	1.000
pH	–0.160	1.000
DO	0.280	–0.086	1.000
EC	0.492	–0.329	0.766	1.000
Ca2+	0.494	–0.520	0.286	0.712	1.000
Mg2+	0.461	–0.483	0.643	0.916	0.815	1.000
Na+	0.427	–0.201	0.895	0.931	0.520	0.818	1.000
K+	0.552	–0.224	0.817	0.927	0.585	0.845	0.951	1.000
SiO2 (aq)	–0.275	–0.411	–0.219	–0.282	0.008	–0.112	–0.308	–0.330	1.000
HCO3-	0.429	–0.266	0.037	0.415	0.770	0.484	0.245	0.340	0.097	1.000
Cl−	0.416	–0.311	0.816	0.950	0.629	0.902	0.959	0.920	–0.291	0.232	1.000
NO3−	–0.220	–0.129	–0.063	–0.198	–0.233	–0.102	–0.188	–0.132	0.340	–0.388	–0.184	1.000
SO42−	0.516	–0.416	0.591	0.665	0.632	0.633	0.626	0.648	–0.109	0.337	0.621	–0.087	1.000
Br−	0.411	–0.329	0.809	0.909	0.553	0.880	0.931	0.891	–0.250	0.195	0.969	–0.182	0.537	1.000

shows a correlation matrix of the 14 hydrochemical parameters in groundwater samples. EC shows a high correlation with major dissolved ions such as Ca²⁺, Mg²⁺, Na⁺, K⁺, Cl⁻, SO₄²⁻, etc., and DO is highly correlated to EC, Na⁺, K⁺, Cl⁻, and Br⁻. In particular, Br⁻, Cl⁻, Mg²⁺, and Na⁺, which indicate an association with seawater, are highly correlated.

Table 6. Principal component loadings, eigenvalues, and explained variance for five extracted components of groundwater samples.

Variable	PC1	PC2	PC3	PC4	PC5
Eigenvalue	7.887	1.957	1.566	0.775	0.527
Variability (%)	56.338	13.980	11.187	5.536	3.762
Cumulative (%)	56.338	70.319	81.506	87.042	90.803
	Factor Loadings
T	0.575	–0.125	–0.389	0.594	–0.016
pH	–0.380	0.667	–0.392	0.024	0.409
DO	0.787	0.396	0.263	–0.051	0.010
EC	0.972	0.052	0.020	–0.064	0.067
Ca2+	0.752	–0.544	–0.195	–0.030	0.094
Mg2+	0.935	–0.170	0.099	–0.086	0.105
Na+	0.938	0.266	0.100	–0.087	0.040
K+	0.946	0.178	0.037	0.066	0.124
SiO2 (aq)	–0.257	–0.653	0.497	–0.199	0.106
HCO3−	0.447	–0.620	–0.502	–0.111	0.312
Cl−	0.956	0.174	0.124	–0.123	–0.022
NO3−	–0.227	–0.057	0.769	0.430	0.359
SO42−	0.743	–0.175	0.050	0.342	–0.280
Br−	0.923	0.188	0.157	–0.160	–0.037

and Figure 8 show the PCA results of groundwater samples from both surveys. PCA results for all samples, including the first and second surveys, show that the three major eigenvalues are 56.34% for PC1, 13.98% for PC2, and 11.19% for PC3, explaining 81.51% of the total variance. Among the 14 hydrochemical parameters, PC1 confirms that DO, EC, Ca²⁺, Mg²⁺, Na⁺, K⁺, Cl⁻, SO₄²⁻, and Br⁻ are the main contributors, while PC2 is affected by pH, SiO₂ (aq), and HCO₃^-, and PC3 is affected by NO₃⁻. PC1 may be considered as the amount of main dissolved components in groundwater (Table 6). Figure 8a shows PC1 versus PC2 of the samples. It clearly distinguished from groups C and B/D, and the groundwater quality of C-03 shows the dramatic change from July to October.

Table 7. Variation of principal components (PCs) during seasonal change. Symbols ▲ and ▼ indicate increase and decrease of PCs, respectively.

Group	Sample	PC1	PC2	PC3
C	C-01	▼ 0.691	▲ 0.487	▼ 0.089
	C-02	▼ 0.220	▼ 0.272	▲ 0.266
	C-03	▼ 9.792	▼ 3.490	▼ 2.785
	C-04	▼ 0.139	▲ 0.411	▲ 0.414
	C-05	▼ 1.034	▲ 0.917	▼ 0.471
	C-06	▼ 0.543	▲ 0.400	▼ 0.064
	C-07	▼ 1.182	▲ 0.786	▼ 0.034
	C-08	▼ 1.482	▲ 0.672	▼ 0.551
	C-09	▲ 0.015	▼ 0.201	▼ 0.165
B/D	B-20	▲ 0.138	▼ 0.095	▲ 0.386
	B-11	▲ 0.405	▲ 0.183	▲ 0.573
	D-01	▲ 0.493	▼ 0.209	▼ 0.610
	D-02	▲ 0.090	▼ 0.148	▲0.163
	D-03	▲ 0.027	▼ 0.004	▲ 0.435
	D-04	▲ 0.202	▲ 0.045	▼ 0.297
	D-05	▲ 0.363	▼ 0.386	▼ 0.075
	D-06	▼ 0.054	▼ 0.203	▲ 0.387
	D-07	▼ 0.206	▲ 0.162	▲ 0.192
	D-08	▼ 1.352	▲ 1.209	▼ 0.138
	D-09	▼ 1.959	▲ 1.807	▼ 0.155
	D-10	▼ 0.356	▲ 0.553	▼ 0.203

shows the variation of principal components (PCs) during seasonal changes. The water quality of most samples changed in the direction that the value of PC1 was reduced from July to October. A reduced PC1 value is thought to indicate dilution. In addition, the water quality of most samples, except the C-03 sample, changed upward, as shown in Figure 8a. The upward direction for the increased value of PC2 means that pH, HCO₃⁻, and SiO₂ (aq) concentrations increased from July to October, and this might be indicated by the degree of water–rock or water movement and mixing in the groundwater flow system. Figure 8b shows PC1 versus PC3 of the samples. PC3 values also tended to decrease as PC1 values decreased from July to October. The parameter representing PC3 is NO₃, and it decreased together with the reduced PC1.

According to PCA, the water quality from July to October can change due to the dilution, and the effect is strong in group C. The period between July and October is when there is much precipitation. Therefore, precipitation may recharge much into the aquifer, and the recharged water may dilute the groundwater. Additionally, a lot of groundwater is pumped for irrigating rice paddies, and the irrigation water can flow back into the aquifer and influence the groundwater to be diluted. In the study areas where rice farming is active in summer, groundwater quality changes seem to be largely caused by both precipitation and the irrigated water.

Figure 9 is presented as a conceptual diagram in order to easily understand the factors affecting the groundwater quality in the study area. From the PCA results, dilution (PC1), water–rock interaction (PC2), and contamination (PC3) were derived as the factors that cause water quality changes in July and October. Dilution can occur by precipitation, surface water infiltration, and the irrigated water to rice paddies which is induced by pumping or naturally flowing back into the aquifer. In addition, water–rock interaction occurs in the aquifer under groundwater flow control. Contaminations by fertilizer and livestock waste also affect groundwater quality in the study area. High saline groundwaters at some wells are attributed to the irrigated water from the outer water source during the past severe drought. From the study, we confirm that the changes in groundwater quality have occurred, and they are closely related to irrigation during the rice-growing season.

5. Conclusions

This study was conducted in priority to find out whether changes in water quality due to intensive pumping were observed in rice farming areas. Furthermore, if such a change in water quality was observed, it was attempted to determine the cause of the occurrence. Through long-term monitoring of the groundwater levels, it was confirmed that intensive pumping occurred for a certain period (from May to August) to supply irrigation water for rice farming. Groundwater levels fluctuated due to frequent pumping for irrigation of rice paddies in summer, and groundwater uses were estimated from major pumping wells in the area based on automatic groundwater level monitoring data from September 2018 to September 2019, about one year. Groundwater use in the area is closely related to irrigation for rice farming. Besides precipitation, irrigation water for rice paddies is another factor that increases groundwater levels, which is clearly shown in the rise in levels during pumping.

The groundwater quality analysis results showed a large change in water quality in some wells, but in some wells, the change was insignificant. Water quality types are classified using the major cations and anions dissolved in groundwater. It seems certain that the groundwater in October is more diluted than in July. The seasonal variation of groundwater quality in group C of alluvial wells at shallow depth was very clearly compared with those in group B/D deep wells. The parameter values indicating dissolved components in groundwater (TDS, EC, Ca²⁺, Mg²⁺, Na⁺, K⁺, HCO₃⁻, Cl⁻, Br⁻, SO₄²⁻) are lower in October than in July. Group B/D does not have large seasonal variation. In the isotope analysis results, it was found that the evaporation effect appeared in shallow wells and thus was greatly influenced by external factors.

The PCA results show that the three major eigenvalues were 56.34% for PC1, 13.98% for PC2, and 11.19% for PC3, explaining 81.51% of the total variance. PC1 can be related to the amount of main dissolved components in groundwater. The water quality of most samples changed in the direction of the PC1 value reduction. The PC2 may indicate the degree of water–rock or water–aquifer interaction, and the PC3, which is referred to NO₃⁻ concentration, may be related to contamination. Another factor influencing groundwater quality in the study area is the saline water irrigation from other regions, which was supplied temporarily during the summer drought in 2018. It was evaluated from the comparisons with the Na⁺, Cl⁻, and Br^- components of seawater and groundwater.

This study is meaningful because it quantitatively evaluated the change of groundwater quality in relation with irrigation water by groundwater pumping, precipitation, and other possible factors (saline water supply during the summer drought in the previous year in this study) in rice farming areas for the first time.