Field Study and Numerical Modeling to Assess the Impact of On-Site Septic Systems on Groundwater Quality of Jeju Island, South Korea

Abstract

Septic-derived nitrogen (N) sources have harmful effects on water resources, humans, and ecosystems in several countries. On Jeju Island, South Korea, the rapid increase in personal sewage treatment facilities (PSTFs, also known as on-site septic systems) raises concerns regarding the deterioration of groundwater quality, as groundwater is the sole water resource on the island. Therefore, this study employed a field study and numerical modeling to assess the impact of PSTF effluents on groundwater quality in the Jocheon area of northeastern Jeju. Water quality analysis revealed that the total nitrogen (T-N) concentrations in the effluent exceeded the effluent standards (75–92% PSTFs). The numerical model simulated the transport of N species, showing limited NH₄⁺ and NO₂⁻ plume migration near the surface due to nitrification and adsorption. However, NO₃⁻ concentrations increased and stabilized over time, leaching on the water table with higher levels in lowland areas and clustered PSTFs. The predictive model estimated a 79% reduction in NO₃⁻ leaching when the effluents followed standards, indicating the necessity of effective PSTF management. This study highlights the importance of managing improperly operated septic systems to mitigate groundwater contamination based on an understanding of the behavior of N species in subsurface hydrologic systems.

1. Introduction

Groundwater is a major water resource that provides 50% of drinking water, 40% of irrigation water, and 30% of industrial water worldwide [1]. However, increasing anthropogenic activities such as agriculture, livestock farming, and wastewater discharge degrade groundwater quality [2,3]. Among the numerous types of groundwater contamination, nitrate (NO₃⁻) contamination has been a significant issue in many countries [4,5,6]. Although the N component is an essential nutrient for plant growth, an excessive loading of N elevates the NO₃⁻ concentration in groundwater over a maximum criteria level (MCL) for drinking water of 50 mg/L of NO₃⁻ [7,8] or 10 mg/L of NO₃-N [9,10,11,12]. Therefore, various studies have been conducted to identify the source and transformation of NO₃⁻ based on the stable isotope analysis [6,13], reduce the NO₃⁻ level in groundwater via applying remediation methods [14,15], and predict the groundwater NO₃⁻ concentration for the water quality management by using numerical models [16,17].

Among the various N contaminants, sewage discharge from septic systems has been identified as a major contributor to groundwater NO₃⁻ contamination in many countries [18,19,20,21]. In highly populated coastal areas in Florida, USA, the clustered septic systems are considered as the principal N source with an annual N load of 2.4–2.9 × 10¹⁰ g N/year [18], which present a threat to public health and the ecosystem by polluting groundwater and surface water [22,23]. In Hawaii, USA, approximately 88,000 cesspools (on-site sewage disposal systems, OSDS) are widespread across the islands, directly discharging pathogens and nutrients into groundwater [20]. Lu et al. [21] discovered that poorly managed septic tanks caused an increase in NO₃⁻ levels in groundwater in the Pearl River Delta, China, which underwent rapid urbanization in the early 2000s. In Hungary, effects of the sewage effluent remained in soil and groundwater more than 10 years after closure with detection of high NO₃⁻ and PO₄³⁻ posing prolonged hazardous pollution due to septic effluents [24]. Moreover, contaminants from septic tanks, which are expected to support sewer treatment for 4.9 billion individuals by 2030, contribute to approximately 30% of the NO₃⁻ contamination in groundwater [25,26]. Hence, the effective management of septic systems to mitigate groundwater NO₃⁻ contamination is essential for understanding the behavior of N species in subsurface hydrologic systems.

Numerical modeling has been widely used to estimate the N load from septic systems and simulate the reactive transport of N species in the vadose zone and groundwater system [27,28,29,30,31,32]. In rural areas, mixed influences of multiple N sources, such as agriculture, livestock, and wastewater, are typically observed in groundwater, and quantifying the N load from septic effluent remains challenging. Karlović et al. [29] and Okuhata et al. [30] quantitively evaluated the N load from the septic system to groundwater in mixed land-use areas. Specifically, Okuhata et al. [30] combined the interpretation of simulated NO₃⁻ concentrations in the coastal Keauhou Aquifer, Hawaii, with δ¹⁵N analysis to validate the identification of multiple N sources. They suggested that OSDS accounted for the highest proportion of the N load (54%) compared with other N sources. Humphrey et al. [33] calculated the N loading from an on-site septic system installed in three different types of soil. Sand and sandy loam soils pose a higher risk of N loading than clayey soils. Pang et al. [31] simulated the reactive transport of N species and fecal coliforms discharged from clustered septic tank systems in New Zealand, which revealed a significant cumulative impact on groundwater NO₃⁻ concentrations, although the impact of fecal coliforms was localized. MacQuarrie et al. [32] developed a multicomponent reactive transport model to simulate a wastewater plume at the Cambridge site in the USA and discovered that the most significant biogeochemical changes occurred in the unsaturated zone, with complete nitrification converting NH₄⁺ to NO₃⁻.

Jeju Island, formed by multiple volcanic eruptions, is a tourist destination known for its stunning scenery and pristine environment on the Korean Peninsula. The island is primarily composed of porous volcanic rock, which allows rapid groundwater recharge; however, it is also vulnerable to surface contamination and land-use effects [34,35,36]. KOSIS [37] and JSSGP [38] reported that the population and visitors increased by 13% and 23%, respectively, over a decade (2013–2023), leading to an increase in wastewater discharge. Owing to an increase in population and visitors, many buildings have been constructed in mid-mountainous areas with elevations of 200–600 m a.s.l. However, public sewage networks are not sufficiently connected; therefore, personal sewage treatment facilities (PSTFs, known as on-site septic systems), which can purify wastewater from residential and commercial facilities, have been rapidly installed. PSTFs are the most numerous point sources, with a total of 10,977 in 2022, showing a two-fold increase over a decade (5141 in 2010). On Jeju Island, the effluent from the PSTFs directly infiltrates underground because of the lack of perennial streams due to its hydrogeological characteristics. JRI [39] examined the water quality of ten PSTFs effluents and, except for two facilities, almost all samples exceeded the effluent standard regulated by the Jeju Special Self-Governing Province, as total nitrogen (T-N) levels were exceeded three times. Similarly, JSSGP and Korea University [40] reported that manure and sewage impacted groundwater resources for tap water supply in 31 of 113 wells, whereas chemical fertilizers affected 68 wells.

Concerns have been raised about groundwater pollution on the island due to improperly treated effluents from the PSTFs. Calculating and predicting the quantitative impact of PSTFs on groundwater is crucial for groundwater quality management against the NO₃⁻ contamination. Therefore, this study aimed to (1) estimate the N loading rate based on field data from PSTFs in the Jocheon watershed, which is located in the northeastern part of Jeju Island and is characterized by the concentrated distribution of PSTFs, based on the T-N analysis in effluent samples from the PSTFs and the effluent discharge amount, (2) predict spatiotemporal behavior of the N species in the unsaturated and saturated zones using a numerical model considering reactive transformation of NH₄⁺, and (3) simulating the NO₃⁻ leaching on the water table by major N sources such as the PSTF effluents and chemical fertilizers applied from crop fields. Based on the results of the PSTF effluent quality and the numerical model, this study proposed appropriate PSTF management plans for Jeju Island.

2. Study Area

Jeju Island is located in the southwestern part of the Korean Peninsula and was formed by multiple volcanic eruptions over 1.8 Ma [41]. In Jeju, groundwater is the sole water resource because perennial streams are rarely developed on the island due to the hydrogeological characteristics of the highly permeable volcanic rocks [42]. Approximately 96% of the total water use relies on groundwater for drinking, irrigation, and other purposes [43].

The Jocheon study area is situated in the northeastern part of Jeju Island (Figure 1a,b) and covers 145 km². The annual rainfall in Jocheon was 1517 mm in 2022 (measured at the Daehul weather station [43]), which was higher than that for the entire island (1362 mm). Additionally, the groundwater recharge rate in the Jocheon watershed was 55%, whereas the mean recharge rate for the entire island was 46%. Figure 1c displays the geological distribution of the study area through an A to A′ cross-section of the boreholes (Figure 1b, mint-filled circles). The two major geological layers of volcanic rocks and the Seogwipo Formation (SGF) comprise the subsurface geology of the study area. The SGF, which underlies volcanic rocks, is composed of sand, bioclastic shells, basaltic tuff, and mudstone, and is characterized by low permeability [42]. Volcanic rocks, which are the major geological units on Jeju Island, overlie the SGF. Clinkers, joints, and fractures developed in the volcanic rock layers act as the main channels for groundwater flow [35]. The clay unit (paleosol) is interbedded between the volcanic rock layers, indicating periods of volcanic dormancy [44].

The Jocheon study area is covered with 10 km² of urban area (7% of the total land area), 47 km² of agricultural land (32%), and 88 km² of forests and grasslands (61% [45]). In the agricultural land, tangerine, buckwheat, and soybeans are commonly farmed and received 890,377 tons of N fertilizers in 2021 accounting for 5.5% of the total N use (16,110,222 tons) on the entire island. In the study area, there are 1128 PSTFs with a total capacity of 4775 m³/day, accounting for 10% of the total PSTFs on Jeju Island (Figure 1b). According to the daily treatment capacity, there were 934, 186, and 8 PSTFs < 5 m³/day, between 5 and 50 m³/day, and ≥50 m³/day, respectively. On Jeju Island, PSTFs with ≥50 m³/day capacity (1% of the total PSTFs in Jocheon) were required by law to hire a professional treatment manager until April 2024, whereas PSTFs with <50 m³/day capacity (99%) are managed by building owners who may lack of understanding of sewage treatment. Owing to the ordinance amendment, from May 2024, a PSTF with ≥20 m³/day capacity is required to hire a professional treatment manager.

3. Materials and Methods

3.1. PSTFs and Groundwater Sampling

The PSTF effluents from 40 facilities in the study area were sampled between March and June 2022. Temperature, electrical conductivity (EC), pH, and dissolved oxygen (DO) were measured in situ using a multi-parameter instrument (ProQuatro, YSI, Yellow Springs, OH, USA). The effluents were collected in a 2 L polyethylene sterilized sampling bag to analyze the biochemical oxygen demand (BOD) and total nitrogen (T-N). BOD and T-N were analyzed at the KOTITI Testing & Research Institute (Gwacheon-si, Republic of Korea). Additionally, NH₄⁺ was analyzed in an additional 10 PSTFs effluent samples (Figure 1b, magenta-filled triangles) twice (March and June 2022) to compare the T-N concentration.

Groundwater was also sampled from 10 groundwater wells to assess the influence of PSTFs effluents on water quality (Figure 1b, blue-filled stars). Two groundwater sampling campaigns were conducted in March and June 2022. Temperature, EC, pH, and DO were measured in situ after purging the wells until the parameters stabilized. Samples were filtered using a 0.45 μm pore size filter and stored at 4 °C until before the chemical analysis. NO₃⁻ and NO₂⁻ were analyzed at the Korea Basic Science Institute (Cheongju-si, Republic of Korea), and NH₄⁺ was analyzed at the KOTITI Testing & Research Institute (Gwacheon-si, Republic of Korea).

3.2. Estimation of Discharge and N Loading from the PSTFs Effluents

To estimate the N loading from the PSTF effluent, the sewage discharge of each building (1128 in total) with installed PSTFs and the N loading unit of the building were used. The sewage discharge rate was calculated based on the tap water usage data of the building in 2022 (provided by Jeju Special Self-Governing Province). The conversion ratio from tap water to sewage is 0.8, according to a previous study [46]. The N loading unit was calculated based on building usage and PSTF capacity using the 40 PSTF effluent T-N concentration. Eleven types of building usage were considered (Figure S1), and the capacity was divided into three classes: <5 m³/day, between 5 and 50 m³/day, and ≥50 m³/day. PSTFs with a capacity exceeding 50 m³/day were supposed to discharge 20 mg/L of T-N, which is the maximum allowed T-N level, because they were managed by experts. PSTFs with a capacity below 50 m³/day were expected to discharge a range of 1.7 mg/L to 53.8 mg/L T-N, depending on their capacity and building usage (Table S1).

3.3. Model Settings

A three-dimensional numerical model was used to assess the behavior of N pollutants from the PSTF effluent in the unsaturated and saturated zones. HydroGeoSphere [47] was used to simulate variably saturated groundwater flows and the reactive transport of N species (NH₄⁺, NO₂⁻, and NO₃⁻) in this study. The governing equation for three-dimensional variably saturated groundwater flow [47] is as follows,

$$-\nabla \left(q\right)\pm Q={\displaystyle \frac{\partial }{\partial t}}\left({\theta }_s{S}_w\right)$$

(1)

$$q=-K{k}_r\nabla (\psi +z)$$

(2)

where q is the fluid flux (L/T) given by Equation (2), Q is a volumetric fluid flux per unit volume representing source/sink (1/T), θ_s is the saturated water content (dimensionless), and S_w is the degree of water saturation (dimensionless), t is the time (T), K is the saturated hydraulic conductivity tensor (L/T), k_r is the relative permeability with a function of S_w, ψ is the pressure head (L), and z is the elevation head (L).

Also, the governing equation for three-dimensional transport of solutes in a variably saturated porous medium [47] is as follows,

$$-\nabla \left(qC-{\theta }_s{S}_{w}D\nabla C\right)+{\left[{\theta }_s{S}_{w}R\lambda C\right]}_{par}\pm Q_c=\left[{\displaystyle \frac{\partial \left({\theta }_s{S}_{w}RC\right)}{\partial t}}+{\theta }_s{S}_{w}R\lambda C\right]$$

(3)

where C is the solute concentration (M/L³) of the current species among multiple species, D is the hydrodynamic dispersion tensor (L²/T), λ is a first-order decay constant (1/L), par is the parent species for a decay chain, and Q_c is a solute exchange with the outside of the model domain representing source/sink (M/L³/T). R is the retardation factor (dimensionless).

The boundary of the model domain was set according to the coastline and groundwater flow directions. The model domain covered an area of 154 km² with vertical and horizontal dimensions of 12 km and 15 km, respectively (Figure 1b). The surface of the model domain was defined based on the topology using digital elevation model (DEM) data. The elevation ranged from 0 m a.s.l. on the coastline to 563 m a.s.l. at the highest point in the southwestern region (Figure 2a). A uniform elevation of the domain bottom was set at −300 m a.s.l. The model grid was constructed using a finite element triangular mesh, and the mesh around the monitored PSTFs and groundwater wells was refined. The model lithology was constructed using two layers, volcanic rocks and the SGF (Figure 2b), based on geologic logs obtained from groundwater wells in the study area [48]. The input parameters for the model are provided in

Table 1. Input parameters used for the model.

Geological Layer	Hydraulic Conductivity		Porosity	Specific Storage	Van Genuchten Function Parameter			Dispersivity
	Kx = Ky (m/s)	Kz (m/s)	n (−)	Ss (m−1)	Swr	α (m−1)	β	aL (m)	aT (m)
Volcanic rocks	4.0 × 10−4	9.0 × 10−5	0.35	1.0 × 10−5	0.38	3.50	3.18	100	10
SGF	5.0 × 10−7	5.0 × 10−8	0.40	1.0 × 10−4	0.49	0.70	1.68	30	3

, as referenced in previous studies [49,50,51].

The boundary condition (BC) of a constant head (Figure 2c) was set along the coastline of the northern part of the model domain at 0 m. No flow BCs were assigned along the upstream, left, or right sides of the outer domain boundaries. Uniform groundwater recharge conditions were set at the top domain surface based on yearly precipitation data from 1991 to 2020 recorded at the weather station (Figure S2). The recharge rate was calculated as 55% [52] of the yearly mean precipitation in the different elevation zones of 0–200, 200–400, and 400–600 m a.s.l. Groundwater pumping at the agricultural wells was assigned using groundwater usage data from 2001 to 2020 (provided by the Korea Rural Community Corporation in Jeju).

For the N species transport, the N loading of the T-N discharge from 1280 PSTFs (Figure S3) was used for modeling. The T-N loading rate was converted to the NH₄⁺ concentration based on the sewage discharge and water quality analysis results of the PSTFs, showing a linear relationship (Equation (4), Figure 3). The NH₄⁺ that infiltrates underground is rapidly oxidized and converted into NO₂⁻ and NO₃⁻ [53]. To simply simulate the nitrification process of NH₄⁺ from the PSTF effluent, sequential first-order decay reactions [54] were utilized (Equation (5), Table S2). The denitrification transformation was not considered in our model, as the denitrification process does not occur in the groundwater of Jeju Island, according to the previous study [6]. The NH₄⁺ component has a positive charge, allowing it to be adsorbed onto the negatively charged surfaces of soil particles [55]. Consequently, the behavior of high-concentration NH₄⁺ pollution discharged into sewage, removal, and retardation of NH₄⁺ components due to adsorption was observed [54]. The adsorption of NH₄⁺ was also considered in the model using a retardation factor (R).

$${NH}_4^{+}=0.82\times TN(R^2=0.93)$$

(4)

$${NH}_4^{+}\stackrel{{\mu }_1}{\to }{NO}_2^{-}\stackrel{{\mu }_2}{\to }{NO}_3^{-}$$

(5)

The variables μ₁, and μ₂ represent the first-order decay constant, respectively. Values of first-order decay constants and retardation factor used in the model are listed in Table S2, with referenced values from a previous study [56].

In the study area, N chemical fertilizers applied to agricultural fields also contribute to the enrichment of NO₃⁻ in groundwater, in addition to PSTFs [57]. Therefore, the influx of NO₃⁻ from agricultural activities was incorporated into the solute transport model. Modeling of the transport of N species was conducted under the following four scenarios: In scenario 1, the N loading was assumed to be caused only by the PSTFs effluent. Scenario 2-1 only considered the N loading from chemical fertilizer applied to agricultural land. In scenario 2-2, the N loading was assumed to be caused by both the PSTF effluent and N fertilizer. For areas with agricultural fields (Figure 1b), the observed NO₃⁻ concentration in the soil layer, due to the use of chemical fertilizers in the Gosan area of Jeju Island [58], was referenced. The amount of N fertilizer applied in 2017 in the Gosan and Jocheon areas [57] was used to adjust the NO₃⁻ concentration in the upper soil layer of the agricultural fields in the Jocheon model area. In scenario 3, the discharge of the PSTFs effluent was assumed to follow the T-N standard of 20 mg/L (Table S3) to evaluate the effect of managing the water quality of the effluent. The loading of agricultural N sources was not considered in scenario 3.

4. Results and Discussion

4.1. PSTFs Effluent Quality

Temperature of the PSTF effluents ranged from 11.1 to 27.2 °C (

Table 2. Descriptive statistics of water quality parameters in the PSTFs effluent.

Water Quality Parameter	Unit	Mean	Median	STDEV	MIN	MAX
Temperature	°C	19.9	20.1	4.3	11.1	27.2
EC	μS/cm	764	670	344	119	1711
pH	-	7.1	7.2	0.6	4.3	7.9
DO	mg/L	2.6	2.0	2.2	0.1	8.9
BOD	mg/L	106.3	76.5	99.4	0.8	387.0
T-N	mg/L	48.9	43.9	28.5	3.2	131.1
NH4+	mg/L	43.7	31.8	33.5	ND 1	125.5

¹ ND: Not detected.

), approximately equivalent to room temperature. EC ranged from 416 μS/cm to 1711 μS/cm, which is similar to the sewage EC of 1480 ± 131 μS/cm investigated by Geary and Lucas [59]. The pH showed a neutral range for an average of 7.0 ± 0.7, the appropriate condition for microbial activity in nutrient degradation [60]. DO averaged 2.9 ± 2.6 mg/L, sufficient for aerobic degradation [60]. BOD and T-N concentrations varied from 5.4 to 107.8 mg/L and 3.2 to 131.1 mg/L, respectively.

A total of 22 PSTFs (92% of 24 PSTFs with a capacity <5 m³/day) and 12 PSTFs (75% of 16 PSTFs with a capacity ≥5 m³/day) exceeded the effluent standards for BOD at <20 mg/L and <10 mg/L (Table S3), respectively. A total of 13 PSTFs (81% of 16 PSTFs with a capacity ≥5 m³/day) exceeded the effluent standards for T-N (<20 mg/L, Table S3). The NH₄⁺ concentration in the PSTF effluents ranged from not detected (ND) to 125.5 mg/L. According to seasons, NH₄-N and T-N from the effluent in March ranged from 1.0 to 97.4 mg/L and 17.2 to 111.0 mg/L (Figure S4a), respectively. In June, NH₄-N and T-N exhibited similar ranges from ND to 125.5 mg/L and 4.4 mg/L to 131.1 mg/L, respectively. Seasonal differences in the NH₄-N and T-N concentrations in the effluent were not observed. The major N species in the PSTF effluents was NH₄⁺, constituting a contribution ratio ranging from 0.0 to 75.6% (median: 61.3%) implying that the nitrification process, in which NH₄⁺ is oxidized to NO₃⁻, did not significantly occur during PSTF treatment. In addition, as shown in Figure 3, T-N and NH₄⁺ had a linear relationship, with 82% of the T-N being NH₄⁺ (R² = 0.93).

4.2. Groundwater Quality

In groundwater samples from the study area, temperature, EC, pH, and DO were averaged at 14.9 ± 1.8 °C, 97.9 ± 22.2 μS/cm, 7.9 ± 0.4 and 10.2 ± 1.0 mg/L, respectively (

Table 3. Descriptive statistics of water quality components in groundwater.

Water Quality Component	Unit	Mean	Median	STDEV	MIN	MAX
Temperature	°C	14.9	15.3	1.8	8.8	16.7
EC	μS/cm	98	97	22	68	157
pH	-	7.9	7.9	0.4	7.0	8.6
DO	mg/L	10.2	10.1	1.0	7.8	12.3
BOD	mg/L	1.1	1.0	0.5	0.2	2.2
T-N	mg/L	1.2	1.1	0.8	0.1	2.9
NH4+	mg/L				ND 1	ND
NO2−	mg/L				ND	ND
NO3−	mg/L	5.2	4.3	3.2	1.7	12.2
Cl−	mg/L	8.7	8.8	1.1	7.1	11.5

¹ ND: Not detected.

). The levels of BOD and T-N ranged from 0.2 mg/L to 2.2 mg/L and from 0.1 mg/L to 2.9 mg/L, respectively, which were significantly lower than those found in PSTF effluents. According to the conditions required for denitrification [61], a DO level below 2.0 mg/L is suitable for denitrifying bacteria to use NO₃⁻ as an electron acceptor. Additionally, BOD/T-N ratio of groundwater in the Jocheon area, which averaged 2.01 ± 2.93, is insufficient to cause denitrification using heterotrophic bacteria [62]. Therefore, denitrification was unlikely to occur in groundwater of the study area. NH₄⁺ and NO₂⁻ were not detected in the groundwater samples, indicating that the organic N compounds discharged from the PSTF were sufficiently oxidized during infiltration through the subsurface medium.

Average NO₃⁻ and Cl⁻ concentrations in the groundwater were measured as 5.2 ± 3.2 mg/L and 8.7 ± 1.1 mg/L (Table 3), respectively. The NO₃⁻ concentration was higher than the background water quality level in Jeju Island (5.5 mg/L [63]), while the Cl⁻ was close to the natural water quality level (average: 8.8 mg/L [64]). Koh et al. [34] suggested that a linear increase in NO₃⁻ compared to Cl⁻ indicated the influence of anthropogenic contamination in the groundwater on Jeju Island. The relationship between Cl⁻ and NO₃⁻ in Jocheon groundwater also showed a trend similar to that suggested by Koh et al. [34], indicating the influence of anthropogenic N contaminants (Figure 4). As see the spatial distribution of NO₃⁻ in the groundwater (Figure S4b), the NO₃⁻ levels increased along the groundwater flow through the clustered PSTF area. Groundwater wells (UG01, UG03, and MG03), which are located upstream of the clustered PSTF area, showed lower NO₃⁻ levels (1.7–3.7 mg/L) compared to the wells in the downstream area (DG01, DG05, and DG04, NO₃⁻ concentration: 4.1–11.6 mg/L). This indicated that the effluent from the clustered PSTFs likely accumulated NO₃⁻ in the groundwater as it flowed downstream. However, various N sources, such as chemical fertilizers, livestock waste, and septic wastewater, have been reported to affect NO₃⁻ in the groundwater [57,65,66,67]. Therefore, the combined influence of N sources, in addition to septic wastewater, can occur in the study area.

4.3. Calculation of T-N Loading by PSTFs Effluent

The T-N loading units of the PSTFs were calculated based on the T-N results of the 40 PSTFs, considering their building usage and capacity (Table S1 and Figure S1a). The T-N loading rate was multiplied by the unit load of T-N and the sewage discharge quantity of 1280 PSTFs. Therefore, the total amount of T-N discharge quantity was 15,540 kg/y. According to building usage, 9510 kg/y (61%) of T-N was from residential buildings, 2332 kg/y (15%) was from business facilities, and 2085 kg/y (13%) was from accommodation facilities (Table S4). Regarding the PSTF capacity, 10,358 kg/y (67%) of T-N was from PSTFs with a capacity of less than 5 m³/day, 4988 kg/y (32%) of T-N was from PSTFs with a capacity between 5 and 50 m³/day, and 194 kg/y (1%) of T-N was from PSTFs with a capacity of ≥50 m³/day. Scattered T-N loading from the PSTFs was observed throughout the study area (Figure S3). Clustered PSTFs are located in several regions and have overlapping T-N loadings, which have a cumulative impact on groundwater quality [31,68]. In addition, the PSTFs were also located in mid-mountainous areas (elevation ranges from 200 to 600 m a.s.l.), and septic discharges of 974 kg/y T-N were estimated (Figure S3). On Jeju Island, the mid-mountainous area is considered the major groundwater recharge region [69,70], therefore, it is important to prevent groundwater contamination by PSTF effluents.

4.4. Numerical Modeling Results

4.4.1. Model Calibration

Calibration of the numerical model was conducted for the groundwater head and NO₃⁻ concentration using a trial and error method. For the transport model of N species, only the NO₃⁻ component was used for the model calibration as the other N species, NH₄⁺ and NO₂⁻, were not detected in the groundwater samples from the study area (Table 3). The groundwater head was calibrated using the groundwater levels observed in the monitoring wells (blue circles in Figure S2) by adjusting the hydraulic conductivity of each geological layer, volcanic rock, and SGF. A high R² value of 0.9269 was obtained by comparing the simulated and observed heads (Figure 5), indicating that our model sufficiently represented the groundwater flow field in the study area. The highest groundwater level was simulated in the southwestern highland (>150 m a.s.l.) and gradually decreased toward the northern coastal area (0 m a.s.l.), as shown in Figure 6a. Groundwater generally flows from the southwest to the north. An unsaturated zone, with saturation lower than 0.4, developed, and below the water table, a fully saturated condition was simulated by the model (Figure 6b).

Unlike the head calibration, an R² of 0.5958 was obtained in the NO₃⁻ calibration from the scenario 1 model (Figure 7a), which considered N loading only from the PSTF effluent. The simulated NO₃⁻ concentrations in scenario 1 were lower than the measured NO₃⁻. However, scenario 2-2, which included additional N sources such as PSTF effluent and chemical fertilizers, exhibited improved calibration results with an R² value of 0.6140 (Figure 7b). Despite this improvement, several wells (DG05, DG04, and MG01) showed lower simulated NO₃⁻ values than the observed NO₃⁻, implying a lack of consideration of the spatially heterogeneous characteristics of agricultural activities in our model.

4.4.2. N Species Transport

The vertical distribution of the N species (NH₄⁺, NO₂⁻, and NO₃⁻) over time, as a simulated result (scenario 1) of the NH₄⁺ decay chain, is presented in Figure 8. After the release of the effluent from the PSTFs, NH₄⁺ exists at higher concentrations near the surface over time (Figure 8a–c). However, vertical infiltration is limited because of attenuation by nitrification and adsorption onto the soil medium. MacQuarrie et al. [32] also discovered restricted vertical and horizontal movement of the NH₄⁺ plume at the Cambridge site, Canada, showing the occurrence of nitrification completion of less than 1 m in the unsaturated zone from the septic system. Similar to the NH₄⁺ plume, the NO₂⁻ concentration is elevated near the surface (Figure 8d–f) as NH₄⁺ converts to NO₂⁻ and then rapidly converts to NO₃⁻. NO₃⁻, the final product of nitrification, gradually migrates into the subsurface over time (Figure 8g–i) as its movement is not inhibited by adsorption or chemical reactions in the unsaturated zone. The NO₃⁻ concentration increased and stabilized, reaching its highest level after 10 years, showing the highest NO₃⁻ concentration immediately below the NH₄⁺ plume.

Figure 9 shows the spatial distribution of the simulated NO₃⁻ concentration leaching to the water table. In scenario 1, considering the PSTF effluent as the only N source, the downstream area reached a maximum NO₃⁻ concentration of 94 mg/L (Figure 9a), which exceeds the drinking water standard for NO₃⁻ in South Korea (45 mg/L and 10 mg/L as NO₃-N). Leaching of higher NO₃⁻ concentrations tended to occur in the clustered PSTF areas with elevations lower than 200 m a.s.l. A clustered septic tank system has a cumulative impact on groundwater [31,71]. Pang et al. [31] demonstrated that density increases in a septic system accumulate NO₃⁻ in groundwater, owing to a reduction in dilution effects. The results of our model indicated that, although clustered PSTFs were also located in highland areas (elevation > 200 m a.s.l.), lower NO₃⁻ concentrations were simulated to leach into the water table. In high-elevation regions, a higher precipitation rate facilitates NO₃⁻ dilution in groundwater as the PSTF effluent infiltrates through a thick unsaturated zone over a longer period. Meanwhile, the low-elevation region receives lower precipitation, and a shallow unsaturated zone develops, enabling the rapid arrival of the NO₃⁻ plume with minimal dilution.

In scenario 2-1, the N loading was only from chemical fertilizer, and higher concentrations of NO₃⁻ exceeding 100 mg/L were simulated in the northwestern lowland area (Figure 9b) where agricultural areas were densely located (Figure 1b). Compared to the NO₃⁻ distribution in scenario 1, the chemical fertilizer significantly elevated groundwater NO₃⁻ in the study area more than PSTFs effluents. In scenario 2-2 (Figure 9c), assuming two N sources of PSTF effluent and chemical fertilizer usage, the distribution of NO₃⁻ exceeded the background water quality threshold of 5.5 mg/L [63]), covering a broader area than that in scenarios 1 and 2-1. Owing to the diverse N-related human activities, the influence of multiple N sources on groundwater NO₃⁻ has been discovered, and it is challenging to discriminate septic-derived N from various N sources. Multiple stable isotopes (such as N, B, and S) and hydrophilic organic wastewater constituents (such as artificial sweeteners and pharmaceuticals) have been widely used to detect the effects of wastewater [72,73,74,75]. Therefore, further studies, including various source tracers, are required to differentiate human wastewater NO₃⁻ in the groundwater of the study area.

To evaluate the effect of managing the PSTFs, scenario 3 was simulated with the assumption that all discharged PSTFs followed effluent standards. All input data for modeling (such as, hydrologic parameters, boundary conditions, and septic discharge rate) were the same as in scenario 1; however, the T-N unit load of the PSTFs was assigned as 20 mg/L (converted into NH₄⁺ based on Equation (4)), which is the threshold of the PSTFs effluent standard. The simulation results of scenario 3 (Figure 9d) showed similar spatial NO₃⁻ leaching on the water table to scenario 1; however, the maximum NO₃⁻ concentration in the groundwater decreased from 94 mg/L to 27 mg/L (71% reduction). Therefore, PSTFs managing to observe effluent standards is essential for mitigating the degradation of groundwater quality by the septic-derived NO₃⁻.

4.5. Suggestion of PSTFs Management Strategies

Based on our model results, the PSTFs effluent had a significant impact on groundwater quality in the study area. Therefore, managing PSTFs is essential for preserving groundwater on Jeju Island. Similar to Jeju Island, many countries have experienced groundwater contamination from NH₄⁺, NO₃⁻, and fecal coliforms from on-site septic systems. Septic-to-sewer conversion has been implemented to address septic-related environmental problems [76,77]. In Rockledge City, Florida, USA, a reduction in groundwater NO₃⁻ by 70% was achieved after septic-to-sewer conversion [76]. Buszka and Reeves [78] predicted timescales for NO₃⁻ reduction following septic-to-sewer conversion in Charotte Harbor, Florida, USA, using numerical modeling approaches; they estimated 2–3 years for a 50% reduction and 8–10 years for a 90% reduction. The most promising strategy for managing PSTFs is septic-to-sewer conversion based on priority classification. The Hawaiian government prioritized cesspool conversions, designating top priority (priority 1) areas with a significant risk of human impact, such as source water protection areas [79]. In addition, clustered septic systems significantly degraded groundwater quality, as shown in our model results. Therefore, clustered PSTFs also have priority in sewer conversion.

Building owners on Jeju Island currently manage PSTFs with a capacity of <20 m³/day. JRI [39] and our study showed that effluents from small-scale PSTFs exceeded the effluent standard owing to improper management by owners. Therefore, PSTF owners should receive compulsory education regarding proper PSTF management. In priority PSTF management areas, regular examinations of effluent water quality and operation conditions should be regulated by the government.

5. Conclusions

This study used a numerical model to simulate the reactive transport of N species derived from personal sewage treatment facilities (PSTFs, known as on-site septic systems) in the Jocheon area of Jeju Island, South Korea. Moreover, the chemical analysis of PSTF effluents and groundwater in the study area was also performed to assess the quality of PSTF effluent and groundwater. The PSTF effluents detected high T-N (3.2–131.1 mg/L) and NH₄⁺ (ND–125.5 mg/L) concentrations, and 75–92% of PSTFs exceeded the effluent standards. In groundwater, the T-N concentration was low at 0.1–2.9 mg/L, and NH₄⁺ and NO₂⁻ were not detected, indicating the sufficient oxidation of organic N derived from the PSTFs as it infiltrated through the subsurface. Elevated NO₃⁻ concentrations along with Cl⁻ suggested the influence of anthropogenic contaminants, such as chemical fertilizers and PSTF effluents; however, a further study is needed to discriminate the impact of septic-derived N sources on groundwater.

Based on the model results, the vertical infiltration of NH₄⁺ species was limited, showing the highest concentration in the unsaturated zone near the surface owing to nitrification and adsorption on the soil medium. NO₂⁻ also presented a similar plume behavior over time as NH₄⁺ rapidly converted to NO₃⁻. The NO₃⁻ component migrated through the saturated zone with an increasing concentration over time, showing the highest NO₃⁻ levels immediately below the NH₄⁺ plume. By discharging the PSTF effluents, the NO₃⁻ leached on the water table with a higher concentration in the lowland area (elevation < 60 m a.s.l.) compared to the highland (elevation > 200 m a.s.l.) due to a lower dilution effect and rapid contaminant arrival. In addition, the clustered PSTFs induced accumulative impacts on the groundwater, resulting in elevated NO₃⁻ levels. In addition to the PSTFs, the agricultural N source greatly contributed to groundwater NO₃⁻ contamination in the model. Based on the prediction model results following the standard of PSTF effluents, a 71% reduction (94 mg/L → 27 mg/L) in the highest NO₃⁻ concentration in groundwater was evaluated. Therefore, to manage PSTFs on Jeju Island, regular examinations of effluent water quality and compulsory education of the owners are suggested to prevent degradation in groundwater quality. Ultimately, septic-to-sewer conversion should be implemented in priority areas with a significant risk of human impact and clustered PSTFs.