Effects of Straw Mulching on Nonpoint Source Pollutant Runoff During Snowmelt in Korean Highland Agricultural Areas

Abstract

Nonpoint Source (NPS) pollution refers to water pollution that does not originate from a single identifiable source. In this study, we conducted water-quality monitoring during the snowmelt period from February to March of 2024 and 2025 in Jaun, a highland agricultural area. We analyzed changes in nonpoint source pollutant concentration and evaluated Best Management Practice (BMPs) effectiveness. A year-to-year comparison showed that in 2024, a single intense snowmelt event led to a sharp increase in particulate pollutants, such as TP and SS. In 2025, repeated and gradual thawing resulted in the accumulation and release of dissolved pollutants, including TN and TOC. BMPs such as straw mulching were partially effective in reducing pollutant concentrations. However, in 2025, a lack of proper maintenance led to increased concentration at certain sites. The water quality during the snowmelt period was comparable to that during the summer monsoon season, indicating that snowmelt has a similar potential for generating nonpoint source pollution. The findings provide area-based insights into snowmelt-induced nonpoint source pollution and can form a foundation for developing seasonal water quality management policies.

1. Introduction

Nonpoint Source (NPS) pollution refers to water pollution that does not originate from a single identifiable source [1]. In contrast with point-source pollution, which originates from identifiable sources such as industrial facilities or wastewater treatment plants, NPS pollution typically occurs as water flows over the land surface during rainfall or snowmelt events. The runoff generated mobilizes a variety of pollutants, including nutrients, sediments, organic matter, and pathogenic microorganisms. It flows over agricultural areas, impervious surfaces, and disturbed soils, eventually discharging into rivers, lakes, coastal zones, and groundwater systems [1]. Agricultural nonpoint source (ANPS) pollution is increasingly recognized as a major contributor to global water quality deterioration [2,3]. ANPS pollutants are generally introduced into aquatic systems through surface runoff generated by rainfall or snowmelt, soil leaching, and irrigation return flow. All of these can have adverse effects on water quality and aquatic ecosystems. The input of nutrients such as nitrogen (N) and phosphate (P) can lead to water body eutrophication, resulting in algal blooms and water quality decline. The impact of ANPS pollution varies depending on several factors, including the location of the croplands, soil, topographic slope, precipitation, and agricultural management practices. Even with similar agricultural activities, the amount of pollutant runoff and its effects on water quality can strongly differ depending on the regional conditions [4,5,6,7].

Highland agriculture is actively practiced in the Republic of Korea, particularly in the province of Gangwon. This farming type is typically conducted on steep slopes, where vegetables are cultivated under poor soil conditions. Consequently, large amounts of compost and fertilizers are used to maintain crop productivity [8]. The high elevation and steep slopes in these areas intensify soil erosion during rainfall events [9]. This increases the potential for NPS pollution compared with lowland or flat agricultural areas. In response, in 2007, the Ministry of Environment (MOE) of the Republic of Korea designated areas with a high risk of turbid water generation in the upper Han River region as nonpoint source pollution management areas. A representative example is the Jaun District in Hongcheon-gun, Gangwon State [10]. Jaun District is located in the upper watershed of the Soyang Lake, which is a major drinking water source for the Seoul metropolitan area. Approximately 95.9% of the district consists of highland areas with elevations greater than 500 m above sea level. With an average slope of 43.1%, the area experiences extensive soil and nutrient runoff from agricultural land during rainfall events, which contributes to pollutant inflow into Soyang Lake [10]. To mitigate these issues, various best management practices (BMPs) have been implemented in the Jaun District alongside community-based governance initiatives. Straw mulching is one of the most widely recognized management practices in highland agricultural areas. The impact energy of rainfall or snowmelt on topsoil is reduced by covering the soil surface with rice straw without planting vegetation, which prevents soil erosion and reduces surface runoff [11,12,13]. In addition, structural BMPs such as gabion retaining walls and onion net barriers have been installed to stabilize slopes and capture sediment during runoff events. Non-structural BMPs have also been promoted through active community participation, including local monitoring activities, routine maintenance of installed facilities, and educational programs to encourage sustainable farming practices.

In Korea, in nonpoint pollution source management areas, to date, most studies have focused on runoff caused by rainfall. Meanwhile, pollutant runoff associated with snowmelt has been relatively overlooked. However, in high-altitude regions such as Jaun District, winter snowfall and spring snowmelt are major hydrological components of stream runoff, and the potential for NPS pollution resulting from these processes cannot be overlooked. Unlike rainfall events that typically generate short-term, intense runoff, snowmelt runoff occurs gradually as the accumulated snow melts. In contrast with rainfall events that typically generate short-term, intense runoff, snowmelt runoff occurs gradually as the accumulated snow melts. As the temperature rises, the moisture content of the previously frozen soil surface increases, leading to the mobilization of suspended solids and pollutants from the loosened topsoil into runoff [14]. High-altitude regions tend to have steeply sloped agricultural areas and high snowfall. During the snowmelt period, the accumulated snow can rapidly melt over a short period, generating a high volume of surface runoff [15,16,17].

Lucianetti et al. [18] used a stable isotope mixing model in the Italian Alps and found that up to 94% of the spring-to-early summer runoff originated from snowmelt. Similarly, Zappa et al. [19] reported that the peak streamflow coincided with the snowmelt period in the Swiss Alps, with the influence becoming more distinct at higher elevations. Cade-Menun et al. [20] reported losses of nitrogen, phosphorus, and sediment from croplands during snowmelt in Saskatchewan, Canada. This suggests that pollutant loads associated with snowmelt runoff can vary substantially depending on land use practices. Redding and Devito [16] reported that slope aspect, soil texture, and autumn soil moisture content influenced snowmelt runoff in the Boreal Plain region, western Canada. Their findings support the need for site-specific snowmelt management strategies that consider topography and soil. Snowmelt runoff in highland agriculture likely strongly influences hydrological processes and pollutant dynamics, and runoff can vary by season and land use type. Su et al. [21] reported that snowmelt runoff is a major source of phosphorus and sediment loss in agricultural watersheds and, in some cases, can generate greater pollutant loads than rainfall events. Most existing BMPs are primarily designed to address rainfall-included runoff and have limited effectiveness in mitigating snowmelt runoff.

However, to date, research on ANPS pollutants in Korea has predominantly focused on runoff during summer rainfall events [11,22,23]. Meanwhile, field-based studies analyzing snowmelt-induced runoff and the effectiveness of mitigation practices remain relatively limited. Choi [24] reported that when snowmelt occurs intensively over a short period, it can generate runoff equivalent to rainfall events exceeding 40 mm. The average runoff volume was also found to be statistically similar to that of a 25.5 mm rainfall event. Kwon et al. [25] reported that in the Doam Lake watershed, which is situated in the same highland agricultural region as the present study and where all cultivated fields are located at elevations above 400 m, snowmelt runoff significantly increased nonpoint source (NPS) pollutant concentrations. Specifically, event mean concentrations (EMC) during snowmelt were 81.3 mg/L for suspended solids (SS), 0.15 mg/L for total phosphorus (TP), and 3.46 mg/L for total organic carbon (TOC), representing 20.6-fold, 2.8-fold, and 2.2-fold increases, respectively, compared to baseline winter conditions. These findings emphasize the potential for substantial NPS pollution during snowmelt periods in highland watersheds and reinforce the necessity of targeted monitoring and management, as addressed in the present study. However, the institutional and technological measures to address winter NPS pollution in highland agricultural areas remain inadequate. Although some studies have used soil and water assessment tool (SWAT)-based modeling to predict hydrological and water-quality changes associated with snowmelt [26,27,28], there is a substantial amount of field-based research that analyzes water quality through sampling during snowmelt periods and evaluates the effectiveness of mitigation practices.

We conducted a field-based analysis of NPS pollutants in stream runoff during the snowmelt period in a highland agricultural area, focusing on the Jain District of Gangwon State, South Korea. We aimed to assess the effectiveness of NPS mitigation practices (BMPs), such as straw mulching, during the winter season to evaluate the practical applicability of snowmelt period management strategies. Considering the limitations of flow and precipitation measurements at the study site, the timing of snowmelt events was estimated based on decreases in snow depth and rising air temperatures. Pollutant runoff was analyzed by examining the changes in water-quality parameters associated with the identified snowmelt periods. Based on this analysis, differences in pollutant concentrations were compared and interpreted according to the presence or absence of mitigation facilities to provide baseline data for developing appropriate NPS management strategies during the snowmelt period.

2. Materials and Methods

2.1. Study Area

We conducted this study in the Jaun District of Gangwon State, which covers an area of approximately 133.18 km² (Figure 1). Jaun District is located in the upper watershed of the Soyang Lake Basin, where the Jaun Stream flows into the Naerin Stream and is eventually discharged into the Soyang River. An agricultural area of approximately 11.46 km² is located within the watershed with an average slope of 43.1%. Among the 3563 farmland plots, 1506 were located in highland agricultural areas with slopes exceeding 20% and elevations above 550 m. Monitoring was conducted at five sites (St.1–St.5) located within a concentrated highland agricultural area in the upper reaches of the Jaun Stream. St.1 was minimally affected by external influences and equipped with a gabion-retaining wall. St.2 and St.4 were sites where straw mulching was applied. Meanwhile, St.3 represented a control site without ant mitigation measures. St.5 is located downstream, where a tributary joins the mainstream.

2.2. Field Monitoring and Analytical Methods

Considering periods of rising temperatures that indicate a high likelihood of snowmelt, monitoring was performed 11 times between February and March of 2024 and 2025. Runoff water at five locations (St.1 to St.5) during each event. In the field, water temperature, pH, electrical conductivity (EC), and dissolved oxygen (DO) were measured using a multiparameter water-quality meter (YSI ProPlus), and turbidity (Turb) was measured using a portable turbidity meter (Hach 2100P). Samples were collected in 3-L containers, transferred into sterile bottles, and transported to the laboratory under 4 °C.

Laboratory analyses of total suspended solids (TSS), total organic carbon (TOC), total nitrogen (TN), and total phosphorus (TP) were performed in accordance with the Standard Methods for the Examination of Water Pollution Process Tests [27]. TSS was determined gravimetrically using GF/C filters and drying at 105 °C. TOC was measured by high-temperature combustion, TN by UV/visible spectrophotometry, and TP by continuous flow analysis.

2.3. Meteorological Data Collection

The meteorological and precipitation data were obtained from the Korea Meteorological Administration Automated Weather System (AWS) station in Naemyeon. Snow depth data were collected from the Daegwallyeong station, which is part of the Automated Synoptic Observing System (ASOS). This was the nearest snow-measuring station to the sampling site. Located approximately 4 km from the study site, the Naemyeon AWS was used to provide temperature and precipitation data. The Daegwallyeong ASOS station is located approximately 32 km in a straight line from the sampling sites, and the data were considered suitable for representing snow accumulation in high-altitude mountainous regions. We estimated the snowmelt timing based on the interrelationships between changes in water-quality concentrations, reduction in snow depth, and rising air temperatures.

Snow depth reduction was calculated as the cumulative decrease from 00:00 on the day before sampling to 23:00 on the sampling day to match the reference period used for calculating the daily mean air temperature. Both snow depth variation data from the Daegawllyeon ASOS and air temperature data from the Naemyeon AWS were used in combination to determine the snowmelt occurrence.

2.4. Comparison with Previous Studies

Water-quality monitoring data (turbidity, SS, TP, TOC) from the Jaun Bridge site, as reported in the Environmental Reports of the Wonju Regional Environmental Office, 2020–2023 [29,30,31,32], were compared with the data from this study. The dataset was collected at the Jaun Bridge site, located approximately 12 km downstream of the Jaun District, in a straight line from St.5. The aforementioned dataset was used to compare the concentrations of NPS pollutants during snowmelt runoff with those during rainfall and non-rainfall periods. Because precipitation data for the sampling dates were included in the report, meteorological data from the Naemyeon AWS station were used as supplementary data.

3. Results and Discussion

3.1. Temperature and Snow Depth Variations in the Study Area

In this study, hourly temperature data from 1 February 2024, to 31 March 2025, were analyzed to assess the correlation between air temperatures at the Naemyeon AWS and the Daegwallyeong ASOS, the two stations closest to the study area (Figure 2a). Variations in snow depth in relation to temperature changes were also evaluated (Figure 2b). Although the two stations showed similar temperature trends, the Naemyeon station generally exhibited slightly higher average temperatures. The correlation between the two stations was analyzed using Microsoft Excel, yielding a coefficient of determination (R²) of 0.7863 and a p-value of less than 0.001, indicating a statistically significant relationship. In 2024, subzero temperatures ranging from −5 °C to −10 °C persisted until mid-February. Between February 13 and 15, air temperatures rose above freezing, resulting in a rapid decrease in snow depth to approximately 60 cm. This triggered the first major snowmelt event. Subsequently, the below-freezing temperatures returned from late February to early March, during which additional snowfall was recorded. A second snowmelt runoff event was observed beginning on 13 March, when temperatures rose above freezing. This led to the gradual melting of the remaining snowpack. In contrast, in 2025, the average temperature dropped sharply to −16.5 °C in early February, and then gradually increased. From early to mid-March, above-zero temperatures and fluctuations in snow depth alternated repeatedly. During this period, three–four dispersed snowmelt runoff events were observed. Cumulative runoff was associated with temporary reductions in snowpack, followed by renewed accumulation as temperatures repeatedly rose above freezing.

Variations in snow depth and temperature are key factors influencing the annual timing, intensity, and duration of snowmelt runoff. In this study, the potential for snowmelt-induced runoff was evaluated based on periods when rising temperatures coincided with decreasing snow depth. Specifically, days with a daily mean temperature of 2.0 °C or higher, or a maximum temperature of 6.0 °C or higher, were used as the primary temperature conditions. In addition, if the two-day cumulative snow depth reduction was 8.0 cm or more, the day was classified as a snowmelt runoff event.

Meanwhile, in cases where the mean temperature ranged from approximately 1.0 °C to 4.0 °C, snowmelt runoff was also identified when there was a marked increase in turbidity and suspended solids (SS) concentrations, even if the snow depth reduction was between 4.0 cm and 7.9 cm, based on complementary judgment criteria. For example, in the second monitoring round of February 2025, the sampling date differed by one day from the first round. Although no reduction in snow depth was observed on the measurement day itself, a significant decrease in snow cover was confirmed at the study site compared to the previous day, and this was clearly reflected in the water quality indicators. As a result, the day was exceptionally classified as a snowmelt runoff event, which is interpreted as being due to the spatial discrepancy between the Daegwallyeong meteorological station (used for snow observation) and the actual sampling location.

3.2. Interannual Variations in Water Quality During the Snowmelt Period

Based on the snowmelt period water-quality monitoring results for 2024 and 2025 (Figure 3), the concentrations of major water-quality parameters increased during snowmelt runoff events compared with non-snowmelt winter conditions. Turbidity, TP, TSS, and TOC showed relatively large variations and elevated concentrations during snowmelt. At the St.5 site, the average concentrations of key water-quality parameters during the two snowmelt events in 2024, including the intensive melt event on February 14, were 61.67 NTU for turbidity, 4.716 mg/L for TP, 64.20 mg/L for TSS, and 5.69 mg/L for TOC. These values represent approximately 16.5-, 9.3-, 9.6-, and 3.7-fold increases in turbidity, TP, TSS, and TOC, respectively, compared with the average concentrations observed during non-snowmelt winter periods in the same year. In 2025, when snowmelt occurred gradually and repeatedly, a comparison between snowmelt and non-snowmelt periods at St.5 showed that turbidity, TP, TSS, and TOC increased approximately 9.6-, 1.0-, 6.9-, and 1.1-fold, respectively (

Table A1. Water-quality monitoring results by site and date during winter and snowmelt periods in 2024 and 2025. Parameters include air temperature, electrical conductivity (Cond), turbidity, total nitrogen (TN), total phosphorus (TP), total suspended solids (TSS), and total organic carbon (TOC).

DATE (DD.MM.YY)	Site	Temp	Cond	Turb (NTU)	TN	TP	TSS	TOC
		°C	µS/cm	NTU	mg/L
14 February 2024 (Snowmelt)	St.1	2.3	24.0	1.81	1.193	0.052	1.7	2.48
	St.2	3.8	30.4	14.7	2.427	0.363	31.8	3.20
	St.3	3.9	65.3	92.0	4.250	11.599	114.0	8.77
	St.4	3.1	72.7	32.0	4.182	2.801	57.0	5.76
	St.5	3.4	65.2	114.0	4.418	9.399	120.0	9.35
20 February 2024 (Snowmelt)	St.1	3.1	23.4	1.44	1.747	0.001	1.5	1.28
	St.2	2.9	26.4	5.52	2.069	0.308	13.9	1.76
	St.3	2.8	39.6	8.78	2.689	0.821	12.8	1.85
	St.4	2.8	42.0	9.88	3.361	0.052	15.8	2.26
	St.5	3.0	45.8	9.34	3.176	0.032	8.4	2.02
27 February 2024	St.1	2.1	25.1	0.56	1.304	0.033	0.6	1.26
	St.2	2.5	41.0	4.08	2.676	0.019	7.8	1.29
	St.3	2.9	67.0	1.38	4.279	0.022	2.3	0.98
	St.4	3.0.	71.8	1.84	4.265	0.022	1.9	1.15
	St.5	2.7	72.3	1.82	4.913	0.066	1.6	1.28
11 March 2024	St.1	2.0	48.1	0.44	1.327	0.047	0.3	1.06
	St.2	2.6	60.3	6.35	2.805	0.319	13.2	1.86
	St.3	3.2	74.4	1.61	4.779	0.345	1.0	0.78
	St.4	3.6	81.1	0.92	3.095	0.352	1.3	1.10
	St.5	3.3	78.2	3.71	4.997	0.693	8.4	1.59
18 March 2024	St.1	5.4	31.5	1.82	1.810	0.094	1.1	1.00
	St.2	6.7	40.7	3.48	2.727	0.235	5.4	1.23
	St.3	8.9	71.0	1.70	4.156	0.744	1.7	1.23
	St.4	8.9	74.9	4.38	4.249	0.583	6.6	1.63
	St.5	9.2	77.1	5.70	4.410	0.755	10.0	1.71
27 February 2025	St.1	1.4	54.0	0.42	4.303	0.007	0.4	0.85
	St.2	1.3	56.7	0.56	4.765	0.008	0.4	1.01
	St.3	1.0	146.0	3.02	12.955	0.015	2.3	1.09
	St.4	0.8	151.9	2.54	13.376	0.014	5.2	1.45
	St.5	1.2	147.9	3.10	13.092	0.016	2.7	2.19
28 February 2025 (Snowmelt)	St.1	2.6	54.2	1.63	4.519	0.008	1.0	1.12
	St.2	3.0	56.3	1.62	4.810	0.020	1.5	1.13
	St.3	3.5	153.9	47.1	12.973	0.0833	32.4	1.81
	St.4	4.0	166.3	83.8	13.917	0.1286	61.8	2.15
	St.5	3.6	159.3	54.1	13.510	0.086	32.6	2.24
7 March 2025	St.1	2.7	50.9	0.69	4.110	0.006	0.2	0.89
	St.2	2.9	54.1	0.7	4.620	0.010	0.8	0.84
	St.3	5.0	147.3	1.59	12.882	0.018	1.8	1.28
	St.4	5.3	155.8	2.58	13.565	0.020	6.4	1.30
	St.5	5.3	155.4	1.72	13.360	0.017	1.5	1.35
12 March 2025 (Snowmelt)	St.1	5.0	47.5	3.23	4.102	0.017	3.9	1.33
	St.2	5.2	71.5	36.1	6.519	0.194	11.8	3.69
	St.3	6.7	120.8	13.4	10.833	0.091	14.8	1.99
	St.4	6.9	128.2	12.5	11.722	0.067	14.1	1.83
	St.5	6.7	122.0	10.9	9.446	0.056	14.6	1.91
21 March 2025 (Snowmelt)	St.1	6.3	51.3	6.61	4.600	0.030	32.9	1.46
	St.2	7.5	65.4	12.4	6.468	0.046	2.6	1.57
	St.3	10.0	130.6	41.9	11.758	0.099	43.0	2.16
	St.4	10.2	145.3	46.8	12.364	0.093	8.6	2.38
	St.5	10.0	128.0	19.9	9.927	0.050	12	1.88
27 March 2025	St.1	6.0	58.4	5.33	4.412	0.131	4.6	1.30
	St.2	6.7	57.7	2.62	4.932	0.228	2.2	1.34
	St.3	7.7	97.7	5.18	6.092	0.109	3.3	1.54
	St.4	7.7	100.9	5.61	6.086	0.119	3.8	1.50
	St.5	7.6	102.7	4.06	6.234	0.098	4.4	1.79

).

Similar findings have been reported previously. Macrae et al. [33] found that during the snowmelt period in an agricultural watershed in eastern Canada, the runoff of TP and SS increased sharply compared to baseline conditions. The soil freezing status and snowmelt rate play critical roles in influencing the runoff of NPS pollutants. Iwata et al. [34] and Niu and Yang [35] reported that when the soil remains frozen during snowmelt runoff events, meltwater cannot infiltrate deeply into the soil. Instead, it flowed rapidly over the surface, resulting in substantial surface runoff within a short period. This process also contributes to TP and SS loss from the soil surface. In frozen soil–snowmelt environments, streamflow often increases sharply during the early snowmelt stages, with simultaneous peaks in TP and SS concentrations. This frequently results in higher pollutant concentrations and loads than those observed during single rainfall events, intensifying pollution in receiving water bodies [36,37].

For TOC, a runoff pattern similar to that of SS and turbidity was observed in 2024. However, in 2025, the highest concentration was recorded at the midstream site (St.4), followed by a slight decrease at the downstream site (St.5). This difference is likely attributable to TOC comprising both particulate organic carbon (POC) and dissolved organic carbon (DOC). Therefore, it is influenced by sediment runoff and various factors such as straw decomposition, organic matter accumulation, and leaching. In 2024, the occurrence of rapid snowmelt accompanied by an abrupt rise in temperature likely resulted in the melting of the surface soil layer, leading to runoff dominated by POC and solid sediments. In contrast, in 2025, gradual and prolonged snowmelt may have favored the leaching of dissolved substances. This could explain the increased TOC concentrations observed in the midstream section, suggesting a relatively higher contribution of DOC to runoff [38]. However, this study did not directly quantify DOC and POC, and therefore, interpretations regarding TOC composition remain speculative. TOC, as a composite parameter of both particulate (POC) and dissolved (DOC) organic carbon, is known to be influenced by distinct hydrological pathways such as surface runoff and subsurface flow [39,40]. Therefore, the spatial variation in TOC observed in this study may reflect a shift in dominant flow paths depending on snowmelt dynamics. Future studies should incorporate direct measurements of DOC and POC to clarify their relative contributions and better understand the mechanisms of organic carbon transport during snowmelt events. This can be explained by TOC being a composite parameter consisting of both POC and DOC. Therefore, it is influenced by multiple runoff pathways. This aligns with previous studies suggesting that DOC runoff can exhibit distinct spatial distribution patterns depending on the snowmelt mode. Croghan et al. [41] reported that during gradual snowmelt and rising groundwater levels, DOC concentrations increased sharply, with a spatial distribution distinct from that of turbidity or SS. TN also shows a sensitive response to temperature changes and snowmelt patterns, as it is predominantly transported in dissolved forms [42]. The average TN concentration at St.5 increased from 4.4 mg/L in 2024 to 10.9 mg/L in 2025. This can be attributed to the cumulative runoff of dissolved nitrogen under repeated thawing conditions. To statistically assess the differences in TN concentrations between 2024 and 2025, a Mann–Whitney U test was conducted using the scipy.stats package in Python 3.12.7. The result showed a p-value of 0.000073, which is below the significance level of 0.05, indicating a statistically significant difference in TN concentration distributions between the two years. This result supports the observation that TN concentrations increased substantially during the 2025 snowmelt period.

In addition, Mann–Whitney U tests were performed to examine the responses of individual water-quality parameters according to both the presence of snowmelt and its intensity/trend. For TOC and SS, the differences according to snowmelt presence alone were statistically significant (p = 0.0000023 and 0.0000102, respectively), suggesting strong responses to snowmelt events. Furthermore, these parameters exhibited even more pronounced variability depending on the snowmelt intensity and trend. On the other hand, TN and TP did not show significant differences when classified only by the presence or absence of snowmelt (p = 0.780 and 0.563, respectively). However, TN exhibited statistically significant differences when categorized by snowmelt pattern, implying its sensitivity to the characteristics of snowmelt such as rate and duration. These findings highlight that water-quality responses during snowmelt can vary considerably by parameter and that simple binary classification of snowmelt presence may be insufficient. A more nuanced approach considering melt intensity and persistence is essential for accurately capturing snowmelt-driven pollutant dynamics.

Snowmelt runoff and responses of key water-quality parameters differed each year according to the snowmelt event type. In 2024, an intensive snowmelt event occurred on February 14, accompanied by substantial soil runoff. This led to a sharp increase in the particulate TP concentration. This indicates that a rapid increase in temperature exceeding 10 °C over a short period can induce intensive runoff of particulate pollutants due to a substantial influx of snowmelt. In contrast, in 2025, a prolonged and gradual snowmelt process led to a dominant pattern of cumulative runoff of dissolved pollutants, with a higher TN concentration. At the St.5 site, during the intensive snowmelt event in 2024, the average concentrations of total nitrogen (TN) and total phosphorus (TP) were observed as 3.80 mg/L and 4.716 mg/L, respectively. In contrast, during the prolonged and gradual snowmelt in 2025, TN increased to 11.56 mg/L, representing approximately a 3.0-fold increase. Meanwhile, TP decreased to 0.052 mg/L, that is, approximately a 90.7-fold decrease (Figure 3). These differences indicate that the forms and concentrations of pollutants in snowmelt runoff vary depending on the type of snowmelt event. This variation is closely associated with the degree of soil freezing and thawing.

Rapid snowmelt promotes the mobilization of surface soil particles and suspended solids, increasing the runoff of particulate pollutants [20]. Meanwhile, gradual snowmelt facilitates the leaching of soil organic matter, resulting in a progressive increase in the concentration of dissolved substances such as DOC [43]. These findings support the notion that the contribution of pollutants to snowmelt runoff may vary depending on the intensity and duration of thawing patterns. This highlights the need to develop tailored nonpoint source (NPS) management strategies based on specific water-quality characteristics during snowmelt.

3.3. Spatial Variability of NPS Pollutants and BMP Effectiveness During Snowmelt

The turbidity, SS, TP, and TOC concentrations increased from the upstream site (St.1) to the downstream site (St.5) (Figure 4). This pattern indicates that soils and NPS pollutants originating from highland agricultural areas were transported and accumulated downstream. St.1, which was located at the highest elevation, was equipped with a gabion-retaining wall that effectively limited the inflow of external pollutants. Therefore, the water quality observed at this site represents the characteristics of pure snowmelt water unaffected by additional surface runoff. Although straw mulch was applied at St.2, soil particles entering through retaining wall gaps and sediments accumulated on the streambed under low-flow conditions were resuspended and transported during the rapid increase in flow caused by snowmelt. This resulted in a temporary increase in pollutant concentrations. St.3. Here, no mitigation facilities were installed, exhibiting high concentrations of water-quality parameters owing to the direct observation of soil runoff induced by snowmelt. In contrast, St.4, located downstream of St.3 and adjacent to agricultural fields, showed relatively low concentrations. This was likely because of the straw mulch application. This is likely attributable to the combined effects of surface coverage by straw mulching and dilution caused by increased discharge. St.5, where no mitigation facilities were installed and a tributary joining St.4 and St.5, showed elevated pollutant concentrations, potentially because of the reintroduction of contaminants from the tributary.

In 2024, the mitigation effects were maintained as expected (Figure 4i). However, in 2025, field observations showed that straw mulching at St.4 had separated, allowing soil particles to enter through the gaps. Similarly, at St.2, unlike in the previous year, elevated concentrations were observed owing to the runoff of soil particles accumulated in the gaps of the retaining wall (Figure 4j). In 2025, TOC, SS, and TP concentrations increased at both St.2 and St.4. This suggests that the presence of NPS pollution mitigation measures and their continued maintenance play a critical role in improving water quality [44]. It is essential to implement systematic maintenance and inspection procedures to ensure the effective functioning of NPS mitigation facilities. These practices are expected to contribute to long-term water-quality enhancement [45].

Owing to the topographical constraints of the study watershed, flow measurements could not be conducted, limiting the analysis to concentration-based interpretations rather than assessments of runoff loads. This limitation highlights the need for supplementary approaches to evaluate pollutant runoff volumes accurately in future studies. Although snow depth data were obtained from the Daegwallyeong ASOS station located approximately 32 km from the sampling site, discrepancies were observed on certain monitoring dates where snowmelt occurred in the field but no corresponding decrease in snow depth was recorded (

Table 1. Snow depth reduction by observation round in 2024 and 2025.

Year	1st (cm)	2nd (cm)	3rd (cm)	4th (cm)	5th (cm)	6th (cm)
2024	23.3	1.3	1.5	0	1.1	-
2025	0	0 * (Snowmelt observed)	6.4	11.9	40.7	0

Note: * Although snowmelt was observed at the study site, no corresponding decrease in snow depth was recorded at the ASOS station.

). This is likely attributable to localized meteorological differences. To improve the accuracy of snowmelt timing estimations, future studies should consider installing additional snow sensors near the study site or operating small-scale meteorological stations in the watershed.

3.4. Comparative Analysis of Seasonal Runoff

To compare the snowmelt runoff observed during 2024–2025 with those under rainfall and non-rainfall conditions, raw monitoring data were obtained from the nonpoint pollution source management area monitoring and evaluation reports (Mandae, Gaa, and Jaun Districts, Doam Lake Watershed) published by the Wonju Regional Environmental Office of the Ministry of Environment (2020–2023) [29,30,31,32]. The analysis was conducted at the Jaun Bridge monitoring site in Jaun District of Hongcheon. This site is situated approximately 6 km downstream of St.5, which is the lowest monitoring point in the present study. In previous studies, snowmelt monitoring was conducted four times (13 March 2021, and 13 January 2023), rainfall monitoring was carried out 84 times between July 2019 and July 2023, and non-rainfall monitoring was conducted 63 times during the same period. However, rainfall observations during Typhoon Maysak (2–3 September 2020), which recorded more than 100 mm of precipitation, were excluded from the analysis because they were classified as extreme. A comparison of the water-quality parameters under different runoff conditions using box plots (Figure 5) showed that snowmelt runoff exhibited the highest concentration ranges for all parameters, including TP, TOC, SS, and turbidity. This is likely attributable to the increased soil runoff from fallow lands during the winter season and the transport of pollutants during snowmelt, particularly in the absence of mitigation measures following fertilizer application in preparation for spring cultivation.

Table 2. Comparison of snowmelt runoff analysis between prior research and this study.

	Monitoring Date	Turb (NTU)	TP (mg/L)	SS (mg/L)	TOC (mg/L)
St.5	14 February 2024	114.0	9.399	120.0	9.4
Jaun bridge	1 July 2020	116.9	0.622	123.1	10.4
Jaun bridge	4 July 2021	183.8	0.887	272.6	10.2

summarizes rainfall events from previous studies that exhibited water-quality concentrations similar to those observed during the first snowmelt monitoring in 2024 at Site 5 in the present study [29,30,31,32]. Rainfall observation data at Jaun Bridge, used for comparison, were collected during the monsoon season on 1 July 2020 (total precipitation: 65 mm) and 4 July 2021 (total precipitation: 70 mm). These events showed similar TOC and SS concentrations to those observed during snowmelt runoff in the present study. Although there are limitations to direct comparisons due to differences in monitoring periods and locations, snowmelt runoff can likely generate NPS pollutant concentrations comparable to those observed during summer monsoon runoff events.

Snowmelt runoff from highland agricultural fields during winter is not merely a hydrological phenomenon, but also a major source of NPS pollution, substantially increasing pollutant concentrations during seasonal transitions. When pre-cultivation management and mitigation measures in agricultural fields are insufficient, snowmelt runoff may lead to transient pollution events with higher concentrations than those observed during rainfall periods. Therefore, it is necessary to establish tailored NPS pollution management strategies that account for seasonal runoff characteristics, including those associated with snowmelt.

4. Conclusions

In this study, we empirically analyzed NPS pollutant runoff during snowmelt and evaluated the effectiveness of BMPs in highland agricultural areas. Pollutant concentrations varied depending on the snowmelt pattern, with particulate pollutants dominating during rapid thaw events and dissolved forms becoming more prominent under prolonged and repeated melting conditions. These findings highlight the need for snowmelt-specific water-quality management strategies that are tailored to the seasonal dynamics of pollutant forms and hydrological processes.

Snowmelt runoff can lead to NPS pollutant concentrations comparable to or exceeding those observed during summer rainfall events. This highlights the importance of recognizing snowmelt as a critical driver of seasonal water-quality deterioration and integrating it into rainfall-centered NPS management frameworks. Spatial comparisons among the monitoring sites showed that BMPs, such as straw mulching and gabion walls, contributed to reductions in pollutant levels. However, field observations have indicated that their effectiveness is substantially reduced in the absence of regular maintenance. These results suggest that, in addition to the installation of BMPs, continuous inspection and site-specific management systems are essential to ensure long-term functionality and pollution control.

Future research should focus on installing additional field flowmeters and meteorological observation equipment to analyze the relationship between runoff volume and water quality. These data can form a foundation for estimating pollutant loads and improving the input parameters for water-quality modeling. Moreover, future studies should consider fractionating TOC into DOC and POC components through direct measurement, in order to improve understanding of carbon dynamics under various snowmelt scenarios. The findings have key academic and policy implications and can form the basis for developing snowmelt period water-quality management policies and mitigation technologies.