Groundwater Response to Snowmelt Infiltration in Seasonal Frozen Soil Areas: Site Monitoring and Numerical Simulation

Abstract

Spring snowmelt has a significant impact on the hydrological cycle in seasonally frozen soil areas. However, scholars hold differing, and even opposing, views on the role of snowmelt during the thawing period in groundwater recharge. To explore the potential recharge effects of spring snowmelt on groundwater in seasonal frozen soil areas, this study investigated the vadose zone dynamics controlled by soil freeze–thaw processes and snowmelt infiltration in the Northeast of China for 194 days from 31 October 2020 to 12 May 2021. Responses of groundwater level and soil moisture to snowmelt infiltration show that most snowmelt was infiltrated under the site despite the ground being frozen. During the unstable thawing period, surface snow had already melted, and preferential flow in frozen soil enabled the recharge groundwater by snowmelt (rainfall), resulting in a significant rise in groundwater levels within a short time. The calculated and simulated snowmelt (rainfall) infiltration coefficient revealed that during the spring snowmelt period, the recharge capacity of snowmelt or rainfall to groundwater at the site is 3.2 times during the stable thawing period and 4.5 times during the non-freezing period.

1. Introduction

Seasonally frozen soil is about 50.5% of the exposed lands in the earth’s Northern Hemisphere [1]. In seasonally frozen soil areas, water is stored as snow over the winter and rereleased into the environment during snowmelt in spring, serving as an essential component of the hydrological cycle [2]. Snowmelt water is a crucial water resource in seasonally frozen ground regions, with the portion that infiltrates and enhances soil moisture being critical to alleviating spring drought issues in arid and semi-arid areas [3]. Furthermore, compared to surface runoff, the infiltration of snowmelt water more readily facilitates the rapid transport of pollutants absorbed by snow from the surface to groundwater and deeper soils [4]. Spring snowmelt infiltration plays a significant role in formulating spring irrigation plans and controlling the diffusion of pollutants, with the amount of snowmelt infiltration being an essential reference indicator.

Spring snowmelt water, including rainfall, may infiltrate the frozen soils or generate runoff that accumulates in surface water bodies such as rivers, ponds, and depressions. Therefore, during the snowmelt period, the infiltration capacity of frozen soil plays a crucial role in determining the fate of the snowmelt water. There are three main categories of infiltration in frozen soils [5]: (1) Restricted: The pores in the soil filled with ice lenses result in minimal infiltration of snowmelt, primarily leading to surface runoff [6]. (2) Limited: The frozen layer hinders snowmelt infiltration, resulting in a reduced infiltration rate compared to the non-frozen period. The extent of this reduction depends on factors such as the initial soil moisture content before freezing and the soil temperature during infiltration [2,7,8,9,10,11]. (3) Unlimited: During the snowmelt period, atmospheric pores or cracks in the frozen soil create preferential flow pathways, significantly increasing the infiltration rate compared to the non-frozen period [12,13,14,15]. Current research on snowmelt infiltration primarily focuses on the freeze–thaw processes of frozen soil. The main topics include changes in frozen soil permeability and influencing factors, mechanisms of moisture infiltration under freeze–thaw conditions, and snowmelt-induced runoff. However, there is little research on the relationship between snowmelt infiltration and groundwater resources [8]. Based on the site investigations, there remain controversial and contradictory hydrological responses to snowmelt infiltration in seasonal frozen soil areas [12]. Some studies show that soil frost has an insignificant effect on the infiltration of snowmelt water [15,16,17], while others suggest that it reduces the groundwater recharge of snowmelt water [10,18]. Additionally, some studies propose that soil meltwater during the thawing period is a significant factor contributing to the rise in groundwater levels [19,20]. Under the backdrop of global climate warming, regional responses to climate change vary, and the distribution of winter snowfall is expected to become increasingly uneven [21,22]. There remains a lack of understanding of groundwater recharge processes relating to spring snowmelt infiltration in cold regions under existing and future climate conditions [6,21,22]. More insight is therefore needed into the recharge effects of spring snowmelt infiltration on groundwater in seasonal frozen soil areas.

It is worth mentioning that snowmelt events and freezing (or thawing) periods are usually used as the introductory primary period to calculate the groundwater recharge of snowmelt water [10,17,19,23]. For the actual exploitation and utilization of groundwater, considering the water budget within cold areas, the former is frequent and complex; the latter is broad and imprecise. A suitable method is needed to delineate the various stages of the freeze–thaw process, serving as the foundation for calculating and comparing snowmelt infiltration amounts.

The primary objective of this study was to characterize the soil freeze–thaw processes controlling snowmelt water infiltration and, based on freeze–thaw period division, evaluate the recharge effects of spring snowmelt infiltration by water balance and the numerical simulation method. The secondary objective was to examine the possible mechanism of spring snowmelt infiltration affecting the timing and magnitude of groundwater recharge in seasonal frozen soil areas.

2. Materials and Methods

2.1. Study Region

The perennial site (125°18′24″ E, 43°52′59″ N), located in the Chaoyang Campus of Jilin University in central Changchun, northern China, was included in this study (Figure 1). The semi-humid climate is typical of the site: cold winters, annual precipitation (591 mm), and high evaporation in spring (Figure 2). The maximum frost depth is approximately 1.69 m, normally freezing in mid-November and thawing in mid-March. The groundwater at the site is classified as Quaternary terrace alluvial-lacustrine loess-like silty clay pore water. The aquifer system, composed of gravel and loess-like silty clay, has good water storage capacity. During the study period, the average groundwater depth was 2.7 m. Groundwater flows from southwest to northeast with a hydraulic gradient of 0.005, and the lateral runoff exchange capacity is relatively weak. Groundwater recharge is primarily derived from atmospheric precipitation infiltration and lateral inflow, while groundwater discharge mainly occurs through evaporation and lateral outflow.

The study focused on eight observation wells: C1, C2, C3, C4, C5, C6, CH1, and CH2 (Figure 1). At the time of the study, the land cover consisted of grassland and shrubs. The soil at the site is classified as a miscellaneous fill with a thickness of 0.03 m, underlain by gravel and loess-like silty clay ranging from 0.03 m to 30 m. This study emphasizes monitoring surface and subsurface hydrological conditions in the eight monitoring wells during the winter and spring of 2020–2021.

2.2. Methods

2.2.1. Meteorological and Land Surface Measurements

Daily temperature and precipitation data at the site were collected from the China Meteorological Data Network. Visual monitoring methods were employed to assess snow melting.

2.2.2. Vadose Zone and Groundwater Monitoring

Vadose zone conditions were monitored using an instrumented monitoring system (PICO-BT, IMKO, Ettlingen, Germany), which recorded soil moisture and temperature at depths of 0.5 m and 1.5 m below the surface in the central area of the site (Figure 3). The groundwater levels of the eight observation wells were measured once a day for a total of 194 measurements.

2.2.3. Specific Yield and Porosity Measurements

A two-step pumping test was conducted at the site from 21–26 September 2020, to measure specific yield. CH1 served as the pumping well, while C1 and C4 were used as observation wells to monitor groundwater levels. Calculations were performed using the Thiem formula method. The Boulton model was used to calculate the hydrogeological parameters at the site. The specific yield was 0.03, the hydraulic conductivity was 0.5 m/d, and the area of the study site (F) was 872.74 m². Prior to the start of the study, the site’s porosity was measured using the cutting-ring method.

2.2.4. Evaluation of Snowmelt (Rainfall) Infiltration Coefficient

(1)

Water Balance Method

Water balance calculations were used to assess the significance of snowmelt infiltration in groundwater recharge. The infiltration processes that were relatively evident during the study were utilized. The schematic diagram is shown in Figure 4.

The calculation formula is as follows:

$$SyF{\displaystyle \frac{\mathsf{\Delta }h}{\mathsf{\Delta }t}}=Q_{in}-Q_{out}+Q_{snowmelt}+Q_{rain}$$

(1)

where Sy represents the specific yield of the aquifer, F is the area of the study site (m²), $\mathsf{\Delta }h$ denotes the groundwater level change value (m), $\mathsf{\Delta }t$ is time (d), $Q_{in}$ is the lateral runoff inflow (m³/d), $Q_{out}$ is the lateral runoff outflow (m³/d), $Q_{snowmelt}$ is the groundwater recharge by snowmelt infiltration (m³/d), and $Q_{rain}$ is the groundwater recharge by rainfall infiltration (m³/d).

The water balance method selects distinct periods of groundwater level rise caused by snowmelt (rainfall) events, which are relatively short. The surface fluctuation at the site is minimal, the aquifer medium is homogeneous loam, the groundwater hydraulic slope is minimal, and the groundwater lateral runoff is slow, allowing us to consider that $Q_{in}=Q_{out}$.

$$SyF{\displaystyle \frac{\mathsf{\Delta }h}{\mathsf{\Delta }t}}=Q_{snowmelt}+Q_{rain}$$

(2)

For the stage of snowmelt infiltration only, Equation (2) can be further simplified to Equation (3).

$$Q_{snowmelt}=SyF{\displaystyle \frac{\mathsf{\Delta }h}{\mathsf{\Delta }t}}$$

(3)

Equation (4) is used to calculate the amount of snowmelt water (rainwater) on the site.

$$Q_{precipitation}={\displaystyle \frac{XF}{\mathsf{\Delta }t}}$$

(4)

where $Q_{precipitation}$ is the snowmelt (rainfall) on the site (m³/d) and X is the amount of snowmelt water (rainwater) (m).

Equation (5) is used to calculate the snowmelt (rainfall) infiltration coefficient based on the calculation of groundwater recharge from snowmelt and rainfall infiltration.

$$\alpha ={\displaystyle \frac{Q_{snowmelt\left(rainfall\right)}}{Q_{precipitation}}}={\displaystyle \frac{sy·\mathsf{\Delta }h}{X}}$$

(5)

where α is the snowmelt (rainfall) infiltration coefficient.

(2)

Numerical Simulation Method

MODFLOW is a physically based groundwater model that utilizes a finite difference method to solve Darcy’s equation for groundwater flow and is commonly used to simulate soil infiltration and groundwater level fluctuations caused by snowmelt and rainfall in cold regions [24,25,26]. Visual MODFLOW 4.6 software discretized the study area (Figure 5). The modeling period was from 31 October 2020 to 12 May 2021, totaling 194 days. The aquifer is generalized as a single layer since the study area is small and the lithology remains constant. There are no pumping wells at the study site. Under conditions of no anthropogenic influence, the vertical flow of groundwater is ignored, and only the horizontal flow is considered, representing a two-dimensional unstable flow. To more accurately generalize groundwater’s lateral inflow and outflow, the model boundary is simplified the specific head boundary. The initial groundwater flow field is shown in Figure 6.

The calculation formula is as follows:

$$\alpha ={\displaystyle \frac{w_{Infiltration}}{w_{precipitation}}}$$

(6)

$$w_{precipitation}=XF$$

(7)

where α is the snowmelt (rainfall) infiltration coefficient, $w_{Infiltration}$ is the infiltration of the snowmelt (rainfall) on the site in the one stage of freeze–thaw cycle (m³), and $w_{precipitation}$ is the snowmelt (rainfall) on the site in the one stage of freeze–thaw cycle (m³).

3. Results

3.1. Variation Characteristics of Temperature and Division of Freeze–Thaw Period

The stages of the freeze–thaw period at the site are classified based on air temperature and soil temperature states, with air temperature determining the start point of the freezing process and ground temperature defining the endpoint of the thawing process. As shown in Figure 7, these stages can be divided into five categories: (1) unstable freezing period (early November–18 November), (2) stable freezing period (19 November–11 February), (3) unstable thawing period (12 February–21 March), (4) stable thawing period (22 March–11 April), and (5) non-freezing period (after 12 April).

The characteristics of each period are as follows:

(1)

Unstable freezing period

At the beginning of November 2020, the air temperature intermittently dropped below 0 °C (Figure 7). The soil temperature continued to decrease, and unstable freezing was observed on the land surface.

(2)

Stable freezing period

After 19 November, the air temperature remained continuously below 0 °C (see Figure 7, point A). The rate of decrease in soil temperature at 0.5 m increased and was significantly higher than that at 1.5 m. The soil temperature at 0.5 m fell below 0 °C on 3 January (Figure 7, point B) and reached a minimum value of −1.6 °C (Figure 7, point C) on 20 January, after which it began to rise slowly. Meanwhile, the soil temperature at 1.5 m continued to decline.

(3)

Unstable thawing period

On 11 February, the air temperature intermittently rose above 0 °C (Figure 7, point D), indicating the onset of the unstable thawing process. The soil temperature at 0.5 m increased to 0 °C and stabilized between 0 °C and 0.5 °C after 28 February (Figure 7, point E). Thus, two inferences can be drawn: (1) There may still be frozen soil below 0.5 m that has not completely melted, but most of the frozen soil has thawed during this period; (2) Refreezing of infiltrated water may occur in the frost zone.

(4)

Stable thawing period

After 22 March, the daily minimum temperature stabilized above 0 °C (Figure 7, point F), meaning there were no conditions for refreezing at the soil surface. The soil temperature at 0.5 m entered the range above 0 °C after 2 March, maintaining around 0 °C until 11 April (Figure 7, point G), then rapidly rising from 0.7 °C to 1.2 °C. The air temperature did not significantly increase during this time, suggesting that the residual frozen soil below 0.5 m had melted.

(5)

Non-freezing period

After 12 April, the air temperature remained above 5 °C. The soil temperature at 0.5 m continued to rise, surpassing that at 1.5 m after 19 April (Figure 7, point H). At this time, the soil temperature at 1.5 m began to rapidly increase from the low-value range (3.3–3.5 °C) and started to respond to the air temperature.

3.2. Variation Characteristics of Soil Moisture

The soil moisture at 0.5 m showed significant changes during the freeze–thaw period, but the infiltration process did not elicit a response in soil moisture at 1.5 m (Figure 8).

3.2.1. Soil Moisture at 0.5 m

The characteristics of soil moisture at 0.5 m in each period are as follows:

(1)

Unstable freezing period

During the unstable freezing period, the daily maximum temperature often remained above 0 °C, and the frozen soil had not formed steadily, resulting in no influence on the soil moisture at 0.5 m. The soil moisture remained stable between 28.5% and 28.6% before 19 November. The precipitation amounts from 17–20 November were 0.4 mm, 6 mm, 25.9 mm, and 2 mm, respectively (Figure 8). Subsequently, the soil moisture increased on 19 November, reaching a maximum value of 29.67%, followed by a slight decrease. The response was rapid, with no time delay observed between precipitation and the corresponding rise in soil moisture at the site (Figure 8, point A; Figure 9).

(2)

Stable freezing period

Soil moisture significantly decreased after 29 December, with temperatures approaching 0 °C (Figure 8, point B), indicating that soil freezing effects caused the reduction in soil moisture. Liquid water in soil pores froze into ice, resulting in a rapid decrease in soil moisture.

(3)

Unstable thawing period

During this period, soil moisture remained relatively stable. The soil moisture at 0.5 m responded to two warm spells. The average daily temperature from 12–14 February was above 0 °C (the daily maximum temperatures reached 6.9 °C, 3.9 °C, and 2.1 °C, respectively). Correspondingly, the soil moisture at 0.5 m increased slightly. A two-day delay was observed between the maximum air temperature and the peak soil moisture (Figure 10).

The average daily temperature from 26–28 February was above 0 °C, with a maximum temperature reaching 8.9 °C on 28 February (Figure 8, point C). Simultaneously, a significant snowpack depletion occurred at the site during this warm spell. During this process, soil moisture increased at depths up to 0.5 m, reaching a maximum value of 20.55% on 3 March (Figure 8, point D). A time delay of three days was observed from the maximum air temperature to the peak soil moisture.

The melting of frozen soil was not observed due to the short duration of both warming events. The rise in soil moisture measured at 0.5 m is interpreted as a response to snowmelt resulting from high daytime temperatures. These soil moisture responses at 0.5 m occurred after the soil temperature reached 0 °C, indicating that pore ice was still present during infiltration. The observed time delays between maximum air temperature and peak soil moisture suggest the low infiltrability of the frozen soil.

(4)

Stable thawing period

Soil moisture increased before the soil temperatures responded, as latent heat transfer prevented soil temperatures from rising above 0 °C until all pore ice had melted entirely (Figure 8, point F). Once the pore ice in the soil was fully melted, the soil moisture could return to the levels before the soil froze (Figure 8, point E).

(5)

Non-freezing period

Soil temperature rose rapidly when the frozen soil at a depth of 0.5 m was completely melted. Meanwhile, soil moisture exhibited a continuous decreasing trend over time due to the effects of infiltration and evaporation.

3.2.2. Soil Moisture at 1.5 m

During the entire winter and spring snowmelt period, no response of soil moisture at 1.5 m to temperature and precipitation was observed at the site (Figure 8). In contrast, the variation of soil moisture at 0.5 m exhibited complex dynamic characteristics. The soil at 1.5 m did not freeze during the winter because the soil temperature remained between 3.3 °C and 3.5 °C. Located near the groundwater surface within the capillary zone, the soil at 1.5 m reached a saturation state with an average moisture content of 32.4%, essentially equivalent to silty clay’s porosity.

3.3. Response Characteristics of Groundwater Level

From early November 2020 to May 2021, the response characteristics of groundwater level can be divided into five stages (Figure 11).

(1)

Stable period (31 October–17 December 2020)

This stage occurs during the unstable and early stable freezing periods, where the groundwater level exhibits a relatively stable state. Following precipitation on 19 November, the groundwater level and soil moisture increased, reaching their maximum values (Figure 9; Figure 11, point A). No time delay was observed, indicating that the groundwater level at the site responded very quickly to precipitation during the unstable freezing period when frozen soil had not yet formed (Figure 9).

(2)

Decline period (18 December 2020–4 March 2021)

This stage occurs during the stable freezing period, with an increasing depth of frozen soil. The groundwater level continued to decline until it reached its minimum value on 4 March 2021 (Figure 11, point B). Due to soil freezing, groundwater migrates upward driven by the water potential gradient, leading to a decline in groundwater levels. There was also a slight rise in the groundwater level from 17–25 January, which may have been related to snowmelt infiltration resulting from a brief temperature increase above 0 °C. This phenomenon indicates that the frozen soil in the vadose zone still has infiltrability.

(3)

Rapid rising period (5–19 March 2021)

The average daily temperature was above 0 °C from 26–28 February, reaching a maximum of 8.9 °C on 28 February. The groundwater level at the site began to rise sharply on 5 March, reaching a maximum value of 29.6 cm on 9 March (Figure 11). At this time, the soil temperature at 0.5 m remained close to 0 °C, suggesting frozen soil was still present when the groundwater level rose. During this period, visual monitoring of surface snow indicated complete snowpack depletion at the site. This sharp rise is interpreted as a response to spring snowmelt infiltration, which accomplished groundwater recharge before the complete thawing of frozen soil. Following this sharp rise, a noticeable recession of groundwater was observed, indicating that the recharge process was brief. Due to the site being located within an urban area surrounded by impermeable artificial surfaces, the recharge of groundwater from snowmelt water can be approximated as point-source recharge. The predominant groundwater flow is outflow from the site, resulting in a noticeable decline of groundwater after a sharp rise.

(4)

Slow rising period (20 March–19 April 2021)

The air temperature gradually increased, and the frozen soil continued to melt. The groundwater level rose slowly, reaching its maximum value on 19 April (Figure 11, point C) due to the infiltration of frozen soil water and precipitation.

(5)

Decline period (after 20 April)

The groundwater level declined again as the air temperature rose, accompanied by evaporation and regional groundwater runoff following the complete thawing of the soil. In this stage, the response speed of the groundwater level to precipitation was rapid, but the response range was relatively small.

3.4. Evaluation of Snowmelt (Rainfall) Infiltration Coefficient

3.4.1. Water Balance Method

The average value of the groundwater level across all observation wells is used to calculate the groundwater level change. The water balance is not calculated for the non-freezing period, which experiences high evaporation (Figure 2), as this would be influenced by evaporation. Four relatively distinct infiltration recharge periods for groundwater during snowmelt (rainfall) events have been identified (Figure 12). The components of the water balance for these snowmelt (rainfall) events are presented in

Table 1. Calculation results of snowmelt (rainfall) infiltration coefficient.

Time	Period	Water Level Rise/mm	Accumulated Precipitation/mm	$\mathit{s}\mathit{y}·\mathsf{\Delta }\mathit{h}$ /mm	Snowmelt (Rainfall) Infiltration Coefficient
15 November 2020–22 November 2020	unstable freezing period	135	34.3	4.05	0.12
16 January 2021–26 January 2021	stable freezing period	105	9	3.15	0.35
4 March 2021–9 March 2021	unstable thawing period	341.25	12.8	10.24	0.80
19 March 2021–23 March 2021	stable thawing period	65	7.8	1.95	0.25

.

The snowmelt (rainfall) infiltration coefficient during the unstable thawing period is 3.2 times that during the stable thawing period.

During the stable freezing period, the daily maximum temperature briefly exceeded 0 °C during the rising water level (Figure 8), leading to a winter snowmelt event. At this time, evaporation (Figure 2) and the amount of snowmelt water were minimal, resulting in low infiltration rates but a high infiltration coefficient.

During the stable thawing period, precipitation was only 7.8 mm during this stage. On the other hand, the average evaporation in March and April ranged from 100 mm to 200 mm (Figure 2).

During the unstable freezing period, the accumulated precipitation from 17–20 November 2020 was 34.3 mm, with a single-day precipitation peak of 25.9 mm on 19 November (Figure 8). Over a short period, this heavy rainfall led to exceed-infiltration runoff, with part of the rainwater leaving the site as surface runoff. Consequently, the snowmelt (rainfall) infiltration coefficient was relatively low.

3.4.2. Numerical Simulation Method

For the model, the bias between the observed and simulated groundwater levels was 1.3 cm, with a root mean square error (RMSE) of 0.061. Figure 13 displays the results of the simulated and observed groundwater levels. Infiltration recharge in the model was represented using the snowmelt (rainfall) infiltration coefficient as a model parameter, which was validated against the numerical model simulation results. The simulated groundwater levels showed good agreement with the observed levels.

Groundwater recharge from rainfall or snowmelt infiltration during each period was calculated using the model. The infiltration amounts, precipitation, and snowmelt (rainfall) infiltration coefficients are presented in

Table 2. Calculation results of snowmelt (rainfall) infiltration coefficient.

Time	Period	Accumulated Precipitation/mm	${\mathit{w}}_{\mathit{p}\mathit{r}\mathit{e}\mathit{c}\mathit{i}\mathit{p}\mathit{i}\mathit{t}\mathit{a}\mathit{t}\mathit{i}\mathit{o}\mathit{n}}$ /m3	${\mathit{w}}_{\mathit{i}\mathit{n}\mathit{f}\mathit{i}\mathit{l}\mathit{t}\mathit{r}\mathit{a}\mathit{t}\mathit{i}\mathit{o}\mathit{n}}$ /m3	Snowmelt (Rainfall) Infiltration Coefficient
31 October 2020–18 November 2020	unstable freezing period	8.6	7.506	0.808	0.11
19 November 2020–11 February 2021	stable freezing period	43.1	37.615	10.542	0.28
12 February 2021–22 March 2021	unstable thawing period	14.8	12.917	6.937	0.54
23 March 2021–11 April 2021	stable thawing period	0.1	0.087	/	/
12 April 2021–12 May 2021	non-freezing period	24.8	21.644	2.524	0.12

.

The results of the numerical model and water balance method are broadly consistent, indicating that the lateral comparison results across the stages are reliable. The snowmelt (rainfall) infiltration coefficient during the unstable thawing period is 4.5 times during the non-freezing period. The snowmelt (rainfall) infiltration coefficients in each period exhibit the following relationship: unstable thawing period > stable freezing period > stable thawing period > non-freezing period > unstable freezing period.

Notably, discharge was significantly greater during the stable freezing period than recharge (

Table 3. Calculation results of Visual MODFLOW-Zone budget module.

Time	Period	Infiltration /m3	Groundwater Evaporation/m3	Inflow /m3	Outflow /m3	Variation of Groundwater Storage/m3
31 October 2020–18 November 2020	unstable freezing period	0.8	0	33.31	30.69	2.7
19 November 2020–11 February 2021	stable freezing period	10.5	0	97.03	128.87	−18.92
12 February 2021–22 March 2021	unstable thawing period	6.937	0	33.83	49.75	−7.27
23 March 2021–11 April 2021	stable thawing period	/	0	23.95	20.28	2.97
12 April 2021–12 May 2021	non-freezing period	2.52	22.02	34.52	20.51	−5.48

). This indicates that the numerical simulation results align with actual conditions, as the groundwater level consistently dropped during this period according to the measured data from observation wells. The continuous increase in frozen soil thickness resulted in upward groundwater movement and a downward shift in the water table. However, the numerical model cannot capture this water level change, and the upward movement of water is calculated as groundwater discharge. Additionally, during the unstable thawing period, the substantial recharge of groundwater from meltwater results in outflow being higher than inflow.

4. Discussion

4.1. Recharge Effects of Spring Snowmelt Infiltration on Groundwater

The calculations (Table 1) and simulations (Table 2) indicate that during the unstable thawing period, spring snowmelt (and rainfall) infiltration is more effective at recharging groundwater than rainfall during the non-freezing period, with snowmelt water being the primary component of this infiltration. This finding aligns with other studies concluding that a significant portion of annual snowmelt contributes to groundwater recharge [16,27,28,29,30]. However, the relatively short observation time in the non-freezing period in this study may result in an underestimation of the snowmelt (and rainfall) infiltration coefficient.

A likely explanation for this result is the low evaporation during the spring snowmelt period (Figure 2). Snowmelt occurs when vegetation is primarily dormant and temperatures are low, minimizing evapotranspiration [30]. Thus, this initial recharge pulse through the frozen ground may escape the effects of evapotranspiration and shallow groundwater cycling, contributing disproportionately to deeper recharge [23]. In contrast, evapotranspiration peaks during non-freezing due to higher temperatures and active vegetation. As shown in Table 3, the groundwater level response to precipitation in the non-freezing period is weak because of significant evaporation and lateral groundwater runoff. The accumulation of snow over months can provide sufficient water for groundwater recharge during the snowmelt period, whereas intermittent rain events often fail to deliver enough moisture [31]. Snow penetrates the ground more quickly than rainwater because it melts gradually and releases water slowly. Additionally, rainfall infiltrating the snowpack penetrates the ground more readily [27].

4.2. Possible Mechanism of Spring Snowmelt Infiltration

The infiltrability of frozen soils significantly affects the quantity of water available for groundwater recharge at the site. Many studies suggest that frozen soil obstructs the infiltration of snowmelt water [2,10,13]. The infiltrability of frozen soils is strongly influenced by initial saturation [2,32,33] and the temperature of the frozen soil [34]. Comparing the initial saturation and temperature of the frozen soil here with other studies provides valuable insights. During the spring snowmelt period, the average soil temperature at 0.5 m at the site is −0.3 °C. The average soil water content at 0.5 m before freezing is 28.12% (22–25 December 2020), and the measured soil porosity at the site is 48.45%, leading to a calculated initial saturation of 0.58 before freezing. These values are lower than those reported by Stuurop et al., who found that cases of severe infiltration reduction due to freezing were generally associated with an initial saturation of at least 0.6 or 0.75, an absence of thaw (0 °C top boundary temperature), and a freezing temperature of at least −0.5 or −1 °C for five days [2]. Watanabe and Osada reported that frozen soil below −0.5 °C is practically impermeable [32]. These findings indicate that a certain amount of infiltration flow was maintained in the frozen soil when spring snowmelt infiltration occurred in this study, which is consistent with observations.

Several observations linked spring snowmelt infiltration to preferential flow through the frozen ground rather than infiltration flow in the frozen soil. The responses of groundwater level and soil moisture to snowmelt infiltration show that the majority of the snowmelt was infiltrated under site despite the ground being frozen/partially frozen, as evidenced by temperatures within the frost zone remaining at or below 0 °C during spring snowmelt event (Figure 7). Groundwater recharge events occurred when the soil in the frost zone was still partially frozen, suggesting that the ponding of water and filling of the unfrozen pore space can activate preferential flow pathways and cause the channeling of snowmelt to groundwater before the complete thawing of the soil profile [23].

Further evidence for a dominant role of preferential flow includes the relative independence of the infiltration rate from soil frost conditions. The average daily infiltration rate of spring snowmelt in this study is 68.25 mm/day, much more significant than that reported by Li et al. for fields in northeast China, who noted that the average daily maximum infiltration rate of the thawing period frozen soil is approximately 14.4 mm/day [3]. The likely explanation for this inconsistency is that conventional tillage in fields destroyed shallow macropore networks that enable preferential flow through the frost zone [35].

The development of root holes and biopores in the soil due to the undisturbed growth of grass on the site enhances soil infiltrability, particularly during the snowmelt period when the soil is frozen [13,36]. Increased permeability may occur as the soil shrinks or cracks after the freeze–thaw cycle, potentially resulting in preferential flow [37]. Both of these macropore network pathways can facilitate preferential mass transport to groundwater. In conclusion, the preferential flow in frozen soil allows spring snowmelt water to recharge groundwater before soil thawing, leading to the observations made in this study.

4.3. Limitations

The spatial scale of this study is too small, which may have led to the insufficient consideration of horizontal flow. Strictly speaking, even for very short calculation periods, there is still horizontal flow into and out at the site.

5. Conclusions

This study aimed to characterize the soil freeze–thaw processes associated with snowmelt, highlighting the recharge effects of spring snowmelt infiltration and analyzing the potential recharge mechanisms at the site. Based on the characteristics of air and soil temperatures, the freeze–thaw period in seasonal frozen soil areas can be divided into five stages: unstable freezing period, stable freezing period, unstable thawing period, stable thawing period, and non-freezing period. Field data suggest that despite the soil remaining frozen throughout the spring snowmelt event, infiltration into frozen soil is the primary sink for spring snowmelt water. The preferential flow within frozen soil may allow snowmelt water to recharge groundwater before the ground thaws, leading to a sharp rise in groundwater levels. This is consistent with the research findings of Mohammed [23], which suggest that preferential flow in frozen soil enabled meltwater to bypass a portion of the soil profile and groundwater recharge prior to ground thaw. Calculations and simulations indicate that spring snowmelt (and rainfall) infiltration during the unstable thawing period, where snowmelt water is the main component, is more effective at recharge groundwater than rainfall during the stable thawing period and the non-freezing period. However, due to a lack of observations in summer and autumn, complete information on possible groundwater recharge during the different periods is not available. Long-term research on the response characteristics of groundwater to spring snowmelt infiltration and rainfall is necessary to understand better the recharge effects of spring snowmelt on groundwater in seasonal frozen soil regions.