冻融过程对半干旱地区农田水土流失的影响

A R T I C L E I N F O

Handling Editor - Dr. B.E. Clothier

Keywords: Farmland Semi-arid area

Freeze-thaw process

Soil water content (SWC) Soil temperature (ST)

A B S T R A C T

Seasonal freeze-thaw processes affect soil water migration and distribution, especially in semi-arid agricultural areas. These processes play an important role in mitigating harsh environmental conditions and wind erosion. Soil water content (SWC) and soil temperature (ST) were monitored at different depths (0–2 m) and investigated under freeze-thaw conditions from November 2018 to May 2019 in a semi-arid agro-pastoral region of northern China. The initial SWC was the main factor that affected the freeze-thaw process. During the freeze-thaw process, differences in soil thermal conductivity caused the soil to thaw faster than the freezing process, and the upper soil layer (0–60 cm) was significantly affected by temperature changes. Changes in the potential energy of water and pore pressure gradient caused the migration of soil water to the upper layer, which led to a slight decrease in SWC in each layer before ST dropped to the freezing point. The vertical migration distance of soil water exceeded 70 cm, and the SWC above a depth of 100 cm increased significantly, as water was mainly obtained from the soil layer below a depth of 200 cm. Soil compaction was reduced when affected by freeze-thaw processes and the soil particles were more fragmented, leading to wind erosion and dust events. Our results partially explain the occurrence of wind erosion in spring and provide a scientific basis for predicting soil water status and appropriate farmland management strategies.

1. Introduction

The freeze-thaw process in soil has an important impact on the agro- ecological environment (Wang et al., 2019). This process promotes the migration of soil water and changes the water distribution and complex hydrothermal coupling (Ala et al., 2016; Evett et al., 2012; St¨ahli et al., 2001). Especially in spring, freeze-thaw processes change soil structure and affect soil particle migration, which makes surface soil more prone to soil wind erosion (Ban et al., 2016). The change in upper soil water content (SWC) also influences the choice of irrigation strategy and resulting crop growth during spring (Dai et al., 2019). The migration of water and soil particles also changes carbon and nitrogen cycling (CN cycle) in the soil (D’Odorico et al., 2003). With the increased CN cycle interaction with soil erosion, some of the most important phases of these cycles, such as decomposition, leaching, and plant uptake are impacted (Porporato et al., 2003). Ice crystal growth within soil voids forces particles apart, and ice pressure may compress or rupture soil

aggregates, increasing the risk of soil erosion (Ferrick and Gatto, 2010), thereby affecting the microclimate and bringing about ecological changes (Huang et al., 2020).

In the process of soil freezing, the soil water freezes into ice, forming a freezing front below 0 ºC (Hou et al., 2020). The soil freezing causes the soil water to move from unfrozen areas to the freezing front (Nagare et al., 2012), which leads to a significant increase in SWC in the frozen soil. Furthermore, the SWC distribution before the soil freezes in early winter affects the soil water status in the following year (Luo et al., 2003), leading to an increase in field storage and a reduction in irriga- tion needs (Chen et al., 2013). The rise in spring temperature causes the upper soil to thaw first, while lower soil gradually thaws as the tem- perature increases. Hence, the SWC increase in the upper soil is more obvious (Yang et al., 2008), and can become an important water source for plants and crops, especially in an area with low precipitation. Accurately evaluating freeze-thaw processes in different soils and re- gions is a critical part of ecological science, agriculture, and forestry

* Corresponding author.

E-mail address: jggd33@163.com (G. Jia).

https://doi.org/10.1016/j.agwat.2021.106876

Received 15 November 2020; Received in revised form 18 March 2021; Accepted 19 March 2021

Available online 31 March 2021

0378-3774/© 2021 Elsevier B.V. All rights reserved.

management, as well as the development of management strategies for agricultural freeze areas. Many researchers have studied the mechanism by which SWC moves during the soil freeze-thaw process in different areas and with different conditions (Nagare, 2011; Wang et al., 2019). Evaluation of the effects of different SWC on the freeze-thaw process showed that SWC and soil properties had a great influence on freezing time and ground thermal condition. At the same time, liquid water and ice are in equilibrium under temperature changes due to the soil freezing, regardless of the amount of water in the ice phase. The content of liquid water may be relatively stable (Low et al., 1968).

In agricultural freeze zones, the amount of water that migrates up- ward during the freeze-thaw progress is controlled by the initial SWC, while a lower SWC makes the soil more susceptible to freezing (Chen et al., 2013). However, the migration characteristics of soil water under freeze-thaw conditions in semi-arid agricultural areas have not been explored. There is also less research on characteristics influencing soil water migration, such as temperature change and related spatio-temporal conditions on agricultural soil wind erosion events (Wang et al., 2019). In China, seasonal freeze-thaw events mostly occur in arid and semi-arid regions (Chen et al., 2013), which have a crucial impact on the growth of crops. However, large-scale soil degradation has caused serious problems including wind erosion and air pollution on farmland (Sun et al., 2018, 2019; Chang et al., 2019), as well as negative impacts on the regional environment. Low precipitation and significant seasonal changes as well as low SWC cause wind-induced soil erosion and low crop yield (Huang et al., 2020; Sun et al., 2019). Previous research indicated that the groundwater level is declining annually (Nakayama et al., 2010; Sun et al., 2018). This phenomenon may cause

the area to be in an extended dry state, causing severe wind erosion on farmland. However, there is still a lack of understanding of the freeze-thaw process and soil water dynamics in this area. To address these gaps, this research studied the characteristics of ST and water migration during the soil freeze-thaw process on farmland in a semi-arid region to provide a reference for regional farmland management and wind-induced soil erosion prevention and control.

2. Materials and methods

2.1. Study site

Research was conducted in Zhangbei Country (40◦01′ 42◦17′ N, 114◦05′ 115◦20′ E), which is located in a semi-arid agricultural freeze- thaw area in north China (Fig. 1a). The annual precipitation is 320 mm and the altitude is about 1750 m. Precipitation is temporally concen- trated, with about 65.4% of precipitation occurring from June to

September (Fig. 2). There is a large difference between daytime and nighttime temperatures. Annual average temperature increased every year of the study, with values ranging between 1.8 and 5.2 ◦C (R2 0.53, p < 0.01), and they increased significantly from 1996 to 2006. The

annual average wind speed is 6 m/s. Windy weather usually occurs

every spring, especially in April. In the past 60 years, the average rela- tive humidity was 57.1%. Winter is cold and lasts up to five months (November to March), with average temperatures reaching 16 ℃. The soil depth ranges to 450 cm, and is mainly composed of sandy soil. The main forest species is Populus simonii Carr (Fig. 1b), which were first planted in the 1960 s to prevent wind-induced soil erosion.

Fig. 1. Map showing the study location; (a) Zhangbei County; and (b) indicates farmland wind-based soil erosion.

Fig. 2. Climate diagram for Zhangbei County covering the period from 1980 to 2018.

2.2. Freeze-thaw process monitoring

The root systems of the main crops and plants were distributed in the range of 0–200 cm in the study area (Liu et al., 2020), leading us to investigate the 0–200 cm soil profile. Environmental data was measured simultaneously during the freeze-thaw process using a standard auto- matic weather station located in an open space 20 m away from the study site. Precipitation (mm), temperature (℃) and relative humidity (%) were measured at 15-minute intervals. Data logging used a battery-powered HOBO U30 NRC data logger (ONSET Computer Cor- poration, USA), which is charged daily using a photovoltaic power source.

Freeze-thaw processes were measured 10 m away from the farmland. The ST (℃) and SWC (%) were measured at 15-minute intervals through ten 5TE probes (Decagon, USA) placed at ten soil depths (10, 30, 50, 70, 90, 110, 130, 150, 170 and 190 cm) in the study site. Data logging used a battery-powered EM50 data logger (Decagon, USA). The state of the soil was based on the ST: when the mean daily ST was below 0 ℃ (Guo et al., 2011a), the soil began to freeze, otherwise it starts to thaw. Measure- ments were taken from 1st Nov. 2018–26th May 2019. Four soil stages

were defined, which included freezing (soil was in the process of freezing), completely frozen (daily maximum soil temperature was < 0 ◦C), thawing (soil was in the process of thawing), and completely thawed (daily minimum soil temperature was >0 ◦C) (Guo et al., 2011b). The freezing point represents a soil temperature of 0 ◦C.

One-way analysis of variance (ANOVA) and the least significant method (LSD) were used to test the significance of the differences. The Pearson correlation coefficient was used to analyze the correlation be- tween air temperature and ST.

2.3. Thermal regime of soil profiles

The temperature transfer between different soil layers was analyzed using the soil heat transfer model proposed by Hillel (2003), which in- dicates the heat transfer process in shallow soil during the freeze-thaw process and reflects the time lag in heat transfer. The diurnal variation of soil temperature can be expressed by:

T(z, t) = T + A0 × sin(ωt — z/d)/ez/d

where T(z, t) is the temperature at z as a function of time t, T is the average temperature at z depth, and A0 is the amplitude of the surface temperature fluctuation (the range from maximum or minimum to average temperature). Finally, ω is the radial frequency. The constant d

is called the damping depth, which is related to the thermal properties of

the soil and the frequency of the temperature fluctuation (Hillel, 2003). This study assumed that surface soil temperature fluctuations were symmetrical.

3. Result

3.1. Stratification of initial soil characteristics

The initial SWC and ST profile are shown in Fig. 3. The lowest ST occurred at the surface of the soil, at 1.58 ℃. The ST reached a maximum value of 5.95 ℃ at 1 m depth and decreased significantly when soil depth exceeded 1 m. ST changes were not obvious within the range of 1–2 m, and ST maintained around 3 ℃. At the beginning of soil freezing, the ST at all depths was above 0 ℃. Each value was higher than the air temperature. The SWC also reached the maximum at 1 m depth, and SWC at 90 cm depth was nearly nine times higher than that at a depth of 110 cm. The SWC values at 110, 130, 150, 170 and 190 cm

were 2.02%, 1.98%, 3.38%, 1.58% and 2.12%, respectively.

3.2. The variation process of ST

Surface soil (0–20 cm depth) first started to freeze on 10th Nov. 2018 (Table 1). ST was about 10 ℃ at a depth of 180 cm. The upper soil froze earlier than the deep soil (Fig. 4b), and for every 20 cm increase in depth, soil freezing was delayed for about seven days (Table 1). The frozen depth reached 100 cm after 42 days and after 67 days for 200 cm. The freezing process was not steady once freezing was initiated, since the air temperature fluctuated. The air temperature was steady below 10 ◦C from 10th December, and there were two significant cooling events on 1st and 19th Dec. (Fig. 4a). During the freezing process, ST

fluctuated significantly between —15 and —10 ℃ in the upper profile (0–60 cm). The lowest ST was 18.55 ◦C at the surface layer and

2.95 ◦C at a depth of 200 cm.

Surface ST rose above 0 ℃ and soil started to thaw on 16th Mar. 2019 (Fig. 5). Temperature fluctuations in this period were monitored under near-freezing/thawing conditions. Soil at depths of 100 and 200 cm began to thaw after 18 and 26 days, respectively. The first full thaw of the soil occurred in the surface soil on 25th April. Compared with the surface soil, the total thawing time at 100 cm and 200 cm was delayed by 28 and 35 days, respectively (Table 1). Soils with a depth of 0–40 cm had greater temperature fluctuations during the thawing pro- cess, and ST fluctuations below 80 cm were not significantly affected by air temperature (Fig. 5). After the thawing progress, the ST at a depth of 100–200 cm was maintained at around 7 ℃. The average time from the start to full thawing was about 26 days, in which the upper soil

Table 1

Fig. 3. Initial soil temperature and soil water content levels for the study site in Zhangbei County.

amplitude of ST in each layer increased significantly (Fig. 7b).

Starting dates of the soil freezing–thawing stages in 2018–2019.

Compared with the time when the ST of the 0–20 cm depth peaked, the ST at 20–40 and 40–60 cm depths lagged by about 4 and 10 h, respec-

Soil

depth

Freezing Completely

frozen

Thawing Completely

thawed

Total

thawing

tively. During the thawing period, the amplitude was smaller than that

of completely frozen stage (Fig. 7c), but it was still larger than that of the

(cm) (month/ day)

(month/day) (month/

day)

(month/day)

time (day)

freezing stage. Compared with depths of 0–20 cm, the peak time of the ST for the 0–20 and 40–60 cm depths lagged by about 6 and 11.5 h,

0–20 11/10 12/31 3/16 4/25 41

20–40 11/19 12/31 4/7 4/25 19

40–60 11/24 12/31 4/15 4/27 13

60–80 12/8 2/5 4/17 4/30 14

80–100 12/13 2/18 4/22 5/23 32

100–120 12/21 2/20 4/20 5/26 37

120–140 12/28 2/23 4/25 5/28 34

140–160 1/3 2/23 4/29 5/27 29

160–180 1/10 2/24 4/27 5/28 32

180–200 1/11 2/27 4/30 5/30 31

(0–80 cm) thawed faster, with an average time of 21 days, and the 80–200 cm depth soil took about 30 days.

In addition, ST was positively correlated with air temperature, except from depths of 160–200 cm (Table 2). Compared with the upper soil, deep soil (below 60 cm) was less affected by air temperature (Fig. 5). Fig. 6 shows the dynamic response of surface ST (0–60 cm) to air temperature. Upper ST (0–60 cm) had a strong linear correlation with the air temperature (y = ax+b). Regression was able to explain changes of more than 68.5%. Coefficient ‘a’ represents the thermal ef- ficiency between air and soil in the equation, which changes with soil properties. Larger values of ‘a’ indicated that the heat transfer between ST and air temperature was faster. AT had a significant relationship with STs in the upper layers (Fig. 6; P＜0.05). The closer to the surface, the faster the ST responded to changes in air temperature.

The soil heat transfer model was used to simulate the change in shallow ST (0–60 cm), which showed sinusoidal fluctuations of ST in the shallow layer (Fig. 7), among which the fluctuations at depths from 0 to 40 cm were more apparent. When compared with the time when the ST of the 0–20 cm depth was at its peak, during the freezing stage ST at 20–40 cm depths lagged by about six hours, and ST at 40–60 cm depths lagged by about 13 h (Fig. 7a); During the completely frozen stage, the

respectively. During the completely thawed period, the amplitude in each layer decreased significantly (Fig. 7d), and the sinusoidal fluctua- tion at 40–60 cm depth was not apparent. The lag time during this period at the 20–40 cm layer was 8 h compared with that at the 0–20 cm depth.

3.3. The variation process of SWC

The changes in SWC at different soil layers are shown in Fig. 8. There were significant differences in SWC in each layer before the freezing process (Fig. 8b), with SWC at a 80–100 cm depth reaching a maximum of 18.99%. Affected by the two cooler-weather temperatures weathers on the 3rd and 5th of December (Fig. 8a), soils at 0–20 cm froze rapidly as air temperature dropped from 2.1 ◦C to 22.6 ◦C in a short period of time, then the frozen layer extended downward. From 10th January, the soil layer in the range of 0–200 cm was completely frozen. The SWC decreased significantly after the freezing front. Since there was a stable phase from freezing to thawing, this phase lasted longer in deeper soil (Fig. 8b). ST of the 0–20 cm depth fluctuated considerably from 16th March to 5th April 2019 (Fig. 8b), leading to a rapid thawing of the surface soil and a significant increase in SWC. During thawing, SWC at depths of 0–100 cm increased slightly compared with that before freezing, which indicated the upward migration of soil water. The SWC at 10, 30, 50, 70 and 90 cm depths increased by 85.7%, 123%, 51%, 43.7% and 29.2% respectively, but it had little effect on the deeper layers. The air temperature began to gradually increase after 5th April, and the SWC increased rapidly. Because there was no significant pre- cipitation in the non-growing season in the study area (Fig. 2; Sun et al.,

2019), SWC was less affected by precipitation.

The SWC within 0–100 cm depth changed significantly during the freeze-thaw process, and the SWC at the 90 cm depth was the highest

Fig. 4. The traces of the soil temperature in Zhangbei Country (10, 30, 50, 70, 90, 110, 130, 150, 170 and 190 cm depth; freeze-thaw progress from 1st Dec.

2018–26th May 2019).

Fig. 5. Dynamic variations in air temperature (a) and soil temperature (b) at depths of 10, 30, 50, 70, 90, 110, 130, 150, 170, and 190 cm throughout the experimental period in Zhangbei County.

during this period (Fig. 9a). SWC below 100 cm decreased significantly. After the freeze-thaw process, the SWC above 140 cm increased, with that at depths of 50 cm increasing the most, by 4.9 mm (Fig. 9b). The

SWC at a depth of 150 cm was reduced by 1.9 mm, while the SWC at 170 and 190 cm were almost unchanged. This phenomenon suggests that soil water at a depth of 150 cm may have moved up to a depth of

Table 2

Pearson's correlation and R2 values between air temperature and soil tempera- ture at each soil depth in Zhangbei County.

Soil depth (cm) Pearson's correlation coefficient R2

0–20 0.948** 0.899

20–40 0.900** 0.810

40–60 0.828** 0.685

60–80 0.727** 0.528

80–100 0.589** 0.347

100–120 0.496** 0.247

120–140 0.380** 0.144

140–160 0.247** 0.061

160–180 0.140 0.020

180–200 0.094 0.009

110–130 cm. During the entire freeze-thaw process, there was a signif- icant difference between the SWC at the 90 cm depth and that at the deeper soil, indicating that the soil water may have been transported vertically upward from deeper soil.

3.4. Freezing and thawing front

The development of the soil freezing and thawing front with depth varied over the study period (Fig. 10), as each layer of soil froze more slowly than it thawed. The freezing front penetrated to a depth of 60 cm for 13 days (12th-25th Nov. 2018; Fig. 10a). However, it took much longer for the thawing front to reach the same depth (23rd Mar. 13th Apr. 2019; Fig. 10b). Unlike the frozen front as it propagated from surface to deep soil in Fig. 8a, the propagation of the thawing front was interrupted at depths of 70, 110, and 170 cm (Fig. 10b). The duration of these melting fronts varied widely at depths of 110–170 cm (Fig. 10b), possibly due to differences in soil properties.

4. Discussion

4.1. Freeze-thaw characteristics of soil

The freeze-thaw process in soil is a complex change of chemical,

physical, and mechanical phenomena, which includes energy, water transfer, and salt accumulation (Li et al., 2008; Qi et al., 2018). Ac- cording to observed SWC and ST data, the freeze-thaw characteristics of soils at different depths were determined. The winter freeze-up stage began when the freezing process of the active layer finished, and it lasted approximately 200 days.

The freeze-thaw process had obvious stratification at a depth of 90 cm. Before freezing, within the depth range of 0–100 cm, SWC gradually increased with depth and the freezing process took place quickly (Fig. 7b). During the freezing process, the freezing peak slowly extended downward, and slowed when it reached a depth of 90 cm, related to the lower SWC at this depth. During the complete freezing stage, soil at depths from 80 to 100 cm maintained a relatively high SWC, which may be due to the release of heat during the solidification of the soil water at the stable freezing stage, leaving the soil in this layer not completely frozen. The above phenomena may be related to differ- ences in the initial SWC (Ala et al., 2016; Wang et al., 2019). Our research confirmed this view. The freeze-thaw processes of soil at different depths were significantly different due to differences in initial SWC.

The SWC below 90 cm was significantly reduced, as it had a greater relationship with soil texture (Arredondo et al., 2018). The SWC of unfrozen soil increased with the increase of clay content (Sta¨hli and Stadler, 1997), as soil particles gradually grow in size with increasing depth in the sedimentary cycle, and the deeper soil consists of coarser soil particles. For this reason, the SWC in upper soil may be higher than that of deeper soils. Furthermore, the water-repellent layer blocks the vertical transport of water and leads to insignificant changes in soil during the freeze-thaw process (Wei et al., 2016).

The transfer rate of soil temperature is related to the thermal con- ductivity. Some studies have pointed out that different soil thermal conductivity is mainly due to SWC (Yi et al., 2014). Since the thermal conductivity of ice is about four times higher than that of liquid water (Campbell, 1985), a lower SWC may make the soil have a higher thermal conductivity. In the soil freezing stage, higher SWC will slow down the freeze-thaw process, and the freeze and thaw time will be extended, as other studies have shown (Wang et al., 2019). When the soil begins to thaw, the ice in frozen soil promotes heat transfer, which confirms that

Fig. 6. Correlation between air temperature and upper soil temperature (0–20, 20–40, and 40–60 cm) in Zhangbei County.

Fig. 7. Diurnal variation of ST at different freeze-thaw stages.

the thawing rate of the soil is greater than the freezing rate in our study.

4.2. The freeze-thaw cycle affects the redistribution of SWC

The dynamic distribution of SWC reflects the water and energy state of the soil, which depends on water-soil interactions and the physical properties of the soil (Chen et al., 2013). During the freezing progress, unfrozen water in the soil moves to the frozen area driven by the water potential difference, which is manifested by soil water accumulation at the frozen layer (e.g. water is transported upward). Meanwhile, the in- crease of soil water at the frozen layer reduces the rate that the soil freezes, which provides enough time for the soil water to transport up- ward. Our research also proved that the freezing rate gradually decreased during the freezing process (Fig. 7b).

When the soil began to freeze, the water potential gradient was the primary factor driving the soil water to migrate vertically. Soil freezing lowers the water potential and creates a free energy gradient along which water will flow from unfrozen soil to frozen soil. Then, before the ST dropped to the freezing point, the water in the frozen soil was redistributed to the freezing front (Dirksen and Miller, 1966). The phenomenon of decreasing SWC was observed at various depths before the temperature dropped to the freezing point (Fig. 7b), which was consistent with others’ results (Dirksen and Miller, 1966). This phe- nomenon indicated that the existence of the water potential gradient is a necessary condition for a certain amount of soil water to flow to the freezing front. The largest soil water potential appeared at a depth of 100 cm, and soil water moved up rapidly under the combined influence of the temperature difference and water potential, making the SWC above 100 cm increase significantly. However the soil water potential was decreased due to the temperature difference and the formation of