The leaching experiments were carried out in the Tarim Basin of southern Xinjiang, Northwest China, between June to September 2018 in the cotton field. The experimental site was located 901 m above mean sea level (Figure 1). The field belongs to continental desert climate with less rain and strong evaporation. The annual precipitation was 53.3–62.7 mm, and the annual evaporation was 2,273–2,788 mm. Compared with evaporation and crop transpiration, the supply of atmospheric rainfall was very little. The fresh water (FW) was taken from the Peacock River with an average total dissolved solids (TDS) of 0.62–0.72 g/L, and the brackish water (BW) was taken from a well located in the test field with TDS of 1.61–1.71 g/L (Table 1). The sodium adsorption ratio (SAR) of infiltration water ranged from 10 to 14 of FW and 8–9 of BW. The main soil type was loamy sand (Table 2). The electrical conductivity of soil was less than 1.0 dS/m with no salinization (Slavich and Petterson 1993; Chen et al., 2018). The field water capacity was 0.2.

The leaching experiments were conducted in 10 randomly selected field plots, each having a size of 225 m² (Figure 2). There were set as 17 infiltration lines with the mode of “one mulch, two infiltration lines, and four rows”. In total, 2 infiltration lines were installed for every four rows of cotton. The distance between the infiltration lines was 55 cm and that between two cotton rows was 20 cm. Infiltration points along each line were spaced 30 cm apart, while the non-mulched area of bare soil was 40-cm wide. Before sowing, flood infiltration of fresh water was conducted to leach salt and preserve soil water on 20 April, with amount of 57.3 mm. Then, cotton seeds were sown and mulched with degradable plastic sheeting. Brackish water was conducted one time per 5 days, for totally 14 times. The emitter flow was 2.2 L/h. The amount of applied infiltration water during the growing stages followed traditional irrigation patterns of cotton (Table 3). In total, 2 groups of control experiments were designed for fresh water of 0.68 g/L and brackish water of 1.66 g/L with 1.0 Q = 572 mm. Other eight groups of MDI experiments leached soil salt by increasing the amount of brackish water with 1.05–1.40 Q.

The soil samples were drilled at 8:00 p.m. before each infiltration for gravimetric water content and EC_1:5 analysis. The sampling interval of each sample was 10 s at a depth of 0–60 cm and 20 s at depth of 60–100 cm. The soil water content was measured by the drying method. The EC_1:5 was referred to the EC of 1–5 soil/water suspensions using a conductivity meter (DDS-307, INESA Scientific Instrument Co., Ltd., Shanghai, China).

A 600 × 100 m section was to present the water–salt transport–based 2D conceptual model (Figure 3). The infiltration point was simplified as line water supply. Under the intermittent infiltration of brackish water, soil salt mainly accumulated in 40–60 cm (Chen et al., 2018; Li et al., 2018; Hu et al., 2020), and the soil was less affected by evaporation below 1.2 m (Heng et al., 2018). The pipe was theoretically suggested to install underground 1.6m, vertical to infiltration line (Figure 3).

The numerical simulation of soil water–salt transport was conducted using HYDRUS (Šimůnek et al., 2016; Katarina et al., 2019). The ground pipe was set as the free drainage boundary, the bare area was the atmospheric boundary, and the dipper was the variable flow boundary (Var.Fl₁). The water quantity data of the infiltration point was assigned by Table 3. There was no flow in the Y direction of the model and under the mulch except infiltration points (Var.Fl₂).

Flow boundary condition:

Lateral boundaries: q y = 0. (1)

Atmospheric boundaries: − K ( h ) ( ∂ h ∂ z + 1 ) = q a ( t )   ( 0 ≤ y ≤ 20,60 ≤ y ≤ 90,130 ≤ y ≤ 170,280 ≤ y ≤ 320,430 ≤ y ≤ 470,580 ≤ y ≤ 600 ,   Z = 160 ) . (2)

Var.Fl₁: − K ( h ) ( ∂ h ∂ z + 1 ) = q d ( t ) ( 50 ≤ y ≤ 60,90 ≤ y ≤ 100,200 ≤ y ≤ 210,240 ≤ y ≤ 250,350 ≤ y ≤ 360,390 ≤ y ≤ 400,500 ≤ y ≤ 510,540 ≤ y ≤ 550 ,   Z = 160 ) . (3)

Var.Fl₂: − K ( h ) ( ∂ h ∂ z + 1 ) = 0     ( 20 ≤ y ≤ 50,60 ≤ y ≤ 90,100 ≤ y ≤ 130,170 ≤ y ≤ 200,210 ≤ y ≤ 240,250 ≤ y ≤ 280,320 ≤ y ≤ 350,360 ≤ y ≤ 390,400 ≤ y ≤ 430,470 ≤ y ≤ 500,510 ≤ y ≤ 540,550 ≤ y ≤ 580 ,   Z = 160 ) . (4)

Solute boundary conditions: − θ ( D y z ∂ c ∂ y + D z z ∂ c ∂ z ) + q c = q d c d ( 50 ≤ y ≤ 60,90 ≤ y ≤ 100,200 ≤ y ≤ 210,240 ≤ y ≤ 250,350 ≤ y ≤ 360,390 ≤ y ≤ 400,500 ≤ y ≤ 510,540 ≤ y ≤ 550 ,   Z = 160 ) , (5) where h is the water pressure head, cm; θ is the volumetric water content, cm³/cm³; c is the salinity of the soil water, g/L; t is time, day; x and z are the spatial coordinates, cm; K is the unsaturated hydraulic conductivity, cm/d; D was the dispersion coefficient, cm²/d; q _y , q _d , and q _a are the water flux of lateral boundaries, infiltration points, bare soil surface, cm/day; c _a was the salinity of the infiltration water, g/L.

The soil hydraulic properties were described using the standard equations of Van Genuchten (1980). The estimated hydraulic parameters were listed in Table 4. The initial longitudinal dispersion coefficient (D _L ) and transverse dispersion coefficient (D _T ) were defined by previous literature (Wang et al., 2014; Singh et al., 2019). The final diffusion coefficient was set to 1.58 cm²/day after verification and adjustment based on the experimental values (Wang et al., 2017; Ranjbar et al., 2019). The linear isotherm adsorption coefficient K _d was set as 0.28 cm³ g⁻¹ (Kadyampakeni et al., 2018).

We simulated the spatial distributions of soil water and salt in the two-direction under mulched intermittent infiltration. Furthermore, the model values of HYDRUS-2D were compared with actual observed values (Figures 4,5). The model could display the dynamic variation of soil moisture and salinity better with infiltration events. A t-test was carried out between the measured and simulated values, and the results showed that the significance level of p > 0.05, indicating that the simulation results of the solute transport model were verified successfully (Table 5).

Soil samples were collected the day before infiltration treatment and obtained soil moisture dates, which could better reflect the water holding capacity of the soil (Figure 6). Within a single infiltration interval of 5 days, the migration distance of soil water in unsaturated zone was more than 40 cm. Due to the water uptake by plant roots, the soil moisture content was almost less than the field capacity of 0.203 at depth of 0–60 cm before each infiltration, and the water mainly existed at the depth of 60–100 cm. Compared two control experiment schemes, the soil water migrated faster under fresh water treatment and the soil reached saturation at the depth of 100 cm with moisture content of 0.35. Brackish water changed the soil structure and reduced the soil permeability because of high sodium ion content and large sodium adsorption ratio. The soil water holding capacity rate was lower than 0.2 after 5 days with 1.20Q–1.40Q brackish water treatments.

The soil salt accumulation at the depth of 0–30 cm was lower under water leaching in the whole growth period (Figure 7). The EC_1:5 of soil in the root zone was basically less than 0.55 dS/m, which was no salinized (Table 6). There was massive salt accumulation at the depth of 30–60 cm. The average salinity of the 10 experimental groups was 0.539 ds/m, 0.979 dS/m and 0.684 ds/m respectively at the depth of 30–40, 40–50, 50–60 cm. The maximum salinity was accumulated at the depth of 40–50 cm. When the groundwater level was greater than the limit evaporation depth, the soil in the root zone formed a “low salt zone” at the depth of 0–30 cm. The cumulative increase of soil salt at the depth of 30–50 cm reached largest, and decreased below 50 cm.

Comparing the soil salinity of each infiltration event to the first leaching salt by fresh water (20 April) showed that the variation of EC_1:5 at depth of 0–30 cm was less than 0.5 dS/m (Figure 8). The root distribution was measured and time-variable through plant growth which could be accounted for by modifying the ratio between evaporation and transpiration (Wang et al., 2014). Soil salinization mainly occurs in 30–60 cm, and the variation of EC_1:5 exceed 4.0 dS/m. Root uptake as a key way to transport soil water into atmosphere contributed mainly to soil salinization. The average variation of EC_1:5 showed that the soil salt little changed before and after brackish water infiltration. The brackish water intermittent infiltration under plastic mulch could effectively improve soil salinization and protect the cultivated environment.

Soil salinity was evaluated using the electric conductivity of the saturation extract as EC _e , which could be calculated from measured values of EC_1:5 using EC _e = 3.64 EC_1:5 (Slavich and Petterson 1993; Wang et al., 2014). We evaluated the salinization degree of soil based on Table 6 (Fitzpatrick 1980; Akça et al., 2020).

The maximum salinity of 10 experimental treatments at depth of 40–50 cm was 3.07, 2.46, 1.99, 0.274, 0.32, 0.38, 1.64, 1.43, 2.41, 0.31 dS/m respectively. Fresh water infiltration would still lead to soil moderate salinization. The soil salinity decreased first and then increased with increasing infiltration water amount of 1.0–1.35Q under brackish water infiltration. The total salt of soil was the lowest with water treatment of 1.15 Q = 657.8 mm. The EC_1:5 of soil was less than 0.55 dS/m, reaching the degree of no salinization. Whether the amount of water was too large or too small, it would lead to salt accumulation. The leaching intensity was not enough with low water amount, resulting in the accumulation of salt in the topsoil. Numerous ions were input into the soil through large amount of water, which could not be absorbed by plants and lead to root salt stress. The optimization of soil environment achieved the best effect with 1.15 Q water treatment.

Long-term infiltration causes the salt of topsoil to penetrate into the deep soil layer and then salinize the groundwater. At present, the drainage channel with the depth of 1.8 m in the experimental field could not meet the salt discharge requirement. Therefore, underground pipe drainage system should be adopted to receive the upper leaching water and salt, collect and recycle it, desalinate soil, and protect groundwater. We simulate water–salt of soil under a leaching–drainage system for 10 years with 1.15 Q = 657.8 mm treatment (Figure 9).

Soil water gradually pushed downward under brackish water intermittent infiltration, and soil salinity is gradually concentrated to the edge of the wetting front. Under the leaching of multiple infiltration points, the wetting front overlaps with each other to make the salt migrate downward, forming a “low salination soil layer”. The salt mainly accumulated to 1.10–2.99 dS/m at the depth of 40–60 cm during 0–30 days, and the soil salinity under mulch area was lower than 0.55 dS/m at depth of 0–10 cm. After 120 days, no salinization soil area expanded to depth of 40–45 cm. During non-infiltration period (121–365 days), the soil salt moved upward under evaporation effect, and the no salinization area was reduced and showed mild salinization. Figures 7E–H showed the prediction results of 2–10 years. The salt was gradually leached and discharged through underground pipe. The soil total salt under mulch was less than 0.55 dS/m, which could not only provide a low salt area for cotton growth but also control soil salt accumulation and protect soil as well as groundwater environment.