Effects of Magnetized Water on Infiltration and Salt Redistribution in Undisturbed Saline–Alkali Soil Columns from the Hetao Irrigation District: Dependence on Irrigation Water Source

Abstract

Soil salinization and freshwater scarcity are two of the most pressing challenges limiting sustainable agriculture in arid and semi-arid regions worldwide. Magnetized irrigation has been proposed as a physical water treatment that introduces no chemical additives and may improve water infiltration and mitigate secondary salinization; however, its effectiveness under contrasting irrigation water sources remains poorly understood, particularly in undisturbed saline–alkali soils. This study investigated the effects of magnetized irrigation water from two sources—Yellow River water (HW) and subsurface drainage water (DW)—on soil water infiltration and water–salt transport in the Hetao Irrigation District, Inner Mongolia. Undisturbed soil columns were irrigated under three magnetic field intensities: 0, 3500, and 5000 Gauss (Gs). Cumulative infiltration, infiltration rate, wetting front advancement, soil moisture, salt storage, and ion redistribution were analyzed. Magnetization affected infiltration and water–salt redistribution, and the magnitude of the response differed between the two water sources. Mean infiltration rate increased by 11.6–13.9% under magnetization relative to non-magnetized controls (p < 0.001), and infiltration duration decreased by 6.1–7.1% (p = 0.020), with no significant difference between the 3500 and 5000 Gs treatments (p = 0.903). The water source × magnetic intensity interaction was not significant for infiltration rate or duration. Salt storage in deep layers was reduced under magnetized treatments: by 24.74% at 40–50 cm under HW at 5000 Gs and by 29.18% at 50–60 cm under DW at 3500 Gs (both p < 0.05 relative to the respective non-magnetized controls). Magnetized irrigation was associated with reduced Na⁺ and Cl⁻ accumulation in the 0–30 cm layer and with a higher relative proportion of Ca²⁺ and Mg²⁺ in the soil water extract. These findings provide experimental evidence on how the response to magnetized irrigation varies with irrigation water source in undisturbed saline–alkali soils and support further evaluation of subsurface drainage water reuse in salinized farmland.

1. Introduction

With the intensification of global climate change and continuous population growth, land salinization and water scarcity have become key constraints on food security and sustainable agricultural development [1]. Arid and semiarid irrigation areas in China, particularly Xinjiang and Hetao, have long faced limited water resources, severe soil salinization, and poor drainage, resulting in farmland yield loss and ecological deterioration [2,3,4]. Studies have attempted to promote desalination by altering water–salt transport pathways; for instance, Xu et al. [5] found that using biochar interlayers effectively intercepts salt transport, but such methods rely on exogenous additives with high costs. Consequently, the scientific utilization of accessible unconventional water sources such as brackish water to improve irrigation efficiency has become a central issue in agricultural water management [6]. Brackish water is considered a viable option for alleviating water shortage owing to its abundant reserves [7,8,9]; however, prolonged excessive brackish water irrigation leads to salt accumulation and secondary salinization risk [10], restricting crop yield and quality [11,12]. Against this background, magnetized water technology has attracted growing attention for its potential to optimize irrigation water properties and enhance soil water–salt regulation while mitigating salt accumulation.

Magnetized water refers to water whose physicochemical properties have been altered by passing through a magnetic field of a specific intensity, modifying its interaction with soil and crop systems [13,14,15]. As a physical treatment that introduces no chemical additives into the irrigation water, magnetized water irrigation has attracted growing attention for its potential to improve infiltration, modify soil structure, regulate ion migration, and increase crop yield [16,17].

At the microscopic level, previous studies have proposed that magnetic treatment may alter hydrogen-bond interactions and the association state of water molecules, which could in turn affect surface tension, viscosity, and ion hydration behavior [13,18,19]. For example, Pang and Deng [20] reported that magnetic fields may weaken hydrogen-bond networks and reduce surface tension and viscosity, potentially improving water migration through porous media and its hydration capacity for salt ions. It should be noted, however, that these mechanisms remain a subject of ongoing scientific debate [15,21], and their magnitude and direction can vary substantially with water quality, soil texture, and magnetic treatment conditions [17,22]. Therefore, the effects reported in the present study are interpreted primarily as macroscopic responses, while the underlying microscopic mechanisms are discussed as hypotheses rather than demonstrated processes.

In field applications, Maheshwari and Grewal [17] and Surendran et al. [23] demonstrated in Australia and India, respectively, that magnetized water improves aggregate stability, increases effective porosity, and enhances infiltration rate. Studies conducted in China confirmed comparable effects in Xinjiang and Heilongjiang [22,24]. Khoshravesh et al. [25] further observed that magnetized water regulates water–salt–fertilizer coupling and alleviates shallow salt re-accumulation.

Despite these advances, three notable deficiencies persist. First, most infiltration experiments employ repacked soil columns [22,23,26], which inevitably disrupt natural pore connectivity and aggregate structure governing preferential flow and salt redistribution [27]. Although several field studies have reported positive effects of magnetized irrigation on soil physical properties and infiltration [15,21,28], whether conclusions obtained from repacked columns can be transferred to undisturbed field soils remains uncertain because repacking removes macropores, compaction layers, and the heterogeneous pore network that constrain water movement in situ. Third, most studies examine only total salt changes at specific depths [4,29] without characterizing the effects on ion composition structure across the full soil profile.

Based on the mechanisms proposed in previous literature, we hypothesized that: (i) the effect of magnetic treatment on soil infiltration and salt redistribution is source-dependent, with different optimal magnetic intensities for irrigation waters of contrasting salinity and ion composition; and (ii) magnetized irrigation preferentially promotes the downward redistribution of monovalent ions (e.g., Na⁺ and Cl⁻) relative to divalent cations (Ca²⁺ and Mg²⁺) within the soil profile, thereby altering the ionic composition of the soil–water extract. To test these hypotheses, we conducted infiltration experiments using undisturbed saline–alkali soil columns from the Hetao Irrigation District under a factorial combination of two water sources (Yellow River water and subsurface drainage water) and three magnetic field intensities (0, 3500, and 5000 Gs). To the best of our knowledge, this is among the first studies to combine undisturbed soil columns with a systematic comparison of contrasting irrigation water sources under magnetized treatment. The results are expected to provide an experimental basis for subsurface DW utilization and magnetized water application in the considered region, as well as insights for sustainable management of salinized soils in arid and semiarid regions of China.

2. Materials and Methods

2.1. Sample Collection

2.1.1. Soil Samples

The experimental soil was collected from the Yumi Science and Technology Experimental Base in Wuyuan County, Bayannur City, Inner Mongolia Autonomous Region, China (41°22′ N, 107°19′ E), as shown in Figure 1. The experimental site is a typical saline–alkali soil of the Hetao Irrigation District, developed under a shallow groundwater table and characterized by salt accumulation in the surface layer. Soil texture was determined following the method of Shirazi et al. [30], ranging from loam in the surface layer to silty clay in the deeper layers, with detailed particle-size distribution and textural classes for each depth provided in Table 1. The terrain is flat, with a typical temperate continental climate. Average annual precipitation ranges from 280 to 360 mm, concentrated mainly between July and September, whereas average annual evaporation ranges from 900 to 2100 mm, far exceeding precipitation and resulting in considerable water deficiency in the region.

Figure 1. Location of Wuyuan County, Inner Mongolia.

Table 1. Initial physical properties of the soil.

Soil Depth	Soil Particle Size Distribution (%)			Bulk Density	Moisture Content (%)	Soil Texture
(cm)	>0.05 mm	0.05–0.002 mm	<0.002 mm	(g·cm−3)
0–10	31.2	46.2	22.6	1.42	14.79	loam
10–20	25.9	50.2	23.9	1.46	17.89	silt loam
20–30	18.6	49.6	31.8	1.52	18.88	silty clay loam
30–40	20.5	46.3	33.2	1.43	22.86	silty clay loam
40–50	4.1	50.1	45.8	1.31	24.79	silty clay
50–60	0.0	52.4	47.6	1.34	12.84	silty clay

Note: Values represent the mean of three replicates (n = 3).

Undisturbed soil sampling was conducted in early June 2025, prior to the summer irrigation season, to ensure that the columns represented the baseline saline–alkali condition of the field. The 18 undisturbed columns were collected from a single field unit (approximately 0.96 m × 2.22 m). Undisturbed soil samples (diameter: 10 cm; height: 60 cm) were collected by inserting a cylindrical sampler directly into the soil using the core sampling method. During sampling, a reinforced soil sampling cylinder was advanced section by section along the soil profile to avoid compaction and structural damage. After transportation to the laboratory, all soil columns were kept intact and stored at ambient temperature (25 ± 2 °C) without pre-drying in order to preserve the natural pore structure, aggregate stability, and initial moisture profile. Physical and chemical properties of the soil (Table 1 and Table 2) were measured on subsamples collected from the same field location and processed following standard procedures. The coefficient of variation (CV) was calculated for bulk density, initial moisture content, and total salt content at each depth across all 18 soil columns (n = 18 per depth) to verify the uniformity of parallel controls; the coefficient of variation for same-layer indicators across all treated soil columns was <5% (within ±2.5%), satisfying the requirements for parallel control. SAR was determined in 1:5 soil–water extracts and remained well below 13 throughout the profile, confirming the soil is saline rather than sodic.

Table 2. Initial chemical properties of the soil.

Soil Depth	Ionic Composition (g·kg−1)						Total Salt Content	pH
(cm)	HCO3−	Cl−	SO42−	Ca2+	Mg2+	Na+	(g·kg−1)
0–10	0.49	0.11	0.15	0.10	0.06	0.09	1.13	8.14
10–20	0.39	0.18	0.38	0.12	0.07	0.17	1.32	8.08
20–30	0.55	0.18	0.19	0.12	0.05	0.19	1.27	8.15
30–40	0.46	0.14	0.14	0.08	0.06	0.12	1.15	8.09
40–50	0.54	0.17	0.12	0.06	0.09	0.16	1.14	8.19
50–60	0.56	0.16	0.10	0.08	0.06	0.15	1.08	8.22

Note: Values represent the mean of three replicates (n = 3). All units are g·kg⁻¹ except pH (dimensionless).

2.1.2. Irrigation Water Samples

HW was drawn directly from a tributary of the Yellow River in Wuyuan County, Bayannur City, while subsurface DW was collected from the aforementioned test site. The subsurface pipes were buried at a depth of 1.7 m, with a pipe diameter of 110 mm and a spacing of 20 m. The total salt content ratio between the two water sources was approximately 1:3. The chemical properties of both water sources are presented in Table 3.

Table 3. Chemical properties of irrigation water.

Irrigation Water	Ionic Composition (g·L−1)						Total Salt Content	pH
	HCO3−	Cl−	SO42−	Ca2+	Mg2+	Na+	(g·L−1)
HW	0.33	0.12	0.23	0.07	0.05	0.08	0.58	8.18
DW	0.69	0.25	0.48	0.16	0.15	0.18	1.73	7.86

Note: HW, river water; DW, subsurface drainage water. Values represent the mean of three replicates (n = 3). All measurements were conducted prior to magnetization treatment.

2.2. Experimental Apparatus

2.2.1. Portable Magnetized Water Circulation Device

The magnetization device is shown in Figure 2. Magnetized water was prepared using a permanent magnet device (The device produced magnetic flux densities of 0.35 and 0.50 T (3500 and 5000 G), with a non-magnetized control at 0 T; all intensities are reported in Gauss with the SI equivalent given at first mention) developed by the Baotou Rare Earth Research Institute (Baotou, China), equipped with five NdFeB permanent magnets. The magnetic field uniformity exceeded 96%, with an effective magnetization length of 50 cm. Magnetization intensity was adjusted by varying the distance between magnets and verified at multiple points along the central axis using a Teslameter (model: HT20, Shanghai Precision Instrument & Meter Co., Ltd., Shanghai, China), with deviations from set values remaining below ±5%.

Figure 2. Schematic of the magnetized water circulation device. (a) Model diagram and (b) photograph of the actual device.

2.2.2. Mariotte Bottle–Soil Column Infiltration Device

The infiltration device comprised acrylic Mariotte bottles and soil columns were custom-made in Hohhot, China (Figure 3). The Mariotte bottle had an inner diameter of 12 cm, a height of 40 cm, and a wall thickness of 7 mm. The water outlet was connected via a hose to a dripper serving as the water distribution device, ensuring uniform and gentle application to the soil surface without splashing or disturbing the surface structure. The soil column had a height of 70 cm and an inner diameter of 10 cm, with sampling holes (diameter: 2.7 cm) spaced at 10 cm intervals, and a filtrate collection trough at the bottom. During each magnetization operation, water was circulated through the magnetization device at a controlled flow rate of 3.0 L·min⁻¹, and the water was circulated for 20 min to ensure sufficient magnetization exposure prior to irrigation.

Figure 3. Schematic of the Mariotte bottle–soil column gravity drip irrigation system.

2.3. Experimental Design

Undisturbed soil columns were irrigated with HW and subsurface DW to simulate field drip irrigation. Six treatments at three magnetization intensities (0, 3500, and 5000 Gs) were established, corresponding to HW0, HW1, and HW2 (HW) and DW0, DW1, and DW2 (DW), respectively, each replicated thrice (18 columns in total). Each soil column was used only once for a single treatment to ensure experimental independence. Magnetization intensities were selected based on pre-experimental results and published literature [17,22].

The experiment was conducted at the Wuyuan County Science and Technology Experimental Base from August to September 2025 (indoor temperature: 25 °C ± 2 °C; relative humidity: 25% ± 5%). A 5 cm quartz sand filter layer was placed at the column base to prevent soil loss [31], and leachate was collected in beakers beneath. Black plastic film covered the soil surface to eliminate evaporation. A single drip irrigation pipe (outer diameter: 16 mm) was inserted vertically into the center of each column, with the dripper positioned 5 mm above the soil surface. Because previous studies have suggested that the physicochemical effects of magnetization on water may persist for several hours after treatment [13], irrigation commenced immediately after magnetization.

A constant-head Mariotte bottle maintained a 5 mm (±0.5 mm) ponding layer on the soil surface. Water level and wetting front position were recorded at intervals of 1 min (0–10 min), 5 min (10–60 min), and 10 min (60–180 min). Drip irrigation was terminated at a cumulative infiltration volume of 700 mL, which had been pre-confirmed as sufficient for the wetting front to approach the column base. The experiment concluded when the wetting front reached the filter layer, followed by immediate stratified sampling and analysis.

2.4. Measurement Indicators and Calculation

2.4.1. Soil Sampling After Infiltration

Upon termination of each infiltration experiment (when the wetting front reached the filter layer at the column base), soil sampling was immediately conducted to minimize post-infiltration redistribution. Each soil column was carefully sectioned into six 10 cm layers (0–10, 10–20, 20–30, 30–40, 40–50, and 50–60 cm) using a stainless-steel soil knife. At each depth, approximately 200 g of soil was collected from the central portion of the layer to avoid edge effects and immediately sealed in airtight polyethylene bags. Samples were transported to the laboratory within 2 h and stored at 4 °C prior to analysis. The leachate collected at the column base was used only as an indicator that the wetting front had reached the bottom of the column; its ionic composition was not analyzed, and no ion mass balance was established.

2.4.2. Cumulative Infiltration Volume and Infiltration Rate

The cumulative infiltration volume V(t) (mL) entering the column up to time t was obtained from the decrease in water volume in the Mariotte bottle:

$$\begin{array}{c}V\left(t\right)=V_0-V_t,\end{array}$$

(1)

where V₀ (mL) is the water volume in the Mariotte bottle at the start of the run and Vt (mL) is the volume remaining at time t.

The cumulative infiltration depth I(t) (mm) was calculated as

$$\begin{array}{c}I(t)={\displaystyle \frac{{10\times (V_0-V}_t)}{A}},\end{array}$$

(2)

where A represents the cross-sectional area of the soil column ($A={\displaystyle \frac{{\mathit{\pi D}}^2}{4}}$).

The mean infiltration rate over a time interval was calculated as

$$\begin{array}{c}F\left(t\right)={\displaystyle \frac{{10\times (V}_{t2}-V_{t1})}{A(t_2-t_1)}},\end{array}$$

(3)

where V_t1 and V_t2 are the water volumes remaining in the bottle at times t₁ and t₂ (t₂ > t₁), expressed in mm·min⁻¹.

2.4.3. Wetting Front Measurement

The wetting front depth was recorded non-destructively through the transparent wall of the column. At each observation time, the vertical distance from the soil surface to the wetting front was read at four positions marked at 90° intervals around the circumference of the column (front, back, left, right), and the arithmetic mean of the four readings was taken as the wetting front depth of that column at that time. The mean wetting front depth was calculated as

$$\begin{array}{c}\overline{D}(t)={\displaystyle \frac{1}{4}}{\displaystyle \sum _{j=1}^4}{D}_j\left(t\right)\end{array}$$

(4)

where D(t) is the wetting front depth at the j-th observation point at time t, and $\overline{D}\left(t\right)$ is t mean wetting front depth.

2.4.4. Salt Content Index

Soil pH was measured with a Mettler-Toledo FE28 pH meter (Mettler-Toledo, Shanghai, China) in 1:5 soil–water suspensions. Salt ion composition was determined in extracts at the same ratio: CO₃²⁻ and HCO₃⁻ by double-indicator titration, Cl⁻ by AgNO₃ titration, SO₄²⁻ by indirect ethylenediaminetetraacetic acid (EDTA) complexometric titration, Ca²⁺ and Mg²⁺ by EDTA complexometric titration, and Na⁺ by flame photometry. Soil electrical conductivity (EC_1:5, mS·cm⁻¹, numerically equivalent to dS·m⁻¹) was measured in the same extracts with a DDSJ-308F conductivity meter (Shanghai INESA Scientific Instrument Co., Ltd., Shanghai, China). Total soil salt content was calculated from EC_1:5 using the empirical relation established for soils of the Hetao Irrigation District:

$$\begin{array}{c}{\mathit{TS}}_k=3.7657{\mathit{EC}}_{1:5}-0.2405,\end{array}$$

(5)

where TS_k represents the total salt content of the kth soil layer (g·kg⁻¹), and EC₁:₅ is the electrical conductivity of the soil extract (mS·cm⁻¹). The coefficient 3.7657 therefore carries units of (g·kg⁻¹)/(mS·cm⁻¹) and the intercept −0.2405 units of g·kg⁻¹.

The salt storage of the soil column profile (SS, kg·m⁻²) was calculated as follows:

$$\begin{array}{c}\mathit{SS}={\displaystyle \sum _{k=1}^n}(T{S}_k{\rho }_{b,k}\Delta z_k)\times {10}^{-2},\end{array}$$

(6)

where ρ_b,k is the bulk density of the kth soil layer (g·cm⁻³), Δz_k is the thickness of the kth layer (cm), and n is the number of layers; because 1 kg·m⁻² = 10 t·hm⁻², the same quantity can be expressed in t·hm⁻² by multiplying by 10.

The sodium adsorption ratio (SAR), an indicator of soil sodicity hazard, was calculated from the concentrations of Na⁺, Ca²⁺, and Mg²⁺ in the soil water extract as follows:

$$\mathit{SAR}={\displaystyle \frac{\left[{\mathit{Na}}^{+}\right]}{\sqrt{\frac{\left[{\mathit{Ca}}^{2+}\right] + \left[{\mathit{Mg}}^{2+}\right]}{2}}}}$$

(7)

where Na⁺, Ca²⁺, and Mg²⁺ represent the concentrations of the corresponding cations in the soil water extract (mmol·L⁻¹).

2.4.5. Soil Moisture Measurement

Gravimetric soil moisture content was determined by oven-drying subsamples at 105 °C for 24 h until constant weight was achieved. Moisture content (θ, %) was calculated as

$$\theta ={\displaystyle \frac{m_{\mathrm{wet}}-m_{\mathrm{dry}}}{m_{\mathrm{dry}}}}\times 100$$

where m_wet and m_dry are the masses (g) of wet and oven-dried soil, respectively.

2.4.6. Multivariate and Hydrochemical Analysis

Hydrochemical facies were classified with a Piper diagram, in which the major cation and anion concentrations of each sample were expressed as charge-equivalent percentages of total cations or anions (meq%) and plotted by treatment. This identified the dominant water type and the shift in ionic composition caused by magnetization relative to the original irrigation water.

Redundancy analysis was performed in R version 4.2.1 (R Foundation for Statistical Computing, Vienna, Austria) with the vegan package to relate soil chemical variables to the experimental treatments. Species data were log-transformed and standardized, and treatment (water source and magnetic intensity) was used as the explanatory variable. Significance was assessed by Monte Carlo permutation (499 permutations), and depth was not partitioned as a separate term. The first two axes explained 56.91% of the total variation.

2.5. Data Processing and Statistical Analysis

All data are presented as mean ± standard deviation (n = 3). Analyses were performed with SPSS 26.0 (IBM Corp., Armonk, NY, USA) and figures prepared with Origin 2025 (OriginLab, Northampton, MA, USA). The experiment followed a two-factor completely randomized design with irrigation water source (W: HW, DW) and magnetic intensity (M: 0, 3500, 5000 Gs) as fixed factors, giving six treatments with three undisturbed soil columns each. The soil column was the experimental unit, and all repeated measurements were first aggregated to the column level. Infiltration variables (Section 3.1) were recorded repeatedly on the same column and were therefore first averaged to column means before analysis, so that the column remained the experimental unit; water source, magnetic intensity and their interaction were treated as fixed effects. The two duration variables were analyzed by two-way ANOVA (W × M) on column means, with simple-effects tests used to resolve significant interactions. Soil moisture, salt storage, ion concentrations and SAR (Section 3.2, Section 3.3 and Section 3.4) were analyzed by three-way ANOVA with water source (W), magnetic intensity (M) and soil depth (D) as factors, in which depth entered the model as an explicit factor rather than as a replicate (W × M × D; error df = 72, Table S1). Relationships among soil–water–salt variables and the treatment factors were examined by redundancy analysis. Pairwise comparisons were performed using Tukey’s test for the infiltration variables (Section 3.1) and Duncan’s multiple range test for all remaining variables, with differences considered significant at p < 0.05.

3. Results

3.1. Influence of Magnetized Water Drip Irrigation on Cumulative Infiltration and Wetting Front Advancement

The cumulative infiltration volume increased nonlinearly over time, whereas the infiltration rate decreased rapidly at the initial stage and then gradually approached a stable value, transitioning from rapid to slow infiltration (Figure 4), with pronounced differences during the early stage. During the first 0–30 min, the cumulative infiltration volumes of HW1 and HW2 increased by 15.20% and 18.53%, respectively, compared with HW0. For subsurface DW, the cumulative infiltration volumes of DW1 and DW2 increased by 14.83% and 8.52%, respectively, compared with DW0.

Figure 4. Infiltration characteristics under magnetized treatments: (a,b) cumulative infiltration volume and (c,d) corresponding infiltration rate for HW (a,c) and DW (b,d) treatments. Error bars represent the standard deviation (n= 3). The alternating background bands indicate the observation intervals, which become progressively wider from left to right as the sampling frequency decreases during infiltration. HW0, HW1, and HW2 represent Yellow River water treated at 0, 3500, and 5000 Gs, respectively; DW0, DW1, and DW2 represent subsurface drainage water treated at 0, 3500, and 5000 Gs, respectively. Statistical details are provided in Table S2, Part A.

The mean infiltration rate was significantly influenced by magnetic intensity (F(2,12) = 24.684, p < 0.001, partial η² = 0.804; Table S2, Part A), whereas neither water source (F(1, 12) = 0.414, p = 0.532, partial η² = 0.033) nor the water source × intensity interaction (F(2, 12) = 0.710, p = 0.511, partial η² = 0.106) was significant. HW1 and HW2 increased the mean rate by 11.6% and 12.9% relative to HW0, respectively; under DW, DW1 and DW2 increased the mean rate by 13.9% and 10.4% relative to DW0.

The steady-state infiltration rate differed significantly between water sources (F(1, 12) = 8.847, p = 0.012, partial η² = 0.424; Table S2, Part A), with DW treatments averaging higher than HW, whereas magnetic intensity had no significant effect (F(2, 12) = 0.128, p = 0.881, partial η² = 0.021) and the interaction was not significant (F(2, 12) = 1.167, p = 0.344, partial η² = 0.163). Under HW, the steady-state rate was 3.0% higher in HW1 and 1.80% higher in HW2 than in HW0; under DW, DW1 was 3.0% lower than DW0, and DW2 was essentially unchanged.

Infiltration duration (the time required to deliver the fixed 700 mL) was significantly affected by magnetic intensity (F(2, 12) = 5.557, p = 0.020, partial η² = 0.481; Table S2, Part A), decreasing from 148.8 min at 0 Gs to 138.2 min at 3500 Gs (7.1%; p = 0.024) and 139.7 min at 5000 Gs (6.1%; p = 0.052, not significant), with no significant difference between the two magnetized intensities (p = 0.903). Water source (F(1, 12) = 1.784, p = 0.206, partial η² = 0.129) and the water source × intensity interaction (F(2, 12) = 1.789, p = 0.209, partial η² = 0.230) were not significant. Magnetization therefore shortened the time required to transmit the same water volume. Because the steady-state infiltration rate was unaffected by magnetization, the shorter delivery time is attributable to faster transient (early-stage) infiltration rather than to an increase in the terminal steady flux.

Magnetization significantly advanced the wetting front (Figure 5). Averaged over the two water sources, the time required for the wetting front to reach the 60 cm bottom of the columns was 207.8, 188.8 and 188.2 min under 0, 3500 and 5000 Gs, respectively, i.e., 19.00 min (9.1%) and 19.67 min (9.5%) shorter than the non-magnetized control (Tukey’s test, p < 0.001; Table S2, Part B; Figure 6), whereas the two magnetized levels did not differ from each other (0.67 min, p = 0.813). The response depended on the irrigation-water source (water source × magnetization: F(2, 12) = 5.614, p = 0.019, partial η² = 0.483), and the water-source main effect was significant (F(1, 12) = 12.472, p = 0.004), with the wetting front reaching the 60 cm bottom earlier under DW than under HW.

Figure 5. Wetting front advancement under magnetized treatments: (a) HW treatments; (b) DW treatments. Data are presented as means ± SD (n = 3). Statistical details are provided in Table S2, Part B.

Figure 5. Wetting front advancement under magnetized treatments: (a) HW treatments; (b) DW treatments. Data are presented as means ± SD (n = 3). Statistical details are provided in Table S2, Part B.

Figure 6. Time required to reach the target cumulative infiltration volume (700 mL) and time required for the wetting front to reach the 60 cm bottom of the columns under different magnetized irrigation treatments: (a) HW treatments; (b) DW treatments. In each panel, the left group of bars is the 700 mL delivery time, and the right group of bars is the wetting front advance time. Error bars represent the standard deviation (n = 3). Different lowercase letters within the same group indicate significant differences among treatments (p < 0.05, two-way ANOVA followed by Tukey’s test).

Figure 6. Time required to reach the target cumulative infiltration volume (700 mL) and time required for the wetting front to reach the 60 cm bottom of the columns under different magnetized irrigation treatments: (a) HW treatments; (b) DW treatments. In each panel, the left group of bars is the 700 mL delivery time, and the right group of bars is the wetting front advance time. Error bars represent the standard deviation (n = 3). Different lowercase letters within the same group indicate significant differences among treatments (p < 0.05, two-way ANOVA followed by Tukey’s test).

3.2. Influence of Magnetized Water Drip Irrigation on Soil Profile Moisture Content

Three-way ANOVA revealed significant main effects of water source (F(1, 72) = 17.690, p < 0.001), magnetic intensity (F(2, 72) = 8.521, p < 0.001), and soil depth (F(5, 72) = 12.617, p < 0.001), along with significant two-way and three-way interactions (all p < 0.001; model R² = 0.926; Table S1). Magnetization treatment significantly altered the vertical distribution of soil moisture after infiltration (p < 0.05; Figure 7). For HW, HW1 significantly increased moisture content by 8.73% and 23.69% in the 20–40 cm layers relative to HW0, whereas HW2 achieved the highest moisture content in the 50–60 cm layer, representing a significant 41.34% increase (Figure 7a). For subsurface drainage water, DW1 significantly increased moisture content by 42.20% in the 10–20 cm layer relative to DW0; however, both DW1 and DW2 significantly reduced moisture content in the 30–50 cm layers, decreasing by 15.50–28.38% and 12.59–25.15%, respectively. DW2 exhibited the highest moisture content in the 50–60 cm layer, with a significant increase of 7.31% (Figure 7b). Overall, magnetization promoted downward water migration for both water sources, although HW primarily enhanced moisture in the middle-to-deep layers, while subsurface DW favored shallow-layer moisture retention.

Figure 7. Vertical distribution of soil moisture content under different magnetization treatments: (a) HW treatments; (b) DW treatments. Error bars represent the standard deviation (n = 3), and the diamond symbols denote the individual replicate values of the three columns within each treatment. Different lowercase letters indicate significant differences among treatments within the same soil layer (p < 0.05, three-way ANOVA followed by Duncan’s multiple range test).

It should be noted that the changes in soil moisture reported here reflect the redistribution of water within the profile rather than a net increase in soil water retention capacity. This distinction is important because the two processes (increased retention vs. altered redistribution) involve different hydrological mechanisms and have different agronomic implications for root-zone water availability.

3.3. Influence of Magnetized Water Drip Irrigation on Salt Storage Distribution in Soil Column Profiles

The distribution of salt storage across the soil profile under different magnetization treatments is presented in Figure 8. Three-way ANOVA revealed that water source, magnetic intensity, and soil depth all had highly significant main effects on soil salt storage (F(1, 72) = 352.645, F(2, 72) = 144.026, and F(5, 72) = 44.984, respectively; all p < 0.001), with significant water source × depth (F(5, 72) = 4.246, p = 0.002), magnetic × depth (F(10, 72) = 12.440, p < 0.001), and three-way interactions (F(10, 72) = 4.522, p < 0.001; model R² = 0.936; Table S1). The greatest reduction in salt storage occurred in the 40–50 cm layer, where HW1 and HW2 significantly decreased by 16.14% and 24.74%, respectively, compared with HW0 (p < 0.05). Overall, HW1 produced a stronger desalination effect in the shallow layers, whereas HW2 achieved more pronounced desalination in the deep layers. For subsurface drainage water (Figure 8b), the salt reduction effect of magnetization was more evident. In the 50–60 cm layer, salt storage under both DW1 and DW2 was significantly lower than under DW0, with reductions of 29.18% and 30.38%, respectively (p < 0.05).

Figure 8. Salt storage distribution across soil layers under magnetized irrigation treatments: (a) HW treatments; (b) DW treatments. Error bars represent standard deviation (n = 3), and the diamond symbols denote the individual replicate values of the three columns within each treatment Different lowercase letters indicate significant differences among magnetic intensities within the same soil layer (p < 0.05, three-way ANOVA followed by Duncan’s multiple range test).

Under the experimental conditions of this study, considering both water sources together, 3500 Gs produced a consistent direction of change (net salt reduction in the 0–40 cm layer under all magnetized treatments) across the observed soil layers, whereas the salt reduction effect of 5000 Gs was less consistent in the shallow layers.

3.4. Distribution Characteristics of Major Salt Ions in Redistributed Soil

Figure 9 shows the vertical distribution of six major salt ions (Figure 9 a–f) and apparent ion removal rate in the 0–20 cm root zone (Figure 9g). Anions were dominated by HCO₃⁻ and SO₄²⁻, followed by Cl⁻, whereas Ca²⁺ remained relatively stable across depths.

Figure 9. Vertical distribution of major salt ions across treatments (a–f) and their removal rates in the 0–20 cm root zone (g). Error bars represent standard deviation (n = 3) Different lowercase letters indicate significant differences among the six treatments within the same ion group (one-way ANOVA followed by Duncan’s multiple range test, p < 0.05).

Under HW irrigation, both HW1 and HW2 reduced Cl⁻ and SO₄²⁻ concentrations in the 0–30 cm depth range (Figure 9a–c), with highly significant main effects of water source (F(1, 72) = 189.81) and magnetic intensity (F(2, 72) = 35.23) and a significant Water × Magnetic interaction (F(2, 72) = 13.77; all p < 0.001; Table S1); HW2 produced surface reductions of 18.2–46.6%. In the root zone (Figure 9g), one-way ANOVA revealed highly significant differences among treatments for all six ions (F(5, 12) = 13.58–89.89, p < 0.001). HW1 and HW2 significantly enhanced the removal of HCO₃⁻ (34.61% and 31.28%, respectively) and Mg²⁺ (28.20% and 23.33%, respectively) compared with HW0 (p < 0.05). Under subsurface DW irrigation, magnetization exerted a similar regulatory effect on Cl⁻ and SO₄²⁻. DW1 reduced their accumulation within 10–30 cm by 21.5–39.9% relative to DW0 (Figure 9d–f), while water source remained a significant source of variation for Cl⁻ (F(1, 72) = 10.74, p = 0.002; Table S1). In the root zone (Figure 9g), DW1 achieved the highest removal rate for HCO₃⁻ (35.27%), while HW2 exhibited the highest Na⁺ removal (35.83%). Notably, Cl⁻ and Mg²⁺ showed negative removal values under DW0 (−11.79% and −11.27%, respectively), indicating net accumulation rather than net removal in the root zone under unmagnetized DW irrigation. Magnetization consistently reversed this trend, with DW1 improving Cl⁻ and Mg²⁺ removal by 16.5 and 15.9 percentage points relative to DW0 (p < 0.05).

The sodium adsorption ratio (SAR), calculated from Na⁺, Ca²⁺, and Mg²⁺, was significantly affected by water source, magnetic intensity, and soil depth (F(1, 72) = 26.11, F(2, 72) = 21.24, F(5, 72) = 7.58; all p < 0.001), with a significant Magnetic × Depth interaction (F(10, 72) = 5.25, p < 0.001; Table S1). Magnetization consistently reduced SAR under both water sources, with the strongest effect at 3500 Gs. The reduction in extract SAR indicates a lower soluble Na⁺ load in these layers; because ESP, exchangeable Na⁺ and CEC were not measured, this cannot be interpreted as alleviation of exchangeable sodicity.

To further examine how magnetization reshapes the overall hydrochemical composition rather than individual ion concentrations, Piper trilinear diagrams were used to visualize the relative proportions of major ions across treatments (Figure 10). Among cations, Na⁺ had the highest concentration, followed by Ca²⁺ and Mg²⁺; among anions, HCO₃⁻ was dominant, with SO₄²⁻ and Cl⁻ as secondary components. Compared with non-magnetized treatments, the distribution of soil salt ions under magnetized water treatments was notably altered: the proportion of Na⁺ decreased overall, while the proportions of Ca²⁺ and Mg²⁺ increased. Moreover, the data points for the magnetized treatments were more tightly clustered than those for the non-magnetized controls (assessed visually, not quantified), indicating that magnetization not only reduces Na⁺ dominance but also stabilizes the overall hydrochemical facies of the soil profile.

Figure 10. Piper trilinear diagrams of soil salt ions under different magnetization treatments. (a–c) HW treatments; (d–f) DW treatments. Colored dots represent different soil depths (0–10, 10–20, 20–30, 30–40, 40–50, and 50–60 cm). The lower-left and lower-right triangles show the relative proportions of cations and anions, respectively; the central diamond represents the overall hydrochemical facies.

3.5. Multifactor Coupling Analysis of Water–Salt Transport Processes

The redundancy analysis (RDA) results are presented in Figure 11. The first two ordination axes cumulatively explained 56.91% of the total variance. The remaining variation was not explained by the treatment terms and may reflect unmeasured variables, nonlinear relationships, and residual variability among the columns. The permutation test confirmed that both the overall model and each explanatory variable were highly significant (* p < 0.05, ** p < 0.01; Figure 11).

Figure 11. Redundancy analysis (RDA) biplot showing the relationships between soil water–salt properties (response variables) and treatment factors (explanatory variables: water source and magnetic intensity). RDA1 explained 46.68% and RDA2 explained 10.23% of the total variance (cumulatively 56.91%); * indicates p < 0.05, and ** indicates p < 0.01.

The RDA1 axis primarily captured the differentiation effect of water source type. The loadings of SO₄²⁻, Cl⁻, HCO₃⁻, and salt storage on this axis were relatively high and aligned with the positions of the DW treatment groups, indicating that these variables were more closely associated with the drainage-water treatments than with the Yellow River water treatments. Among them, SO₄²⁻ had the largest projection on RDA1, reflecting its strong association with the salinity difference between the two water sources. The RDA2 axis was primarily driven by magnetization intensity. Na⁺ concentration and infiltration rate showed the strongest loadings along the negative direction of this axis, together with the magnetization vector. Moisture content was oriented towards the opposite half of the axis, indicating that it varied along a direction distinct from that of the salt-related variables. The DW0 group exhibited relatively high dispersion and was located in the region indicated by the salt-related vectors. The DW0 group was relatively dispersed and was located in the region indicated by the salt-related vectors, whereas the HW1 and DW1 groups deviated farthest from this cluster. This ordination pattern is descriptive: it does not weight the variables by their agronomic importance and therefore does not identify a single treatment as optimal; the intensity showing the largest response differed among variables and between water sources (Section 3.1).

4. Discussion

4.1. Effects of Magnetized Water on the Infiltration Process in Undisturbed Soil

It should be noted that the scientific literature on magnetized water treatment remains inconsistent. While many studies have reported positive effects on soil infiltration, aggregate stability, and crop yield [17,22,23,24], other investigations have found weak, inconsistent, or non-significant responses [15,28]. The present study contributes to this ongoing debate by providing evidence from undisturbed saline–alkali soils under two contrasting water sources.

In this study, the largest infiltration response among the three tested intensities was observed at 5000 Gs for HW and at 3500 Gs for DW, which is broadly consistent with the findings of Duan et al. [24] and Wang et al. [32]. However, the water source × magnetic intensity interaction was not statistically significant for mean infiltration rate, steady-state infiltration rate, or infiltration duration; the interaction was significant only for wetting front advance time. The differences between the two water sources should therefore be interpreted as differences in the magnitude of the observed response to the specific intensities tested rather than as evidence of a source-specific optimum.

This difference may be related to the nonlinear response of water physicochemical properties and ion hydration behavior to magnetic treatment. According to hydrogen bond breaking theory, Lorentz force depolymerizes macromolecular clusters into smaller, more reactive clusters, thereby reducing surface tension and viscosity [20,21]. Within the complex pore network of undisturbed soil, reduced surface tension lowers capillary flow resistance, facilitating water passage through micropores between aggregates [17]. In the low-mineralization Yellow River water, the higher of the two tested intensities produced the largest infiltration response, whereas in the ion-rich drainage water, the intermediate intensity did. One possible explanation, suggested by work on electrolyte solutions [23], is that dissolved ions interact with both the water structure and the applied field; this was not tested in the present experiment and remains a hypothesis. The weaker response of DW2 compared with DW1 may be related to the higher ionic strength of subsurface DW, which could modify ion hydration and water–soil interactions under magnetic treatment. However, this mechanism requires further verification through direct measurements of water physicochemical properties [33,34]. For the high-salinity drainage water, the most effective of the tested intensities was 3500 Gs. Because only three discrete intensities were compared, this identifies the best-performing level under the present conditions rather than a general threshold. The infiltration enhancement observed is smaller than values typically reported for repacked column studies [22,26]. One hypothesis is that repacking homogenizes the soil matrix and removes natural aggregate structure, biopores and preferential flow pathways [27], which could alter the apparent magnitude of the magnetic effect. This was not tested directly in the present experiment, and the comparison should therefore be treated as a hypothesis rather than as an established explanation. In contrast, undisturbed columns preserve compaction layers, macropores, and heterogeneous pore networks that more realistically constrain water movement under field conditions. This methodological distinction represents a key advantage of the present study and supports the practical relevance of our results. Nevertheless, it should be acknowledged that even undisturbed columns cannot fully replicate field-scale processes such as evapotranspiration, rainfall interception, and groundwater interaction.

Although both 3500 and 5000 Gs significantly improved infiltration relative to the control, they did not differ significantly from each other. The selection between them cannot be made on statistical grounds alone. For Yellow River water, 5000 Gs produced larger deep-layer salt reduction and moisture increase, suggesting that the higher intensity may favor deeper redistribution under low-salinity irrigation. For drainage water, 3500 Gs yielded stronger infiltration enhancement and comparable salt reduction. Considering infiltration, salt leaching, and energy input together, 3500 Gs appears more favorable for high-salinity water, while 5000 Gs may be preferable for low-salinity water when deep leaching is prioritized. These are operational recommendations based on observed response patterns rather than statistically demonstrated optima.

4.2. Water–Salt Redistribution and Desalination Mechanism Under Magnetization

Magnetization treatment promoted the downward redistribution of salts within the profile, producing a favorable distribution pattern in which root-zone salinity was considerably reduced relative to deeper layers [4,18,35]. Three-way ANOVA confirmed that water source, magnetic intensity, and soil depth all significantly affected salt storage (F = 352.645, 144.026, and 44.984, respectively; all p < 0.001), with significant Water × Depth and Magnetic × Depth interactions (both p ≤ 0.002). The underlying mechanism of the enhanced downward salt transport effect could potentially be associated with changes in the interaction between water molecules and dissolved ions under magnetic treatment. Previous studies have proposed that magnetic fields may weaken electrostatic interactions between ions and water molecules and possibly compress the ionic hydration shell [36]. The resulting reduction in hydration radius could increase the migration coefficients of Na⁺ and Cl⁻, rendering these ions more susceptible to leaching into deeper layers. This dual mechanism—enhanced water mobility coupled with reduced ion hydration—may together account for the marked increase in deep-layer moisture (e.g., 41.34% at 60 cm under HW2) and the simultaneous decrease in shallow salt storage across treatments. Salt accumulation in the 50–60 cm layer was more pronounced than is typically reported for repacked soil. One possible explanation is that the continuous pore network of the undisturbed columns allowed more rapid downward water movement and salt transport towards the wetting front; because pore architecture and preferential flow were not measured directly, this remains speculative. Although magnetization enhanced downward salt movement, some deeper layers showed higher salt storage than the layers above them [37]. No upward water or salt flux was measured in this single-irrigation experiment, so this cannot be described as salt backflow; whether repeated irrigation would lead to accumulation at depth requires multi-season field observation. Coordinated design of magnetized water irrigation with subsurface drainage systems is therefore recommended to mitigate this risk. However, because we did not directly measure water physicochemical properties (e.g., surface tension, viscosity, NMR spectra) or ion hydration parameters, this mechanistic interpretation should be regarded as a working hypothesis supported by literature rather than a demonstrated process. Further studies combining macroscopic infiltration experiments with molecular-scale measurements are needed to verify this pathway.

4.3. Ion-Selective Migration and Implications for Saline–Alkali Soil Management

Magnetization treatment demonstrated notable selectivity in ion migration. Following magnetized irrigation, HCO₃⁻ remained the dominant anion across all treatments, while Cl⁻ and SO₄²⁻ concentrations decreased in certain treatments. The proportion of Na⁺ generally declined, whereas Ca²⁺ and Mg²⁺ proportions slightly increased. These results indicate that magnetized water promotes the downward migration of Na⁺ within the profile while increasing the relative proportions of Ca²⁺ and Mg²⁺ in the soil extract, thereby optimizing soil ion composition [29]. Similar cation exchange phenomena have been reported in saline–alkali soil remediation studies. Diffuse double-layer theory and previous studies suggest that magnetic treatment may influence ion hydration behavior and colloid–ion interactions [38,39], which could partly explain the observed changes in ion redistribution. After magnetization, the hydration shell of Na⁺ may be compressed, potentially enhancing its mobility [40]. In contrast, divalent cations such as Ca²⁺ and Mg²⁺ have smaller hydration radii and a stronger affinity for negatively charged surfaces, which could favor their retention relative to Na⁺. From the perspective of saline–alkali soil amelioration, a lower exchangeable sodium percentage (ESP), together with a higher Ca²⁺/Mg²⁺ saturation, is widely recognized as a prerequisite for restoring aggregate stability and hydraulic conductivity in sodic soils [41,42,43]. If the shifts observed in the soil water extract were accompanied by comparable changes in the exchange phase, magnetized irrigation could act as a low-cost complement to existing amelioration measures. Because CEC, ESP and exchangeable cations were not measured here, this remains a hypothesis to be tested. For the Hetao Irrigation District, where secondary salinization and freshwater scarcity coexist, these findings raise two questions for field testing rather than management recommendations: whether 5000 Gs magnetization of Yellow River water can enhance pre-sowing salt leaching, and whether 3500 Gs magnetization combined with the existing subsurface drainage network could allow drainage water to be reused while limiting shallow Na⁺ accumulation. This study has several limitations. Indoor undisturbed columns cannot fully reproduce field processes such as evapotranspiration, rainfall, groundwater fluctuation, and root water uptake, and a single irrigation event cannot capture long-term cumulative effects on soil structure and salt balance. Moreover, factors potentially modulating magnetization efficiency (transport distance, temperature, and time lag before irrigation), crop responses, soil microbial and enzymatic dynamics, and other soil types were not addressed here. Future work should combine multi-season, multi-soil field trials with soil microstructural imaging (e.g., CT scanning) and direct measurements of water physicochemical properties to further elucidate the underlying mechanisms. Two further constraints should be noted: each treatment was replicated only three times, which limits the power available to detect interaction effects, and the absence of leachate analysis precluded a complete ion mass balance.

5. Conclusions

Magnetization altered water infiltration in undisturbed saline–alkali soil, and the magnitude of the response differed between the two irrigation water sources. Among the three intensities tested, HW1 and HW2 increased the mean infiltration rate by 11.6% and 12.9%, respectively; DW1 and DW2 increased it by 13.90% and 10.4%, respectively. Although both magnetized intensities significantly improved infiltration relative to the non-magnetized control, they did not differ significantly from each other for infiltration rate or duration (p = 0.903, Table S2). The selection between 3500 and 5000 Gs therefore cannot rest on statistical grounds alone. Considering infiltration enhancement, salt redistribution, moisture retention, and energy input together, 3500 Gs appears more favorable for high-salinity subsurface drainage water under the present conditions, whereas 5000 Gs may be preferable for low-salinity Yellow River water when deep salt leaching is prioritized. These are operational recommendations based on observed response patterns across multiple variables rather than statistically demonstrated optima. Magnetized irrigation was associated with redistribution of salts from shallow layers towards the deeper profile, with salt storage in the 40–50 cm layer under HW2 decreasing by 24.74% and in the 50–60 cm layer under DW1 decreasing by 29.18%. Regarding moisture content, HW2 increased it by 41.34% at 60 cm depth, and DW1 increased it by 42.20% at 10–20 cm. The ionic composition of the soil water extract shifted towards a lower relative proportion of Na⁺ and Cl⁻ and a higher relative proportion of Ca²⁺ and Mg²⁺. Redundancy analysis separated the samples mainly along two axes, the first associated with water source and the second with magnetic intensity; the two axes together explained 56.91% of the total variance. These findings are based on a single irrigation event in undisturbed laboratory columns and should not be extrapolated to other soil types, water qualities, or irrigation systems without field-scale validation.