FORECASTING SOIL SALINITY DYNAMICS UNDER CONSTANT WATER MINERALIZATION AND REDUCED IRRIGATION VOLUMES (KHOVOS DISTRICT, 2023–2030)

АННОТАЦИЯ

В исследовании выполнена прогностическая оценка динамики засоления корнеобитаемого слоя в районе Ховос (2023–2030) при сценариях поэтапного сокращения годовых объёмов орошения при неизменной минерализации поливной воды. Используется прозрачный индикатор солевого баланса и линейная регрессия год‑к‑году (OLS), что позволяет количественно оценить тенденцию роста засоления и условия перехода участка в более высокую категорию засолённости при отсутствии достаточного выщелачивания и эффективной работы дренажа. Явно сформулированы допущения, диапазоны входных параметров и неопределённости. Показано, что «водосбережение» должно сопровождаться управлением солевым балансом: постоянным мониторингом качества воды, сезонными выщелачивающими поливами и подтверждённой работоспособностью дренажной сети для сохранения мелиоративного состояния земель.

ABSTRACT

This study presents a predictive assessment of root‑zone salinity dynamics in the Khovos district (2023–2030) under scenarios that gradually reduce annual irrigation volumes while keeping water mineralization unchanged. Using a transparent salt‑balance indicator and an ordinary least squares fit of year‑to‑year changes, we estimate a steady increase in soil salinity and identify conditions under which fields may shift to a higher salinity class if salt inputs are not offset by leaching and drainage. Assumptions, input ranges, and uncertainty are stated explicitly. The findings emphasize that “water‑saving” must be coupled with salt‑balance control—continuous water‑quality monitoring, seasonal leaching, and verified drainage performance—to sustain the meliorative state.

Ключевые слова: засоление почв; минерализация поливной воды; водосберегающий сценарий; поступление солей; район Ховос; прогнозирование

Keywords: soil salinity; irrigation water mineralization; water‑saving scenario; salt input; Khovos district; forecasting.

1. Introduction

Soil salinization is a growing threat to agricultural productivity and environmental sustainability, particularly in arid and semi-arid regions where irrigation is essential. In such areas, the interaction between irrigation practices and water quality plays a crucial role in determining the salinity dynamics of soils. The Khovos District, characterized by intensive land use and reliance on irrigation, faces increasing challenges from soil salinity due to persistent water mineralization and recent reductions in irrigation volumes driven by water scarcity and conservation policies.

Forecasting soil salinity trends is critical for long-term land management, especially under conditions where water mineralization remains constant and irrigation inputs decrease. These conditions can alter the salt balance in the soil profile, potentially accelerating salinization and reducing land fertility. Despite numerous studies on salinity modeling, there remains a gap in region-specific forecasts that incorporate local climatic, hydrological, and agronomic variables.

This paper aims to model and forecast soil salinity dynamics in the Khovos District for the period 2023–2030 under two primary conditions: constant mineralization of irrigation water and reduced irrigation volumes. The objective is to provide a predictive framework that can inform sustainable irrigation management, mitigate salinization risks, and support adaptive agricultural practices in the region.

Soil salinity is a critical constraint on agricultural productivity, particularly in arid and semi-arid regions dependent on irrigation. Salinization is exacerbated by poor drainage, shallow groundwater, and irrigation with saline water. The challenge of predicting salinity under conditions of constant water mineralization and reduced irrigation volumes has led to the development of various modeling and monitoring approaches.

Dynamic models such as Support Vector Machines (SVMs) have been shown to effectively predict soil salinity levels using input variables like groundwater depth, irrigation water volume, and evaporation. These models outperformed traditional artificial neural networks (ANNs), with a mean relative prediction error of just 6.37% [1].

State-space models grounded in mass balance and empirical data offer another approach for forecasting salinity in soils influenced by shallow groundwater. These models performed well in predicting water content and salinity dynamics even with limited sampling frequency [2].

In Ethiopia’s Rift Valley, ANNs demonstrated superior predictive accuracy (R² = 0.77) for topsoil salinity compared to Partial Least Squares Regression (R² = 0.45), identifying groundwater table depth as the most influential factor [3].

Experimental studies have shown that irrigation with moderately saline water (e.g., 3 g/L) leads to salt accumulation at depths of 100 cm, especially in the absence of rotational irrigation with freshwater [4]. Similarly, simulation models applied to South African catchments highlight that soil salinity is sensitive to hydrological parameters such as infiltration and drainage capacity, which can be integrated into global prediction models [5].

The relationship between groundwater depth and soil salinity was confirmed in a 15-year study in China, where soil salinity increased in areas with shallow water tables. The research suggested a threshold depth of 2.5–5 meters below which salinization becomes significantly active [6].

Salt transport simulations under drip irrigation in the Taklimakan Desert showed increased salinity at greater depths, depending on soil texture. Loam and clay soils exhibited increased accumulation below 60 cm, emphasizing the importance of depth-specific modeling [7].

Recent spatial analyses from Northwest China concluded that maintaining a groundwater depth of 4.0–6.0 m and irrigation volumes around 5,500–6,000 m³/ha/year can mitigate secondary salinization in arid irrigation zones [8].

On a global scale, remote sensing and Bayesian modeling revealed that advanced irrigation methods such as drip irrigation combined with institutional water management reduce salinization over time, particularly in regions with high water stress [9].

The literature consistently demonstrates that soil salinity dynamics are influenced by a combination of irrigation volume, water mineralization, groundwater depth, and soil texture. Under scenarios of constant water mineralization and reduced irrigation volumes—like those projected for the Khovos District—salinity levels are likely to rise unless proactive management strategies are implemented. Forecasting tools such as support vector machines, artificial neural networks, and geostatistical models provide valuable predictive capabilities, especially when integrated with field-specific parameters. Maintaining optimal groundwater depths and adapting irrigation techniques (e.g., rotational or drip irrigation) are critical to controlling salt buildup. Together, these insights highlight the need for data-driven, site-specific approaches to forecast and manage soil salinity for sustainable agricultural productivity.

2. Materials and Methods

2.1. Study area and inputs

The analysis covers fields in the Khovos district (Turkestan and E. Kakkharov farms). Soils are medium loam with a dry bulk density ρ = 1400 kg·m⁻³ (typical of the area). Baseline root‑zone salinity at the start of the projection is S₂₀₂₃ = 0.800% (moderately salted). Irrigation‑water mineralization in practice varies within 1.14–3.99 g·L⁻¹; for projections we fix C = 2.5 g·L⁻¹ (i.e., 2.5 kg·m⁻³). Annual irrigation volume V is assumed to decline gradually from 3220 → 3000 m³·ha⁻¹ (2023→2030) as a representative water‑saving scenario.

Data provenance. Irrigation-water mineralization C was sourced from the MIS portal of the Ministry of Water Resources of Uzbekistan (MIS Portal): https://mis-portal.mwr.uz/report/table (accessed 2025-10-27).

2.2. Deterministic screening indicator

We define the annual salt input to the root zone as:

                                                                        (1)

where

 — mineralization of irrigation water,kg/m³;

— the volume of irrigation water supplied during the year,m³/ha.

If  given for 1ha:

                                                             (2)

For interpretation, a simplified screening indicator of salinity can be written as:

                                                                               (3)

where h is the considered root‑zone thickness (m) and ρ is dry bulk density (kg·m⁻³).

The year-to-year increase in the salinity level can be expressed mathematically as follows:

 , t=2024,2025, …                                                             (4)

2.3. Empirical projection of salinity

Year‑to‑year salinity is projected with an ordinary least squares (OLS) trend anchored at S₂₀₂₃ = 0.8 %:

                                                      (5)

This slope arises from the scenario endpoint S₂₀₃₀ = 1.573% (i.e., ). We report k and display the trend in Fig. 1. Because this series is a deterministic scenario (not stochastic field replicates), formal confidence intervals are not inferable; the uncertainty discussion is moved to Limitations.

2.4. Root‑zone depth and worked example

Unless stated otherwise, we illustrate calculations at h = 0.50 m and ρ = 1400 kg·m⁻³, which are consistent with the baseline S₂₀₂₃ order of magnitude.

Worked example (2025):

1) 

2) 

3) 

4) 

5) 

with C = 2.5 kg·m⁻³ and V = 3150 m³·ha⁻¹, A screening estimate would be 0.1125 % . The projected value from Eq. (5) is 1.021 %. The difference reflects accumulation dynamics and leaching not explicitly modeled in ; We retain Eq. (5) as the formal projection and use s to interpret parameter sensitivity.

Table 1.

Categories of soil salinity

Table 2.

Calculation of soil salinity level

Notes. In Table 2, (M_in) follows Eq. (1). (S_t) follows Eq. (4) and is the plotted series in Fig. 1. The slight decline in with reduced V coexists with a rising because Eq. (4) and Eq. 5 capture net accumulation under insufficient leaching—a process not encoded in the one‑year screening formula (3).

2.5. One‑factor sensitivity (±20%) around 2025 baseline

We quantified which inputs most affect the screening indicator by applying a one-factor perturbation of ±20% around the 2025 baseline, holding the other inputs constant. The baseline represents mid-horizon conditions and is defined as: C = 2.5 kg·m⁻³, V = 3150 m³·ha⁻¹, h = 0.5 m, ρ = 1400 kg·m⁻³.

Computation. For each input X ∈ {C, V, h, ρ}, we compute s(%) = (M_in/(h·ρ)) × 100, with M_in =,[t/ha] using the units shown above. We then evaluate s at X·1.2 and X·0.8, keeping the other three inputs at baseline. This yields the values reported in Table 3 / Fig. 3.

Rationale for ±20%. In the absence of precise priors for parameter variance, a ±20% envelope is a standard engineering/scenario tolerance that approximates plausible operational deviations in water quality (C), seasonal volume (V), and effective leaching/storage (h, ρ). The analysis reveals proportional relationships: s ∝ C and s ∝ V; s ∝ 1/h and s ∝ 1/ρ. Accordingly, increasing h (via seasonal leaching and functioning drains) or lowering C (source mixing/treatment) most effectively reduces the risk signal.

Using 2025 inputs (C = 2.5 kg·m⁻³, V = 3150 m³·ha⁻¹, h = 0.5 m, ρ = 1400 kg·m⁻³), the screening metric responds as follows:

Table 3.

One‑factor sensitivity

Implication. Controlling C (mixing/blending sources, treatment) or increasing effective leaching depth (larger h via seasonal leaching events and functioning drains) most reduces the risk signal.

3. Results

3.1. Trend of projected salinity (Fig. 1)

The OLS projection from Eq. (5) yields an approximately linear increase from 0.800% (2023) to 1.573% (2030) with slope k ≈ 0.110429 %·yr⁻¹. Each point represents the projected mean for that year under the stated assumptions (constant C, steadily reduced V, and no additional leaching program). The straightness of the line reflects the deterministic scenario. In practice, seasonal management could bend this trajectory downward (see Discussion).

Figure 1. Projected root‑zone salinity S(%) for 2023–2030 under constant irrigation‑water mineralization (C = 2.5 kg·m⁻³) and gradually reduced volumes (V). Model: OLS trend; slope k (reported on plot); authors’ calculations

3.2. Salt input vs. projected salinity (Fig. 2)

While M_in declines slightly (8.05→7.50 t·ha⁻¹),  rises (0.800→1.573%). This is consistent with a setting where leaching is insufficient to flush carry‑over salts as volumes are reduced. Thus, reducing V alone—without complementary leaching events and verified drainage performance—does not guarantee stabilization of salinity.

Figure 2. Comparison of annual salt input M_in (bars, left axis) and projected salinity S_t (line, right axis). Authors’ calculations

3.3. Sensitivity summary (Fig. 3,Table 3)

The ±20% one‑factor analysis (around 2025) indicates the strongest levers on the screening metric are C, h, and ρ (the latter two reflecting the effective leaching/storage capacity of the root zone). Operationally, lowering C and/or scheduling seasonal leaching (increasing effective h) are the most direct ways to offset rising S_t.

Figure 3. Tornado chart of the screening metric s at 2025 baseline showing the influence of ±20% changes in C, V, h, ρ. Authors’ calculations

4. Discussion

1. Why S_t rises while M_in falls. With reduced V, percolation and leaching capacity also diminish, so salts previously stored in the profile are not flushed effectively. The empirical trend (Eq. 5) embeds this net accumulation. Hence, volume savings must be paired with measures that increase effective leaching and maintain drainage outflows.
2. What controls matter most. Sensitivity shows that managing C (source mixing, periodic checks), and episodically increasing h via seasonal leaching irrigations (timed when the drainage network is functional) are the most influential levers; simply cutting V is insufficient.
3. How to operationalize. We recommend a trigger‑based plan: if in‑season EC/EM readings or the projected S_t exceed a threshold (e.g., crossing from Class 3 to Class 4), schedule a leaching irrigation and verify the drainage response (water levels, flows, EC at outlets) within the same season. Keep a minimal leaching fraction in routine scheduling when feasible (Table 1).
4. Monitoring package. (i) Quarterly ECₑ profiles at fixed points; (ii) irrigation‑water EC/TDS logs per source; (iii) drain flow and EC logs during/after irrigation and leaching; (iv) satellite indices (e.g., NDVI with a salinity‑sensitive index) to prioritize ground checks.

5. Limitations

The screening indicator omits explicit leaching fractions, drainage fluxes, capillary rise, and intra‑season redistribution; Eq. (5) is an empirical trend, not a mechanistic salt‑balance solution. Therefore, projections should be interpreted as conservative until field‑validation (repeat ECₑ profiles, drain monitoring) confirms local responses. Extending the framework with a mass‑balance or HYDRUS‑type model is a logical next step when data allow.

6. Conclusions

Under constant irrigation‑water mineralization (C = 2.5 kg·m⁻³) and gradually reduced volumes (V), the projection indicates a rise in salinity from 0.800% (2023) to 1.573% (2030) (slope ≈0.110429 %·yr⁻¹), moving from moderately to strongly salted if unmanaged. To avoid adverse meliorative outcomes, water‑saving programs must be coupled with:

• routine water‑quality monitoring and source management (lower C where possible),
• scheduled seasonal leaching tied to drainage capacity,
• continuous drainage performance verification (water levels/flows/EC), and
• trigger‑based adjustments when the projected  approaches class thresholds.

References:

1. Guan X., Wang S., Gao Z., Lv Y. Dynamic prediction of soil salinization in an irrigation district based on the support vector machine. Mathematical and Computer Modelling. 2013;58(3–4):719–724. doi:10.1016/j.mcm.2011.10.026.
2. Wu L., Skaggs T.H., Shouse P.J., Ayars J.E. State space analysis of soil water and salinity regimes in a loam soil underlain by shallow groundwater. Soil Science Society of America Journal. 2001;65(4):1065–1074. doi:10.2136/sssaj2001.6541065x.
3. Gelaye K.K., Zehetner F., Stumpp C., Dagnew E.G., Klik A. Application of artificial neural networks and partial least squares regression to predict irrigated land soil salinity in the Rift Valley Region, Ethiopia. Journal of Hydrology: Regional Studies. 2023;46:101354. doi:10.1016/j.ejrh.2023.101354.
4. Yu S. (Shi-peng Y.). Effects of rotational irrigation with saline water on soil salinity and crop yield. Journal of Irrigation and Drainage (in Chinese). 2010;29(6):83–86. (Title/journal verified; article in Chinese with English abstract.)
5. Elenius M., Bartošová A., Musuuza J.L., Arheimer B. Salinization sources and management strategies in South Africa. EGU General Assembly 2020 (abstract). 2020. doi:10.5194/egusphere-egu2020-9743.
6. Lian H., Sun Z., Xu C., Gu F. The relationship between the distribution of water and salt elements in arid irrigation areas and soil salination evolution. Frontiers in Earth Science. 2022;10:852485. doi:10.3389/feart.2022.852485.
7. Yang H., Chen Y., Zhang F., Xu T., Cai X. Prediction of salt transport in different soil textures under drip irrigation in an arid zone using the SWAGMAN Destiny model. Soil Research. 2016;54(8):869–879. doi:10.1071/SR15169.
8. Li W., Kang S., Du T., Ding R., Zou M. Optimal groundwater depth and irrigation amount can mitigate secondary salinization in water-saving irrigated areas in arid regions. Agricultural Water Management. 2024;302:109007. doi:10.1016/j.agwat.2024.109007.
9. Shi H. To reduce soil salinity: the role of irrigation and water management in global arid regions across development phases. arXiv preprint. 2022; arXiv:2204.02029. (Preprint.)