Skip to content
Frontiers
Search

Original Research

Frontiers in Earth Science ·

Volume 14 - 2026 | https://doi.org/10.3389/feart.2026.1880109

Seasonal differentiation and drivers of vertical soil moisture variation in farmland of the middle Heihe River

  • Wenye Li1,2,3
  • Wenpeng Li1
  • Yuejun Zheng1*corresponding author
  • Haoming Wang3,4,5,6,7
  • Xinlei He8
  • Xusheng Wang2
  • Yan Wen9
  • Rongchang Zhao3,4,5,6,7
  • Jiayin Lu10
  1. 1Department of Water Resources, China Institute of Geo-Environment Monitoring, Beijing, China
  2. 2School of Water Resources and Environment, China University of Geosciences (Beijing), Beijing, China
  3. 3Water Cycle Field Station of The Heihe River Basin, China Geological Survey, Zhangye, China
  4. 4Scientific Research and Innovation Center, Institute of Hydrogeology and Engineering Geology Gansu Provincial Bureau of Geology and Mineral Exploration and Development, Zhangye, China
  5. 5Gansu Geological Engineering Survey Institute Co., Ltd., ZhangYe, China
  6. 6Gansu Inland Arid Area Water Cycle and Water Balance Field Scientific Observation and Research Station, Zhangye, China
  7. 7Geological Society of China Technology Innovation Base for Water Cycle Monitoring and Water Balance Analysis in Inland Arid Areas, Zhangye, China
  8. 8Institute of Loess Plateau, Shanxi University, Taiyuan, China
  9. 9College of Civil and Architectural Engineering, North China University of Science and Technology, Tangshan, China
  10. 10Non-coal Mine Safety Supervision and Management Section, Songxian County Emergency Management Bureau, Songxian, China
Article metrics0 viewsView details
PDFCiteMetrics

Abstract

Soil water scarcity constitutes a pivotal environmental challenge across arid and semi-arid zones both in China and globally. Driven by ongoing climate change coupled with anthropogenic disturbances, such drought-prone areas keep undergoing continuous spatial expansion. The middle reaches of the Heihe River, as a typical irrigated oasis area in arid area, face prominent contradictions between water supply and demand. Farmland soil moisture dynamics play a vital role in ecological security and agricultural sustainability. Understanding the spatiotemporal characteristics and driving mechanisms of soil moisture is urgently needed to optimize agricultural water resource management. Based on multi-depth soil moisture and meteorological observation data from 2012 to 2024 at the Daman farmland station, redundancy analysis (RDA), multiple linear regression, and hierarchical cluster analysis (HCA) were comprehensively employed to systematically identify the driving factors of soil moisture during the growing and non-growing seasons and characterize their seasonal differentiation patterns. We concluded: 1) Soil moisture is significantly greater variability in shallow soil of 2–20 cm than in deep soil of 120–160 cm, and distinct differences in moisture dynamics between different years; 2) RDA analysis indicated that during the growing season, soil moisture variation was mainly correlated with net radiation (Rn) and air temperature (Ta), whereas in the non-growing season it was dominated by soil temperature (Ts) and photosynthetically active radiation (PAR); 3) Regression analysis further showed that Rn had the greatest impact on the moisture of shallow soil, and Ts showed a significant influence during the non-growing season; 4) Cluster analysis revealed distinct seasonal differentiation features: high temperature and humidity in the growing season, and low temperature water storage in the non-growing season, reflecting the region’s typical climatic characteristics. Soil moisture in the middle reaches of the Heihe River is correlate with energy inputs and water replenishment, with markedly different driving mechanisms between seasons. During the growing season, irrigation should be strengthened to mitigate water loss caused by high temperature and intense radiation, while in the non-growing season emphasis should be placed on low-temperature water retention. These findings provide a scientific basis for sustainable water resource utilization and agricultural water management.

Keywords

1 Introduction

Over 38% of the world’s population currently lives in arid and semi-arid regions, which, driven by climate change and human activities, are expanding at a rate of about 1%–3% per year and are projected to cover 50% of the Earth’s land surface by 2,100 (Huang et al., 2016; Jingyu et al., 2020; Guan et al., 2025). However, due to sparse precipitation and high evaporation, arid regions face prominent water supply–demand imbalances. In particular, in inland river basins of arid zones such as the middle reaches of the Heihe River, water resource limitations have led to pronounced ecological vulnerabilities, manifested by soil degradation, loss of vegetation biodiversity, and water shortages. It has become a key focus for sustainable development and climate adaptation. In addressing the imbalance between water supply and demand, soil moisture represents a valuable and exploitable resource. As a critical variable in terrestrial hydrological cycles, it plays a pivotal role in regulating the exchange of energy, water, and carbon between the land surface and the atmosphere (Shiao et al., 2026; Ching-Min et al., 2025).

The majority of studies posit that soil moisture dynamics are driven by complex and interrelated hydrometeorological processes (Botter et al., 2007; Ridolfi et al., 2003; Wang et al., 2017). Factors such as precipitation, wind speed, and air temperature can influence soil moisture over various timescales (Eunsang and Minha, 2014). Specifically, precipitation explains 27.54% of the variance in surface soil moisture (Wang et al., 2026a). However, a widespread decline in soil moisture has been documented globally (Martin et al., 2010), accompanied by the emergence of ‘soil flash droughts’ characterized by rapid desiccation. While some studies attribute these events primarily to atmospheric flash droughts (Yinglan et al., 2019), earlier research posits an opposing causality: that limited soil water availability is the principal driver of the recent decline in global terrestrial evapotranspiration at short temporal scales (<10 years) (Ma and Yuan, 2025). This apparent contradiction highlights the ongoing debate regarding the dominant direction of land-atmosphere coupling.

Farmland soil moisture, defined as the liquid water stored in soil pores within the crop root zone, acts as a critical nexus linking terrestrial hydrological cycles with crop physiological processes (Jinxia et al., 2026). It not only provides the fundamental substrate for photosynthesis and transpiration but also serves as a vital carrier for soil nutrient transport (Huimin et al., 2026; Melo et al., 2026). Existing studies have indicated that soil moisture in the 0–10 cm layer is particularly important for farmland and grasslands, whereas forest and shrub vegetation growth relies more on 10–20 cm soil moisture (Zheng et al., 2025). However, under arid conditions, vegetation increasingly depends on deeper soil moisture (Zheng et al., 2025). Following the cessation of rainfall or irrigation, soil water undergoes redistribution driven by gravity, matric potential, and temperature gradients, exhibiting a pronounced hysteretic effect (Fan et al., 2024; George et al., 2021); that is, changes in soil moisture lag behind variations in meteorological factors, and this hysteresis becomes increasingly significant with soil depth (Meng et al., 2025; Xu et al., 2026). Consequently, surface soil moisture responds quickly to short-term climate events like precipitation, whereas deep soil moisture dynamics are more influenced by long-term hydrological processes, reflecting longer-term water deficits (Xu et al., 2021).

In the Heihe River basin, climate warming has triggered glacier retreat, reducing lake regulation capacity (Triplett and Condon, 2023; Zhan et al., 2014), causing downstream vegetation degradation, and affecting soil water stability. In addition, rapid expansion of farmland has led to a significant increase in irrigation of water use and groundwater extraction (Fishman et al., 2011; Cai et al., 2003). Furthermore, it is a typical oasis irrigation area in China, with agriculture accounting for over 90% of total water use (Cai et al., 2022; Liu et al., 2022; Zhang et al., 2025). However, evapotranspiration far exceeds infiltration (Hakami-Kermani and Hossein, 2025; Yao et al., 2023; Wang et al., 2024). The dynamics of farmland soil moisture in the middle reaches directly affect ecosystem stability and agricultural sustainability. Previous studies have widely explored the effects of factors such as precipitation, temperature, evapotranspiration, and vapor pressure deficit on soil moisture (Yinglan et al., 2019; He et al., 2020; Kang et al., 2025), providing important references. However, most existing studies have been limited to analyses of part of influencing factors and have predominantly focused on hydrological variations during the growing season. Consequently, there is an urgent need to conduct comprehensive coupling analyses that integrate multiple factors across diverse temporal dimensions, particularly in the highly agriculturally developed middle reaches of the Heihe River.

This study focuses on a typical seed maize farming area in the middle reaches of the Heihe River as the research site. By using high-frequency, multi-dimensional, long-term meteorological and soil moisture observation data,it aims to identify the dominant driving factors and evaluate their contributions across different soil depths and during the growing and non-growing seasons. We reveal the regulatory mechanisms of soil moisture dynamics and the coupling between soil profile moisture changes and meteorological factors. Furthermore, we classified the study years based on environmental conditions to characterize the soil moisture regimes and climatic backgrounds associated with each year type. Thereby, we elucidated the climatic features of the middle reaches and the driving factors of farmland water dynamics under climate change. Ultimately, this study provides a scientific basis for climate regulation, soil water resource utilization, irrigation management, drought early warning, and the optimal allocation of water resources in the middle Heihe River.

2 Study area and data sources

2.1 Study area

The study site is Daman farmland station (Figure 1), which is located in Xiaoman Town, Wuxing Village, Daman Irrigation District, Zhangye City, Gansu Province—a flat-terrain oasis farmland station representative of mid-latitude inland oasis seed-maize agriculture in the middle Heihe River basin. It exhibits a temperate continental arid climate, characterized by an annual precipitation of 125 mm, evaporation of 2047.9 mm, mean temperature of 7.3 °C, and relative humidity of 52%. Cultivation at the station is dominated by maize, with soils identified as oasis alluvial. Positioned at 100°37′E, 38°86′N and elevated approximately 1,560 m, the site represents typical seed-maize farmland in the mid-latitude inland oasis of the middle Heihe River basin and functions as a key observation point for regional hydrological and agricultural studies.

Study site location map. **(a)** Location map of Daman farmland station; **(b)** Field photograph of the underlying surface at Daman farmland station; **(c)** Diagram of the meteorological element gradient observation system.
Figure 1. Study site location map. (a) Location map of Daman farmland station; (b) Field photograph of the underlying surface at Daman farmland station; (c) Diagram of the meteorological element gradient observation system.

2.2 Data sources

The data used in this study are from the meteorological element gradient observation system at the Daman superstation in the middle reaches of the Heihe River, which is part of the Heihe River comprehensive observation network dataset (Liu et al., 2018; Liu et al., 2023). Located precisely at the core of the representative local maize cropland, the meteorological element gradient observation system automatically records observational data at 10-min intervals, with a 13-year continuous dataset from 2012 to 2024 utilized in this research. All observed parameters are listed in Table 1. Monitoring a total of 21 variables, the system spans a complete framework that extends from atmospheric dynamic and thermal conditions to surface energy balance, and from moisture inputs to soil properties. This configuration delivers comprehensive data support for investigating land–atmosphere interactions at the regional scale. Maize is cultivated annually from April to late September, with a growing season spanning 130–160 days and continuous irrigation applied throughout (Ogunsola et al., 2025; Feng et al., 2023; Wang M. et al., 2026). Phenologically, the crop development is segmented into six distinct phases: Sowing Stage (SOS), Seedling Stage (SES), Jointing Stage (JOS), Tasseling Stage (TAS), Grain Filling Stage (GFS), and Maturity Stage (MAS).

TABLE 1. Table of research indicators and testing instruments.

CategoryMonitoring indicatorMonitoring instrumentInstallation location and directionSymbol
Temperature-relatedAir temperature/(°C)Temperature and humidity sensorAt a height of 1.5 m, facing due northTa
Surface radiant temperature/(°C)Infrared thermometerAt a height of 1.5 m, vertically downward facing due southIRT
Soil Temperature/(°C)Soil temperature probeMean temperatures at depths of 2 cm, 4 cm, 10 cm, 20 cm, 40 cm, 80 cm, 120 cm, and 160 cm below the ground surfaceTCAV
Humidity - relatedRelative humidity/%Temperature and humidity sensorAt a height of 1.5 m, facing due northRH
Precipitation/mmTipping - bucket rain gaugeAt a height of 0.7 mRain
Soil moisture/% (v/v)Soil moisture sensorAt depths of 2 cm, 4 cm, 10 cm, 20 cm, 40 cm, 80 cm, 120 cm, and 160 cm below the ground surfaceMs_ depth
Wind - relatedWind speed and direction/(m.s-1)Anemometer and wind vane sensorAt a height of 10 m, facing due northWSWD
Pressure and radiationAtmospheric pressure/kPaBarometerInside a waterproof casePress
Net radiation/(W.m-2)Four - component net radiometerAt a height of 1.5 m, facing due northRn
Soil thermal conditionSoil heat flux/(W.m-2)Soil heat flux plateAt a depth of 6 cm below the ground surfaceGs
Average soil temperature/(°C)Soil temperature probeAverage value of measurement values at depths of 2 cm, 4 cm, 10 cm, 20 cm, 40 cm, 80 cm, 120 cm, and 160 cm below the ground surfaceTCAV
Plant - relatedUpward/downward photosynthetically active radiation/(μmol.m-2.s-1)Photosynthetically active radiation sensorOne probe is placed 1.5 m above the canopy, one is vertical upward and the other is vertical downward, facing due southPAR_U, PAR_D

2.3 Data processing

2.3.1 Data preprocessing

To address gaps, missing values, and outliers in the monitoring data, we applied interpolation (by MATLAB 2024a) to replace anomalous values and smooth the time series. Data points were considered anomalies if they deviated from the daily mean by more than three standard deviations, or if they were physically implausible sensor readings (e.g., soil moisture >100%, air temperature >50 °C). Raw data at 10 min intervals were aggregated to daily means to reduce noise and highlight long-term trends and seasonal patterns. A sliding window filter (window width of 5 time points) was then applied to remove any spike outliers. Daily mean values were utilized for analysis to highlight long-term seasonal patterns and reduce computational complexity.

2.3.2 Multicollinearity test

Unfiltered environmental variables typically exhibit severe multicollinearity, characterized by high correlations among two or more indicators (Wang et al., 2026c). Moreover, an excessive number of variables introduces information redundancy, which reduces computational efficiency and increases model complexity (Dormann et al., 2013; Rasel et al., 2021). In this study, Variance Inflation Factors (VIF) were employed to diagnose multicollinearity. Higher VIF values indicate stronger linear dependencies among independent variables. A VIF threshold of < 5 suggests the absence of multicollinearity, while a VIF < 10 is generally considered to indicate an acceptable level of independence among predictors (Yue et al., 2021).

2.3.3 Multiple linear regression analysis

To investigate the relationship between soil moisture and meteorological factors, a standardized multiple linear regression analysis was employed. All predictor variables were normalized using Z-score standardization to eliminate the influence of dimensionality. The overall significance of the model was assessed using the F-test (p < 0.05), while the significance of individual variables was determined based on t-tests and p-values. General form of the model is:where X 1 , … , X k are non-random explanatory variables, y is the random dependent variable, β 1 , … , β k are regression coefficients, and ξ is the random error term. Standardized regression coefficients were used to compare the relative influence of different meteorological predictors.

y = β 0 + β 1 X 1 + β 2 X 2 + … + β k X k + ξ (1)

2.3.4 Redundancy analysis (RDA)

To quantify the explanatory power of environmental factors on the response variables, we used RDA to elucidate the relationships between soil moisture at eight depths (response variables) and meteorological factors (explanatory variables). RDA is a constrained ordination technique based on linear models that maximizes the covariance between the explanatory variables (meteorological factors) and soil moisture. By maximizing this covariance, RDA evaluates the explanatory strength of each environmental factor on the spatial distribution of soil moisture, revealing the driving roles of meteorological factors in soil moisture distribution.

2.3.5 Hierarchical cluster analysis (HCA)

We introduced HCA to explore the similarity and differences in environmental conditions across different years. HCA computes a distance matrix between samples based on environmental factors and then uses Ward’s minimum-variance method to merge clusters step by step, minimizing within-group variance and maximizing between-group variance. A dendrogram is constructed to group years with similar soil moisture and meteorological characteristics. Years with similar characteristics in soil moisture and meteorological characteristics are clustered together, thereby identifying typical environmental patterns for the growing and non-growing seasons and revealing intrinsic differentiation rules of environmental conditions across seasons.

3 Results

3.1 Soil moisture dynamics across depths

The time-series plot of soil moisture at 8 depths from 2 to 160 cm spanning 2012 to 2024 (Figure 2) shows that the soil moisture at the Daman farmland station in the middle reaches of the Heihe River exhibits significant seasonal fluctuations and vertical stratification. Soil moisture at all depths during the growing season generally remains at a high level throughout the year, with shallow layers such as 2 cm, 4 cm, and 10 cm peaking from June to August, reaching maximums of over 40%. Conversely, soil moisture overall declines during the non-growing season, showing particularly low water content during the winter months of December to February, with a minimum of only 4.36%. Vertically, the fluctuation amplitude of shallow soil moisture is much larger than that of the deep layer; the monthly variation at a 2 cm depth can exceed 20%, whereas soil moisture at 160 cm is generally higher with a smaller fluctuation range, mostly remaining between 25.12% and 35.46%, with an average of 29.39%. In addition, significant interannual differences in moisture dynamics exist; moisture during the growing seasons of 2018–2019 and 2022–2023 was generally high, while it was relatively low during 2016–2017 and 2020–2021. The vertical distribution characteristics and spatiotemporal variation patterns of soil moisture during the growing season also show obvious differences. Therefore, this study further analyzed the different growth stages of maize, as shown in Figure 3.

Soil moisture distribution maps of the Daman farmland station during growing and non-growing seasons.
Figure 2. Soil moisture distribution maps of the Daman farmland station during growing and non-growing seasons.
Vertical distribution of soil moisture at daman farmland station during the growing season. **(a)** 2 cm; **(b)** 4 cm; **(c)** 10 cm; **(d)** 20 cm; **(e)** 40 cm; **(f)** 80 cm; **(g)** 120 cm; **(h)** 160 cm.
Figure 3. Vertical distribution of soil moisture at daman farmland station during the growing season. (a) 2 cm; (b) 4 cm; (c) 10 cm; (d) 20 cm; (e) 40 cm; (f) 80 cm; (g) 120 cm; (h) 160 cm.

As shown in Figure 3, regarding vertical distribution, the moisture variability at 2–20 cm is significantly higher than that in the deep layers. Specifically, Figure 3a shows that soil moisture at 2 cm depth has the widest distribution range across all growing seasons, falling between 16.53% and 30.61%. Moisture distribution at the 4 cm and 10 cm (Figures 3b,c) depths is relatively concentrated, mainly ranging between 15.62% and 27.63% (Figure 3b) and 23.47%–36.68% (Figure 3c), respectively. In terms of growing season dynamics, soil moisture at all layers during the SOS is generally low, ranging from 16.52% to 18.82% at the 2 cm depth (Figure 3a). However, shallow soil moisture increases significantly from SES to TAS, and the moisture distribution at the 160 cm (Figure 3h) depth during the JOS stage ranges from 31.41% to 33.29%, marking an increase of nearly 5% compared to the sowing stage. Soil moisture across all depth layers exhibits a trend of initially increasing and then decreasing throughout the six growth stages of maize; for example, the 80 cm soil layer rises from 35.25% to 42.93%, and finally drops to 35.13% (Figure 3g). Nevertheless, the amplitude of soil moisture variation in the 120–160 cm soil layer is much smaller than that in the 2–20 cm and 40–80 cm ranges.

3.2 Relationship between soil moisture and meteorological factors

The RDA ordination plot for the growing season (Figure 4a) indicates that the first two ordination axes cumulatively explain 93.17% of the environmental variation, with RDA1 explaining 79.11% and RDA2 explaining 14.06%. Ta and Rn are the core factors correlated with soil moisture changes during the growing season. Ta and Rn exhibit a parallel variation trend with Ms, showing a positive correlation with changes in soil moisture. The angles between IRT, PAR, and TCAV and deep soil moisture variables like Ms_160 and Ms_120 are close to right or even obtuse angles, indicating a negative correlation trend regarding their influence on deep soil moisture, and TCAV has a minor impact on soil moisture.

RDA analysis result plot of the relationships between soil moisture and meteorological factors. **(a)** Growing season; **(b)** Non-growing season.
Figure 4. RDA analysis result plot of the relationships between soil moisture and meteorological factors. (a) Growing season; (b) Non-growing season.

As for non-growing season, Figure 4b shows that the first two ordination axes cumulatively explain 93.03% of the environmental variation, with RDA1 explaining 83.48% and RDA2 explaining 9.55%. Ta, RH, and PAR_U, as core factors, have a significant impact on soil moisture variations and exhibit a positive correlation. Compared to the growing season, the driving effect of Rn on soil moisture is smaller during the non-growing season, while PAR and Gs show a significant negative correlation with deep soil moisture. On the RDA2 axis, PAR_D exhibits a parallel variation trend with deep soil moisture Ms_160, 120, and 80, which is opposite to the variation in shallow soil moisture.

3.3 Regression analysis results

Prior to regression, we performed multicollinearity analysis using the Variance Inflation Factor (VIF) to detect multicollinearity among predictors (Figure 5). In this study, the VIF values for Ta and PAR during the growing season were larger than 10, indicating severe collinearity, so they were removed. To eliminate the influence of varying dimensions among predictors with different units (e.g., °C vs. W/m2), all remaining variables were standardized using Z-score normalization prior to model fitting. The VIF values for the standardized indicators—IRT and PAR_U during the growing season, and Ta and PAR during the non-growing season—ranged between 5 and 10, indicating mild collinearity. After addressing collinearity, we fitted multiple linear regression models. Model significance was evaluated using the F-test, and the significance of individual predictors was determined using p-values. The resulting regression equations, featuring standardized regression coefficients (β) for the growing and non-growing seasons, are given in Table 2.

Multicollinearity test result plot.
Figure 5. Multicollinearity test result plot.

TABLE 2. Soil moisture regression equations for growing and non-growing seasons.

DepthLinear regression equationR2p
2 cmy *g2* = −0.028x *1* - 0.064x *2* + 0.164x *3* + 0.048x *4* + 0.063x *5* - 0.697x *6* + 0.644x *7* + 0.003x *8* + 0.01x *9* - 0.198x *10* + 0.253x *11* + 0.141x *12*0.533p < 0.05
Y *ng2* = 0.195x *1* - 0.07x *2* + 0.007x *3* + 0.004x *4* - 0.081x *5* - 0.294x *6* + 0.561x *7* - 0.02x *8* - 0.054x *9* + 0.306x *10*0.671p < 0.05
4 cmy *g4* = −0.051x *1* - 0.058x *2* + 0.085x *3* + 0.048x *4* - 0.116x *5* - 0.431x *6* + 0.187x *7* + 0.037x *8* + 0.058x *9* - 0.035x *10* + 0.082x *11* + 0.328x *12*0.620p < 0.05
y *ng4* = 0.024x *1* - 0.042x *2* + 0.076x *3* + 0.028x *4* - 0.037x *5* + 0.1x *6* + 0.207x *7* + 0.088x *8* + 0.165x *9* - 0.172x *10* + 0.165x *11* + 0.474x *12*0.747p < 0.05
10 cmy *g10* = −0.016x *1* - 0.034x *2* + 0.083x *3* + 0.046x *4* + 0.074x *5* + 0.011x *6* + 0.288x *7* + 0.035x *8* - 0.238x *9* - 0.116x *10* + 0.023x *11* + 0.281x *12*0.559p < 0.05
y *ng10* = 0.071x *1* - 0.053x *2* + 0.074x *3* - 0.009x *4* + 0.008x *5* + 0.007x *6* + 0.062x *7* + 0.216x *8* - 0.003x *9* + 0.147x *10* - 0.096x *11* + 0.576x *12*0.746p < 0.05
20 cmy *g20* = −0.013x *1* - 0.029x *2* + 0.034x *3* + 0.042x *4* + 0.009x *5* - 0.053x *6* + 0.091x *7* + 0.057x *8* - 0.171x *9* + 0.004x *10* + 0.034x *11* + 0.302x *12*0.684p < 0.05
y *ng20* = 0.045x *1* - 0.024x *2* + 0.1x *3* + 0.012x *4* - 0.039x *5* + 0.032x *6* + 0.047x *7* + 0.124x *8* + 0.071x *9* - 0.02x *10* + 0.153x *11* + 0.64x *12*0.820p < 0.05
40 cmy *g40* = −0.007x *1* - 0.045x *2* - 0.088x *3* - 0.112x *4* - 0.145x *5* - 0.196x *6* + 0.044x *7* + 0.102x *8* + 0.156x *9* + 0.189x *10* + 0.245x *11* + 0.322x *12*0.696p < 0.05
y *ng40* = 0.028x *1* - 0.048x *2* + 0.137x *3* + 0.013x *4* - 0.032x *5* + 0.2x *6* + 0.23x *7* - 0.086x *8* + 0.123x *9* - 0.097x *10* + 0.107x *11* + 0.549x *12*0.834p < 0.05
80 cmy *g80* = −0.024x *1* - 0.025x *2* + 0.014x *3* + 0.015x *4* - 0.03x *5* - 0.191x *6* + 0.418x *7* + 0.097x *8* - 0.249x *9* - 0.261x *10* + 0.074x *11* + 0.356x *12*0.574p < 0.05
y *ng80* = −0.006x *1* - 0.067x *2* + 0.197x *3* - 0.043x *4* - 0.03x *5* + 0.069x *6* + 0.154x *7* - 0.084x *8* - 0.017x *9* - 0.032x *10* + 0.671x *11* + 0.671x *12*0.755p < 0.05
120 cmy *g120* = −0.055x *1* + 0.002x *2* + 0.02x *3* + 0.001x *4* - 0.046x *5* - 0.154x *6* + 0.274x *7* + 0.121x *8* - 0.291x *9* - 0.155x *10* + 0.025x *11* + 0.279x *12*0.634p < 0.05
y *ng120* = −0.016x *1* - 0.053x *2* + 0.223x *3* - 0.042x *4* - 0.048x *5* + 0.069x *6* + 0.059x *7* - 0.115x *8* - 0.143x *9* + 0.076x *10* + 0.022x *11* + 0.513x *12*0.761p < 0.05
160 cmy *g160* = −0.047x *1* + 0.007x *2* + 0.003x *3* - 0.004x *4* + 0.032x *5* - 0.051x *6* + 0.314x *7* + 0.101x *8* - 0.39x *9* - 0.203x *10* + 0.012x *11* + 0.212x *12*0.569p < 0.05
y *ng160* = −0.001x *1* - 0.063x *2* + 0.159x *3* - 0.045x *4* - 0.05x *5* - 0.336x *6* + 0.07x *7* - 0.106x *8* - 0.146x *9* + 0.107x *10* + 0.004x *11* + 0.497x *12*0.743p < 0.05

1) “g” denotes the growing season, and “ng” denotes the non-growing season; environmental indicators and their corresponding symbols are as follows: x *1* (WS), x *2* (WD), x *3* (RH), x *4* (Rain), x *5* (Press), x *6* (IRT), x *7* (Rn), x *8* (Gs), x *9* (TCAV), x *10* (PAR_U), x *11* (PAR_D) and x *12* (Ts); 2) Values in the equations represent standardized regression coefficients (β).

Regarding the regression performance across soil profiles in the farmland ecosystem, the explanatory power for soil moisture during the growing season was lower than that during the non-growing season at all depths (Table 2). In the growing season, although the adjusted values ranged from 0.534 to 0.696, which might seem modest than non-growing season, these results are consistent with the complexity of soil hydrological processes.

Among the driving factors of soil moisture during the growing season, based on the magnitude of standardized regression coefficients, Rn, RH, and Rain are the core factors. WS and IRT exert negative impacts on soil moisture from 2 to 160 cm during the growing season, but the impact of IRT is significantly greater than that of WS, with its coefficients ranging from -0.011 to -0.697. Rn, Gs, PAR_D, and Ts all show positive impacts on soil moisture during the growing season, with Rn and Ts having a more substantial influence. Rn has a greater impact on surface and deep soils than on the middle layers, acting as the primary influence factor for soil moisture at depths of 2 cm, 10 cm, 80 cm, and 160 cm; for instance, its influence coefficient on soil moisture at the 2 cm depth reaches 0.644. In contrast, the influence of Ts on soil moisture across all layers is relatively balanced, with an influence coefficient of about 0.2. As a direct recharge source, Rain has a significant impact on the 2–20 cm soil layers and mostly shows a positive effect in the growing season equations. The response of the middle soil layer to meteorological conditions such as precipitation decreases notably, whereas deep soil moisture tends to be more sensitive to factors like TCAV and Gs.

Gs and Ts are the key influencing factors for soil moisture during the non-growing season. According to the standardized regression analysis, Ts is one of the most significant factors during the non-growing season, exerting a positive impact on soil moisture at all depths. The contributions of Rn, PAR_U, and PAR_D show a weakening trend from shallow to deep soils but still maintain a regulatory role. The impact of Rn in the non-growing season is generally smaller than in the growing season and even exhibits a negative effect at certain depths. Compared to the growing season, the response characteristics of soil moisture at different depths show clear differentiation during the non-growing season. The shallow soil at 2 cm is significantly and directly affected by Press, WS, and Rn, and its moisture dynamics are closely tied to atmospheric boundary layer processes. The middle soil at 40 cm relies more on the delayed effects of Ts and precipitation recharge, displaying relatively gentle moisture variations. The deep soil at 160 cm is primarily under the combined regulation of Ts and Rn.

3.4 Clustering analysis of seasonal differences

Cluster analysis divided the growing-season years into three clusters (Figure 6a). Cluster 1 is “high soil moisture” year type, characterized by high temperature, high humidity, abundant precipitation, and high soil moisture content; it includes 2015, 2016, 2017, and 2021, representing years with relatively wet and warm growing-season climates. Cluster 2 is “vertical moisture separation” year type, characterized by high net radiation, low precipitation, moderate temperature, relatively high humidity, high deep soil moisture but low surface soil moisture; it includes 2019 and 2020, representing years with plentiful sunshine, scarce precipitation, and strong deep soil water retention during the growing season. Cluster 3 is a type that representing relatively moderate environmental conditions, characterized by moderate temperature, humidity, precipitation, radiation, and soil moisture content (with moderate surface temperature); it includes 2012, 2013, 2014, 2018, 2022, 2023, and 2024, representing relatively dry growing-season conditions with stable environmental indices.

Dendrogram of hierarchical clustering. **(a)** Growing season; **(b)** Non-growing season.
Figure 6. Dendrogram of hierarchical clustering. (a) Growing season; (b) Non-growing season.

For non-growing season (Figure 6b), three clusters also emerged. Cluster 1 is “moderate soil moisture” year type, characterized by moderate temperature, moderate radiation, and relatively high soil moisture content; it includes 2012, 2013, 2014, 2015, 2022, 2023, and 2024, representing years with relatively mild conditions and moderate precipitation. Cluster 2 is “low soil moisture” year type, characterized by high net radiation, relatively low temperature, moderate precipitation, low deep soil moisture, and low surface temperature; it includes 2019 and 2020, representing years with abundant sunshine but low temperature and relatively dry soils. Cluster 3 is “dry soil moisture” year type, characterized by low temperature, little precipitation, low soil moisture content, and low surface temperature; it includes 2016, 2017, 2018, and 2021, representing the driest non-growing-season conditions.

4 Discussion

4.1 Seasonal differences in soil moisture distribution

RDA results show consistent characteristics with the regression results, and a comparative analysis of the RDA results between the growing and non-growing seasons reveals the seasonal differentiation features of the driving factors. The pattern of high soil moisture in the growing season and low moisture in the non-growing season reflects the active water dynamics driven by the combined effects of crop transpiration and precipitation recharge, indicating that low temperatures inhibit evaporation while also restricting water recharge. Among these factors, Ta is the primary driver of soil moisture variation, but its impact is greater during the growing season, previous study have also corroborated this viewpoint (Mao et al., 2023). The relatively low values observed in growing season do not undermine the validity of regression analysis. On the contrary, they indicate that a portion of the soil moisture variation is governed by the selected meteorological drivers, while the remainder reflects the inherent stochasticity of soil moisture dynamics in arid oasis ecosystems. Unlike controlled laboratory experiments, field observations are constrained by unobserved factors such as spatial heterogeneity of soil texture, antecedent moisture conditions, and subsurface flow, which inevitably introduce unexplained variance.

Soil moisture variations at different depths during the growing season are complex (Fang et al., 2018; Nan et al., 2024); continuous irrigation prevents extreme drought in the overall soil profile, but the interactions between factors like precipitation and temperature demonstrate diversity. Consistent with classical soil physics principles, our results confirmed that the magnitude of soil moisture variation attenuates with increasing depth (Figure 2), primarily due to the damping effect of soil matrix and gravitational forces. However, beyond this expected physical attenuation, our study reveals a critical seasonal decoupling in the driving mechanisms. While the ‘decrease with depth’ is a essential results, we found that the primary factors of this vertical profile shift dramatically between seasons. This suggests that the vertical transmission of meteorological signals is not merely a passive physical process but is actively modulated by seasonal energy budgets.

Consistent with recent findings on the Loess Plateau (Wang et al., 2026a), this vertical attenuation is not merely a static property but a dynamic filtering process. Studies suggest that this discrepancy primarily arises because surface soil moisture is more susceptible to short-term perturbations—such as human activities or extreme weather events—which can mask circulation effects. Consequently, precipitation contributes significantly to the variability of surface soil moisture. In contrast, deep soil moisture is less affected by these transient disturbances, thereby more clearly reflecting the cumulative and long-term impacts of circulation indices. This explains why deep soil layers function as a stable buffer zone against short-term climatic fluctuations. This conclusion is consistent with previous findings in the Heihe River Basin (Wenye et al., 2026). In contrast, soil moisture variability during the non-growing season is small, leading to a relatively narrow range of changes in the coefficients of various indicators in the regression models.

4.2 Seasonal driving factors and characteristics

The growing season is a critical period for active plant growth and the coupling of water consumption and recharge processes. Influenced by multiple interacting factors, environmental variation is more complex during this period, leading to a significant differentiation in the dominant influencing factors and their directions of effect on soil moisture at different depths (Nan et al., 2024; Zhang et al., 2024). Soil layers closer to the surface experience frequent heat and energy exchanges with the atmosphere, directly impacting soil moisture distribution. The intense fluctuations in soil moisture at the 2 cm layer during the SES and JOS stages indicate that this layer is most sensitive to meteorological conditions and agricultural activities like irrigation, while deep layers possess stronger buffering capacities against short-term meteorological fluctuations. Notably, the moisture differences across all depths narrow significantly during the stage of MAS, and moisture distribution tends to be uniform, suggesting that soil moisture reaches a new equilibrium state through redistribution after crop harvest.

The growing season centers around energy inputs like Rn and water recharge from precipitation and irrigation. Soil moisture is primarily driven by the complex “radiation-temperature-precipitation” system, this diverges from the conclusions drawn by existing study (Wang et al., 2026a). Rn indirectly correlate with soil moisture by influencing precipitation and evapotranspiration processes, jointly determining the “income-expenditure” balance of soil moisture (Kyle et al., 2024; Koster et al., 2024). As the source of surface energy, the impact of Rn on soil moisture is dual: on the one hand, it accelerates soil evaporation and plant transpiration by raising near-surface temperatures, leading to water loss (López-Olivari et al., 2016); on the other hand, higher Rn may indirectly increase water recharge by promoting precipitation events (Jones and Brunsell Nathaniel, 2009; Koster et al., 2006). The consequent increase in precipitation directly leads to an increase in soil moisture. Consequently, Rn shows a positive correlation with soil moisture.

It is important to note, however, that these relationships are derived from linear statistical models. As emphasized by Wang et al. (2026b), soil moisture responses to climatic drivers often involve threshold effects and nonlinear hysteresis loops. While Rn shows a net positive correlation here, its impact is likely mediated by complex interactions between soil thermal regimes and hydraulic conductivity. During the growing season, the system is dominated by the coupling of vegetation water demand and energy input, whereas in the non-growing season, the system shifts to purely physical processes governed by soil thermodynamics. This fundamental shift in energy budget allocation explains the seasonal decoupling of driving factors observed in our RDA results (Figure 4).

During the non-growing season, plant growth stagnates, and changes in soil moisture are mainly affected by “atmosphere-soil physical processes (Table 2),ˮ including soil physical states like temperature, meteorological conditions like Press and WS, and energy balance indicators like radiation (Sun et al., 2025). The driving mechanism fundamentally differs from that of the growing season. Specifically, temperature and soil heat flux directly determine the retention-loss process of soil moisture. A slight increase in Ts in low-temperature winter environments can alter the energy exchange between the soil and the external environment, promoting water thawing or slowing the freezing rate, thereby changing soil moisture mobility; thus, Ts is one of the primary factors during the non-growing season, and research indicates that optimal Ts facilitates seed germination (Mostafa et al., 2025). Overall, soil moisture is driven by the dual supply and demand of energy and water in the growing season, whereas it is primarily correlate with low temperatures and limited recharge in the non-growing season, with deep soils playing a crucial role in maintaining moisture stability. The effect of Rn in the non-growing season is weaker than in the growing season, and its direct “heating” effect on soil moisture diminishes, but may still indirectly affect water evaporation by influencing soil surface temperatures. Regarding the proportion of variance explained, the explanation rate 83.48% of RDA1 in the non-growing season is significantly higher than that in the growing season — 79.11%, indicating that soil moisture variation in the non-growing season is more concentrated on a few key factors, reflecting stronger system stability.

4.3 Climate and soil moisture patterns from 2012 to 2024

Climate anomalies and agricultural management exert a significant influence on the interannual variability of soil moisture. In this study, Cluster 3 of the growing season (including 7 years: 2012, 2013, 2014, 2018, 2022, 2023, and 2024) and Cluster 1 of the non-growing season (including 7 years: 2012, 2013, 2014, 2015, 2022, 2023, and 2024) are the largest clusters in their respective seasons, with a substantial overlap in the included years. Compared to the years in other clusters, the years within these two clusters exhibit characteristics of moderate temperatures and moderate precipitation, with all indicators falling in the middle range, representing years that are relatively normal and free of extreme climate in the middle Heihe River farmlands. This indicates that in most years during both the growing and non-growing seasons, environmental indicators remain relatively stable and conventional, with steady climatic performance. The combination of Cluster 3 in the growing season which is the baseline soil moisture year and Cluster 1 in the non-growing season which is the medium soil moisture year reveal the seasonal differentiation characteristic of the Daman farmland station—“coordinated temperature and humidity in the growing season, and low-temperature water storage in the non-growing seasonˮ—as well as the climate foundation for the dynamic balance of “precipitation-evaporation-soil storage” on an annual scale (Chunlin et al., 2016; Palmer, 2022). During the growing season, Ta, Rain, RH, Rn, and evapotranspiration are mutually coordinated, ensuring a stable water supply during the critical period of crop growth (Liu et al., 2025). During the non-growing season, consumption is suppressed by low Ta, and a small amount of precipitation recharges and maintains soil moisture, forming a “low input and low consumption” water storage pattern that lays the foundation for the next growing season. This is consistent with previous research findings (Yinglan et al., 2019; Lakshmi et al., 2003).

The 3 years of 2016, 2017, and 2021 belong to Cluster 1 during the growing season, exhibiting high temperatures, high humidity, and high soil moisture; however, they fall into Cluster 3 during the non-growing season, characterized by low Ta, low RH, and low soil moisture. This highlights the stark contrast in moisture conditions between different seasons within the same year. The primary reason is that soil moisture is driven by the seasonal variations of precipitation and evapotranspiration, leading to significant differences in precipitation and temperature between the growing and non-growing seasons in the same year (Shen et al., 2024; Huazhu et al., 2025; Ding et al., 2024). During the growing season, precipitation is relatively abundant and supplemented by irrigation, leading to strong evaporation and transpiration, and the continuous water input helps maintain high soil moisture. However, during the non-growing season, due to low temperatures, scarce precipitation, frozen soil, and dormant vegetation with no transpiration, soil moisture is gradually depleted or maintained at extremely low levels due to a lack of input (Eunsang and Minha, 2014; Baldocchi et al., 2022; Liu and Yang, 2022). The active recharge from precipitation and irrigation in the growing season versus the continuous drying caused by low temperatures and scarce snow in the non-growing season further demonstrates the decisive influence of seasonal water input on the interannual fluctuations of soil moisture.

4.4 Causes of soil moisture discrepancies between different ecosystems

Previous studies conducted by our team on artificial grasslands have provided a baseline (Wenye et al., 2026). However, systematic differences in soil moisture characteristics exist between maize cropland and artificial grassland ecosystems. These discrepancies are primarily attributed to distinct root system architectures and hydrological behaviors. As a fibrous and deep-rooted crop, maize develops root systems extending nearly 1 m vertically (Daniel, 1962). It exerts high-intensity transpiration to consume deep soil water during the growing season (particularly from JOS to GFS), and its soil water dynamics are subject to strong anthropogenic interference (e.g., irrigation and tillage), resulting in distinct seasonal patterns of pulsed consumption and recharge (Liao et al., 2024; Zhou et al., 2021). In contrast, artificial grasslands—typically composed of Gramineae and Leguminosae species—exhibit high root density concentrated in shallow layers (0–40 cm) (Shahnaj et al., 2023; Wang et al., 2020). These systems rely on dense root networks to intercept precipitation and suppress evaporation. Furthermore, due to long-term no-tillage and organic matter accumulation, a surface soil layer with high water retention capacity forms (particularly within 0–10 cm), leading to relatively smooth water consumption and stronger hydrological connectivity in the topsoil, whereas cropland is characterized by more intense vertical water redistribution (Han et al., 2025; Sayantan et al., 2024; Tad et al., 2025). Additionally, root-soil energy interactions induce changes in soil physical properties, which subsequently regulate soil hydraulic behavior and evaporative fluxes, thereby altering the coupling strength between the land surface and the atmosphere (Sayantan et al., 2024; Farzane et al., 2023). Collectively, soil moisture serves as a critical state variable that interacts with diverse surface and eco-hydrological processes, thereby establishing complex feedback mechanisms (Wang et al., 2017; Shmuel, 2021; Goutham et al., 2024).

4.5 Future research directions

This study revealed the seasonal dynamics and main driving mechanisms of farmland soil moisture in the middle reaches of the Heihe River. Although the current analysis based on daily mean values effectively captured long-term trends, it inevitably smoothed out sub-daily fluctuations. Future studies should therefore integrate hourly-scale observations to capture transient boundary layer fluxes before daily averaging. Furthermore, while statistical models identified the covariates of soil moisture, they cannot explicitly quantify the underlying physical transport processes. Subsequent research should couple statistical analysis with Hydrus-1D simulations to explicitly track vertical water movement mechanics.

In oasis agricultural systems, differences in soil properties, irrigation practices, and land use patterns can cause significant spatial heterogeneity in soil moisture dynamics. Future studies could integrate multi-site observations and remote sensing data to more comprehensively characterize soil moisture variations at the regional scale. In addition, in irrigated agricultural systems, human irrigation activities have a significant impact on soil moisture changes. Future research could combine long-term observations with information on irrigation management and ecohydrological modeling to further assess the soil moisture dynamics and their implications for water resource management in arid oasis agricultural systems under the combined effects of climate change and human activities.

5 Conclusion

  • The vertical distribution and spatiotemporal variation patterns of soil moisture show significant differences. Shallow soil moisture of 2–20 cm tends to be more sensitive to seasonal changes, and across maize growth stages it first increases and then decreases, while deep soil moisture remains relatively stable.
  • The driving mechanisms for soil moisture suggest significant seasonal differentiation. In the growing season, soil moisture may primarily correlated with net radiation, temperature, and precipitation, presenting a “high input–high consumption” dynamic balance. In the non-growing season, it is mainly affected by soil temperature and low-radiation conditions, forming a “low input–low consumption” moisture retention mode.
  • The responses of soil moisture at different depths to environmental factors point to a vertical gradient. Shallow soil of 2–20 cm is sensitive to meteorological factors such as net radiation, air temperature, and precipitation, and its dynamics are aed by short-term weather processes. Deep soil of 120–160 cm relies more on long-term hydrological processes and soil heat flux conduction, showing a lagged response to climate variations.
  • The clustering results indicate that 2012, 2013, 2014, 2022, 2023, and 2024 are identified as typical climate and soil moisture years at the Daman farmland station. These years show stark differences in precipitation and temperature between the growing and non-growing seasons, illustrating the seasonal differentiation pattern of “coordinated warmth and moisture in the growing season, and cold-season water storage,” which represents a typical soil moisture evolution pattern in the mid-reaches farmland of the Heihe River.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Footnotes

Edited by: Raji Pushpalatha, Amrita Vishwa Vidyapeetham, India

Reviewed by: Panxing He, Fudan University, China

Sadashiv Chaturvedi, Nanjing University of Information Science and Technology, China

Statements

Conflict of interest

Authors HW and RZ were employed by Gansu Geological Engineering Survey Institute Co., Ltd.

The remaining author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Funding

The author(s) declared that financial support was received for this work and/or its publication. This study was funded by the following: China Geological Survey Project “National Water Resources Investigation, Monitoring, Assessment, and Smart Services” (DD20230075); the Xinjiang Uygur Autonomous Region Water Conservancy Special Development Fund Project (2022.B-004); Chinese Academy of Sciences Strategic Priority Research Program (Category A) “Pan-Third Pole Environment Change and Green Silk Road Constructionˮ (XDA20000000); Zhangye City Major Technology Research Project “Study on Regulation Mechanism of Underground Reservoirs in Typical Basins of the Middle Heihe River” (ZY2024JBGS08); the Ministry-Province Cooperative Project under the Ministry of Natural Resources (2024ZRBSHZ019); Xinjiang Uygur Autonomous Region Key R&D Program “Identification of Water Storage Structures and Precise Water Quality/Quantity Assessment in Tarim Basin Desert Areasˮ (2025B03025-1) and the Joint Open Fund of Water Cycle Field Station of the Heihe River Basin, CGS (No. WCSHR-2025–06); Project of Natural Science Foundation of Gansu Province: “Research on the Evolution Characteristics and Driving Forces of Shallow Groundwater Flow Field in the Middle Reaches of the Main Stream of Heihe River” (24JRRG011); Gansu Geological Engineering Investigation Institute Co., LTD. High-tech Enterprise Research and Development Fund (2025). The author(s) declared that this work received funding from Gansu Geological Engineering Investigation Institute Co., LTD. High-tech Enterprise Research and Development Fund (2025). The funder was not involved in the study design, collection, analysis, interpretation of data, the writing of this article, or the decision to submit it for publication.

Data availability

Publicly available datasets were analyzed in this study. This data can be found here: https://data.tpdc.ac.cn/home.

Author contributions

WyL: Conceptualization, Methodology, Writing – original draft. WpL: Formal Analysis, Writing – review and editing. YZ: Formal Analysis, Writing – review and editing. HW: Formal Analysis, Writing – review and editing. XH: Data curation, Writing – review and editing. XW: Data curation, Writing – review and editing. YW: Data curation, Writing – review and editing. RZ: Supervision, Writing – review and editing. JL: Supervision, Writing – review and editing.

References

  1. Baldocchi D. D., Keeney N., Rey-Sanchez C., Fisher J. B. (2022). Atmospheric humidity deficits tell us how soil moisture deficits down-regulate ecosystem evaporation. Adv. Water Resour. 159, 104100. doi:10.1016/j.advwatres.2021.104100
  2. Botter G., Porporato A., Rodriguez-Iturbe I., Rinaldo A. (2007). Basin-scale soil moisture dynamics and the probabilistic characterization of carrier hydrologic flows: slow, leaching-prone components of the hydrologic response. Water Resour. Res. 43 (2), 1 – 14. doi:10.1029/2006wr005043
  3. Cai X., Mckinney D. C., Rosegrant M. W. (2003). Sustainability analysis for irrigation water management in the Aral Sea region. Agric. Syst. 76 (3), 1043 – 1066. doi:10.1016/s0308-521x(02)00028-8
  4. Cai W., Jiang X., Sun H., He J., Deng C., Lei Y. (2022). Temporal and spatial variation and driving factors of water consumption in the middle Heihe River basin before and after the implementation of the 97 water diversion scheme. Agric. Water Manag. 269 (1), 107727. doi:10.1016/j.agwat.2022.107727
  5. Ching-Min C., Chuen-Fa N., Lin C.-P., Lee I. H., Lo W. C. (2025). Quantification of the variability of the soil moisture field in response to variations in the rainfall field. J. Hydrology 662, 133881.
  6. Chunlin H., Weijing C., Yan L., Huanfeng S., Xin L. (2016). Assimilating multi-source data into land surface model to simultaneously improve estimations of soil moisture, soil temperature, and surface turbulent fluxes in irrigated fields. Agric. For. Meteorology 230, 142 – 156. doi:10.1016/j.agrformet.2016.03.013
  7. Daniel L. (1962). Fused leaf sheath and branched stalk in maize. Acta Biologica Acad. Sci. Hung. 13, 89 – 97.
  8. Ding Z., Wang Y., Ding J., Ren Z., Liao J. (2024). Dynamics of carbon and water vapor fluxes in three typical ecosystems of Heihe River Basin, Northwestern China. Sci. Total Environ. 929, 172611. doi:10.1016/j.scitotenv.2024.172611
  9. Dormann C. F., Jane E., Sven B., Carsten B., Gudrun C., Gabriel C. (2013). Collinearity: a review of methods to deal with it and a simulation study evaluating their performance. Ecography 36 (1), 027 – 046. doi:10.1111/j.1600-0587.2012.07348.x
  10. Eunsang C., Minha C. (2014). Regional scale spatio-temporal variability of soil moisture and its relationship with meteorological factors over the Korean peninsula. J. Hydrology 516, 317 – 329. doi:10.1016/j.jhydrol.2013.12.053
  11. Fan Y., Shi J., Wen S., Nie W. (2024). Soil water infiltration characteristics for drip irrigation of seedlings to reduce aeolian erosion of sandy soils. J. Irrigation Drainage Eng. 150 (6), 04024033. doi:10.1061/jidedh.ireng-10245
  12. Fang Q., Wang G., Baolin X., Liu T., Kiem A. (2018). How and to what extent does precipitation on multi-temporal scales and soil moisture at different depths determine carbon flux responses in a water-limited grassland ecosystem? Sci. Total Environ. 635, 1255 – 1266. doi:10.1016/j.scitotenv.2018.04.225
  13. Farzane M., Sadegh J., Arsalan G., Mehdi M. (2023). Global soil moisture trend analysis using microwave remote sensing data and an automated polynomial-based algorithm. Glob. Planet. Change 231, 104310. doi:10.1016/j.gloplacha.2023.104310
  14. Feng W., Haofeng M., Xie R., Wang K., Ming B., Hou P. (2023). Optimizing deficit irrigation and regulated deficit irrigation methods increases water productivity in maize. Agric. Water Manag. 280, 108205. doi:10.1016/j.agwat.2023.108205
  15. Fishman R. M., Siegfried T., Raj P., Modi V., Lall U. (2011). Over-extraction from shallow bedrock versus deep alluvial aquifers: reliability versus sustainability considerations for India’s groundwater irrigation. Water Resour. Res. 47, W00L05. doi:10.1029/2011wr010617
  16. George K., Soulis K. X., Petros K. (2021). Implications of hysteresis on the horizontal soil water redistribution after infiltration. Water 13 (19), 2773. doi:10.3390/w13192773
  17. Goutham T., Poulamee C., Singh J., Xu S., Kovacs P., Iqbal J. (2024). X-ray computed tomography - measured pore characteristics and hydro-physical properties of soil profile as influenced by long-term tillage and rotation systems. Catena 237, 107801. doi:10.1016/j.catena.2023.107801
  18. Guan Y., Gu X., Dai A., Zhou T., Yuan X., Mishra A. K. (2025). Anthropogenic enhancement of subsurface soil moisture droughts. Nat. Clim. Change 15 (12), 1355 – 1362. doi:10.1038/s41558-025-02458-z
  19. Hakami-Kermani A., Hossein B. (2025). Evapotranspiration and water requirement changes of main crops under climate change conditions in a semi-arid region. Appl. Water Sci. 16 (1), 18. doi:10.1007/s13201-025-02696-8
  20. Han J., Jianye L., Chen Q., Zhang X. (2025). Long-term conservation tillage breaks the plough pan and promotes the development of preferential flow. Geoderma 458, 117329. doi:10.1016/j.geoderma.2025.117329
  21. He J., Yang K., Tang W., Lu H., Qin J., Chen Y. (2020). The first high-resolution meteorological forcing dataset for land process studies over China. Sci. Data 7 (1), 25. doi:10.1038/s41597-020-0369-y
  22. Huang J., Yu H., Guan X., Wang G., Guo R. (2016). Accelerated dryland expansion under climate change. Nat. Clim. Change 6 (2), 166 – 171. doi:10.1038/nclimate2837
  23. Huazhu X., Wang Y., Dong G., Zhang C., Lian Y., Wu H. (2025). Multi-source meteorological data assessment on daily runoff simulation in the upper reaches of the Hei River, Northwest China. J. Hydrology-Regional Stud. 57, 102100. doi:10.1016/j.ejrh.2024.102100
  24. Huimin M., Chesheng Z., Shi H., Zhonghe L., Zhonghui L. (2026). Implementing a dynamic root distribution scheme responsive to soil environment into Noah-MP-Crop: improving simulations of soil water, crop growth, and energy fluxes in agro-ecosystems. J. Geophys. Research-Biogeosciences 131 (4), e2025JG009462. doi:10.1029/2025JG009462
  25. Jingyu Y., Liu H., Huang J., Gao Z., Wang G., Li D. (2020). Accelerated dryland expansion regulates future variability in dryland gross primary production. Nat. Commun. 11 (1), 1665. doi:10.1038/s41467-020-15515-2
  26. Jinxia A., Ying M., Juan P., Guangyao G. (2026). Redistribution of hydrological and dissolved nutrient by self-organized shrubs: strategies to collect rain and nutrient for surviving in drylands. J. Hydrology 666, 134877. doi:10.1016/j.jhydrol.2025.134877
  27. Jones A. R., Brunsell Nathaniel A. (2009). Energy balance partitioning and net radiation controls on soil moisture-precipitation feedbacks. Earth Interact. 13, 2 – 25. doi:10.1175/2009ei270.1
  28. Kang W., Zhang Y., Wenzhi Z., Yang Q. (2025). Soil water regimes and climate sensitivity in hyper-arid deserts of the Hexi Corridor, China: insights from a complete moisture range model. J. Hydrology 662, 133870. doi:10.1016/j.jhydrol.2025.133870
  29. Koster R. D., Fekete B. M., Huffman G. J., Stackhouse P. W. (2006). Revisiting a hydrological analysis framework with international satellite land surface climatology project initiative 2 rainfall, net radiation, and runoff fields. J. Geophys. Research-Atmospheres 111 (D22), D22S05. doi:10.1029/2006jd007182
  30. Koster R. D., Feldman A. F., Holmes T. R. H., Anderson M. C., Crow W. T., Hain C. (2024). Estimating hydrological regimes from observational soil moisture, evapotranspiration, and air temperature data. J. Hydrometeorol. 25 (3), 495 – 513. doi:10.1175/jhm-d-23-0140.1
  31. Kyle L., Di T., Wang H. (2024). Subseasonal forecast skill of evaporative demand, soil moisture, and flash drought onset from two dynamic models over the contiguous United States. J. Hydrometeorol. 25 (7), 965 – 990. doi:10.1175/JHM-D-23-0124.1
  32. Lakshmi V., Jackson T. J., Zehrfuhs D. (2003). Soil moisture–temperature relationships: results from two field experiments. Hydrol. ProcHydrological Process. 17 (15), 3041 – 3057. doi:10.1002/hyp.1275
  33. Liao Qi, Xukai L., Wang R., Du T., Zhao X., Kang S. (2024). Maize yield is associated with abscisic acid and water potential under reduced soil water supply but with indoleacetic acid in genotypic renewal. Plant Physiology Biochem. 217, 109299. doi:10.1016/j.plaphy.2024.109299
  34. Liu Y., Yang Y. (2022). Advances in the quality of global soil moisture products: a review. Remote Sens. 14 (15), 3741. doi:10.3390/rs14153741
  35. Liu S., Xin L., Ziwei X., Tao C., Qing X., Mingguo M. (2018). The Heihe integrated observatory network: a basin-scale land surface processes observatory in China. Vadose Zone J. 17 (1), 180072. doi:10.2136/vzj2018.04.0072
  36. Liu Qi, Niu J., Wood J. D., Kang S. (2022). Spatial optimization of cropping pattern in the upper-middle reaches of the Heihe River basin, Northwest China. Agric. Water Manag. 264, 107479. doi:10.1016/j.agwat.2022.107479
  37. Liu S., Ziwei X., Che T., Li X., Xu T., Ren Z. (2023). A dataset of energy, water vapor, and carbon exchange observations in oasis-desert areas from 2012 to 2021 in a typical endorheic basin. Earth Syst. Sci. Data 15 (11), 4959 – 4981. doi:10.5194/essd-15-4959-2023
  38. Liu H., Aihua L., Yang Y., Yu J. (2025). Advantages of the E-CMRSET model in unveiling evapotranspiration water consumption patterns in the inland river basin. J. Environ. Manag. 384, 125426. doi:10.1016/j.jenvman.2025.125426
  39. López-Olivari R., Ortega-Farias S., Poblete-Echeverria C. (2016). Partitioning of net radiation and evapotranspiration over a superintensive drip-irrigated olive orchard. Irrigation Sci. 34 (1), 17 – 31. doi:10.1007/s00271-015-0484-2
  40. Ma F., Yuan X. (2025). The propagation from atmospheric flash drought to soil flash drought and its changes in a warmer climate. J. Hydrology 654, 132877. doi:10.1016/j.jhydrol.2025.132877
  41. Mao R., Song J., Tang B., Xu W., Kong F., Sun H. (2023). Vegetation variation regulates soil moisture sensitivity to climate change on the Loess Plateau. J. Hydrology 617, 128763. doi:10.1016/j.jhydrol.2022.128763
  42. Martin J., Markus R., Ciais P., Seneviratne S. I., Sheffield J., Goulden M. L. (2010). Recent decline in the global land evapotranspiration trend due to limited moisture supply. Nature 467, 951 – 954. doi:10.1038/nature09396
  43. Melo De, Luciana A. M., Jong Van Lier De, van Dam J. C. (2026). FAO crop coefficient and plant-available water: do soil and plant hydraulic properties matter? Agron. J. 118 (1), E70277. doi:10.1002/agj2.70277
  44. Meng B., Wang J., Meng F., Zheng E., Li T., Tang B. (2025). Stochastic simulation of soil moisture dynamics in farmland in the eastern region of the Songnen Plain. Plos One 20 (1), e0318161. doi:10.1371/journal.pone.0318161
  45. Mostafa O., Mohammad A., Alireza P., Ramin P., Hassan A., Heinz M. (2025). Modelling the effects of warmer, drier winters on dormancy release and germination in summer annual weeds. Weed Res. 65 (6), e70047. doi:10.1111/wre.70047
  46. Nan Li, Skaggs T. H., Ellegaard P., Bernal A., Scudiero E. (2024). Relationships among soil moisture at various depths under diverse climate, land cover and soil texture. Sci. Total Environ. 947, 174583. doi:10.1016/j.scitotenv.2024.174583
  47. Ogunsola O. Q., Bankole A. O., Soboyejo L. A., Adejuwon J. O., Makinde A. A. (2025). Modeling deficit irrigation water demand of maize and potato in Eastern Germany using ERA5-Land reanalysis climate time series. Irrigation Sci. 43 (5), 1127 – 1145. doi:10.1007/s00271-024-00939-1
  48. Palmer L. (2022). Storing frozen water to adapt to climate change. Nat. Clim. Change 12 (2), 115 – 117. doi:10.1038/s41558-021-01260-x
  49. Rasel S. M. M., Chung C. H., Ralph T. J., Saintilan N., Diti I. J. (2021). Application of feature selection methods and machine learning algorithms for saltmarsh biomass estimation using Worldview-2 imagery. Geocarto Int. 36 (10), 1075 – 1099. doi:10.1080/10106049.2019.1624988
  50. Ridolfi L., D'odorico P., Porporato A., Rodriguez-Iturbe I. (2003). Stochastic soil moisture dynamics along a hillslope. J. Hydrology 272 (1-4), 264 – 275. doi:10.1016/s0022-1694(02)00270-6
  51. Sayantan S., Dianna B., Srinivasulu A., Cristine L. S. M., Christine C. M. (2024). Modeling tillage effects on plant-available water by considering changes in soil structure. J. Asabe 67 (3), 589 – 599. doi:10.13031/ja.15695
  52. Shahnaj P., Jason C., Rose T. (2023). Rooting depth and water use of summer cover crops in a semi-arid cropping environment. Eur. J. Agron. 147, 126847. doi:10.1016/j.eja.2023.126847
  53. Shen C., Wang G., Gou Q. (2024). Dynamics of soil water, temperature, and salt and their coupled effects in Haloxylon ammodendron forests of different ages in an arid desert oasis ecotone. J. Hydrology Regional Stud. 55, 101965. doi:10.1016/j.ejrh.2024.101965
  54. Shiao F., Wang W., Zhang Y., Wei Z., Dong J., Weihermüller L. (2026). Fusing ERA5-Land and SMAP L4 for an improved global soil moisture product (1950–2025). Earth Syst. Sci. Data 18 (2), 1061 – 1088. doi:10.5194/essd-18-1061-2026
  55. Shmuel A. (2021). What can we learn from the water retention characteristic of a soil regarding its hydrological and agricultural functions? Review and analysis of actual knowledge. Water Resour. Res. 57 (12), e2021WR031026. doi:10.1029/2021WR031026
  56. Sun W., Sha Z., Bofu Y., Zhang Y., Keenan T. (2025). Soil moisture-atmosphere interactions drive terrestrial carbon-water trade-offs. Commun. Earth and Environ. 6 (1), 169. doi:10.1038/s43247-025-02145-z
  57. Tad T., Daren H. R., Erik W., Cassidy B., Ansley B., Emmanuel D. (2025). Connecting the soil health-water quality nexus under surface-irrigated conservation tillage. J. Environ. Qual. 54 (5), 1090 – 1102. doi:10.1002/jeq2.70013
  58. Triplett A., Condon L. E. (2023). Climate-warming-driven changes in the cryosphere and their impact on groundwater-surface-water interactions in the Heihe River basin. Hydrology Earth Syst. Sci. 27 (14), 2763 – 2785. doi:10.5194/hess-27-2763-2023
  59. Wang T., Franz T. E., Jinsheng Y., Shulski M. D., Ray C. (2017). Evaluating controls of soil properties and climatic conditions on the use of an exponential filter for converting near surface to root zone soil moisture contents. J. Hydrology 548, 683 – 696. doi:10.1016/j.jhydrol.2017.03.055
  60. Wang G., Liu S., Yan F., Shangguan Z. (2020). Adaptive changes in root morphological traits of Gramineae and Leguminosae seedlings in the ecological restoration of the semiarid region of northwest China. Land Degrad. and Dev. 31 (16), 2417 – 2429. doi:10.1002/ldr.3616
  61. Wang L., Yinglan A., Wang G., Xue B. (2024). Improvement of evapotranspiration simulation study in the Hailar River basin under the influence of vegetation dynamics. J. Contam. Hydrology 262, 104324. doi:10.1016/j.jconhyd.2024.104324
  62. Wang J., Zhang S., Yuning C., Shi B., Li X., Li B. (2026a). Divergent hydrological responses of multi-depth soil moisture variability to multiple factors on the Loess Plateau, China. J. Hydrology 670, 135226. doi:10.1016/j.jhydrol.2026.135226
  63. Wang M., Ding Y., Jia Z., Tao K. (2026b). Impact of climate change adaptive strategies on maize yield and irrigation in drought areas. Irrigation Drainage 75 (2), 799 – 814. doi:10.1002/ird.70057
  64. Wang J., Yuning C., Zhang S., Shi B., Zhang L., Du C. (2026c). Quantitative attribution of seasonal runoff variations in the Yellow River Basin under unsteady conditions using a modified Budyko framework. J. Hydrology 675, 135588. doi:10.1016/j.jhydrol.2026.135588
  65. Wenye L., Wenpeng L., Zheng Y., Wang X., Qi X. (2026). Dynamics of soil moisture and its response to meteorological factors at different depths in an arid land, Northwest China. Atmosphere 17 (3), 232. doi:10.3390/atmos17030232
  66. Xu H., Jia S., Zhu W. (2026). Characteristics and influencing factors of soil moisture memory across mainland China. J. Hydrology 671, 135281. doi:10.1016/j.jhydrol.2026.135281
  67. Xu Z. -G., Wu Z.-Y., He H., Guo X., Zhang Y.-L. (2025). Impact of soil moisture depth variation on agricultural drought monitoring in karst and non-karst areas of southwest China: a case study of Yunnan Province. Catena 259, 109392. doi:10.1016/j.catena.2025.109392
  68. Yao R., Dai X., Wang W., Wu P., Huo Z. (2023). Dependence of evapotranspiration validity on shallow groundwater in arid area-a three years field observation experiment. Agric. Water Manag. 286, 108411. doi:10.1016/j.agwat.2023.108411
  69. Yinglan A., Wang G., Liu T., Xue B., Kuczera G. (2019). Spatial variation of correlations between vertical soil water and evapotranspiration and their controlling factors in a semi-arid region. J. Hydrology 574, 53 – 63. doi:10.1016/j.jhydrol.2019.04.023
  70. Yue W., Liu Z., Su M., Gu Z., Xu C. (2021). The impacts of multi-dimension urbanization on energy-environmental efficiency: empirical evidence from Guangdong Province, China. J. Clean. Prod. 296, 126513. doi:10.1016/j.jclepro.2021.126513
  71. Zhan C. S., Jiang S., Sun F. B., Jia Y. W., Niu C. W., Yue W. F. (2014). Quantitative contribution of climate change and human activities to runoff changes in the Wei River basin, China. Hydrology Earth Syst. Sci. 18 (8), 3069 – 3077. doi:10.5194/hess-18-3069-2014
  72. Zhang Z., Zhang Y. L., Henderson M., Wang G., Chen M., Fu Y. (2024). Effect of land use type on soil moisture dynamics in the sloping lands of the black soil (mollisols) region of Northeast China. Agriculture-Basel 14 (8), 1261. doi:10.3390/agriculture14081261
  73. Zhang Z., Guan Q., Bo Z., Xiao X., Zhang J., Yang X. (2025). Multi-objective optimal allocation of agricultural water and land resources in the Heihe River basin: coupling of climate and land use change. J. Hydrology 653, 132783. doi:10.1016/j.jhydrol.2025.132783
  74. Zhengguang X., Wu Z., Hai H. (2021). Comparison of soil moisture at different depths for drought monitoring based on improved soil moisture anomaly percentage index. Water Sci. Eng. 14 (3), 171 – 183.
  75. Zhou H., Zhou G., He Q., Zhou L., Ji Y., Lv X. (2021). Capability of leaf water content and its threshold values in reflection of soil-plant water status in maize during prolonged drought. Ecol. Indic. 124, 107395. doi:10.1016/j.ecolind.2021.107395

Citation

Li, W., et al. (2026). Seasonal differentiation and drivers of vertical soil moisture variation in farmland of the middle Heihe River. Frontiers in Earth Science, Volume 14. https://doi.org/
Received
Accepted
Published
Volume
14 - 2026

Copyright

Copyright © 2026 Li, Li, Zheng, Wang, He, Wang, Wen, Zhao and Lu.

License: https://creativecommons.org/licenses/by/4.0/

*Correspondence: Yuejun Zheng

Article metrics

Views
0
Downloads
0
Citations
0
Shares
0

Related articles

Original ResearchMulti-constraint collaborative seismic imaging and interpretation method for fine characterization of small faults in complex coal mining areasNing Ma, Dong Yan, Liyong Qian, Pengcheng Niu, Xuefeng Tian, Jiandong Lin, Yong WangOriginal ResearchQuantified impact of projected climate change on groundwater recharge and river discharge leveraging the use of open access geospatial dataJulius Incillo Jimenez, Nitin Kumar Tripathi, Ashish Pandey, Sangam Shrestha, Kuo Chieh ChaoOriginal ResearchEffects of pressure on the stable carbon isotope composition of gas generated from continental type-I and type-III kerogen under semi-closed conditionsYanju Li, Hao Yang, Zhonghong Ning, Shengling Jiang, Biyang Feng, Xiulan Zhu, Wanchang Zhang, Xianghe Ma