Separating Climatic and Anthropogenic Drivers of Groundwater Change in an Arid Inland Basin: Insights from the Shule River Basin, Northwest China

Highlights

What are the main findings?

• Groundwater in the Shule River Basin declined persistently from 2003 to 2023 at −0.31 cm yr⁻¹, with the most severe losses in the central and lower reaches.
• Natural variability explained most of the early depletion, but human activities became the dominant driver after 2016, closely linked to cropland expansion, urban growth and GDP.

What is the implication of the main finding?

• The intensifying role of human activities highlights the urgent need for adaptive water management in arid inland basins.
• The integrative framework combining GRACE, land surface models, and socio-economic data offers transferable insights for groundwater sustainability in other water-stressed regions.

Abstract

Groundwater is a vital resource in arid regions, where it sustains agriculture, industry, and livelihoods. In northwestern China’s Shule River Basin, located in the Hexi Corridor, increasing water stress has raised concerns about the sustainability of groundwater use. However, the relative contributions of climate variability and human activities to groundwater depletion in this region remain poorly quantified. This study investigates long-term groundwater storage changes in the Shule River Basin from 2003 to 2023 using GRACE satellite data combined with GLDAS land surface models. A water balance approach was applied to isolate natural (climatic) and anthropogenic contributions to groundwater storage anomalies (GWSAs). In addition, land use transitions and socioeconomic indicators were incorporated to assess the impact of human development on subsurface water dynamics. The results show a persistent downward trend in GWSA, with an average annual loss rate of −0.31 cm·yr⁻¹. Spatially, the central and lower reaches of the basin exhibit the most significant depletion, driven by intensive irrigation and urban growth. Contribution analysis indicates that natural factors accounted for 61% of the groundwater loss across the study period, while anthropogenic drivers became increasingly dominant over time, particularly after 2016, accounting for over 40% of total depletion in recent years. Strong correlations were found between groundwater decline and the expansion of cropland, impervious surfaces, and GDP. These findings highlight the intensifying role of human activities in shaping groundwater trends in arid inland basins. This study provides a data-driven framework to support sustainable groundwater management and offers transferable insights for similar water-stressed regions globally.

1. Introduction

Water groundwater is the largest accessible source of freshwater on Earth and plays a crucial role in maintaining ecological integrity, supporting agricultural productivity, and ensuring water security [1,2,3]. Globally, it supplies approximately 50% of domestic water, 40% of industrial water, and 20% of irrigation water [4,5]. Beyond its direct utilization, groundwater is a key component of the hydrological cycle, regulating runoff, supporting baseflow, and maintaining wetland and vegetation health [6,7,8].

However, the availability and quality of groundwater are increasingly threatened by both natural climate variability and anthropogenic pressures, particularly in arid and semi-arid regions [9,10,11]. These threats have led to groundwater depletion worldwide, causing ecological degradation, land subsidence, and reduced baseflow and even contributing to sea-level rise [12]. Understanding the spatial and temporal dynamics of groundwater change and their driving mechanisms is, therefore, a critical research priority for global water security and sustainable development [13].

Rapid socioeconomic development has significantly increased groundwater extraction in China, resulting in severe reductions in groundwater levels across many regions [14,15]. This is particularly critical in the inland basins of Northwest China, where scarce precipitation and limited surface water availability make groundwater the primary source for agricultural, industrial, and domestic needs [11,16,17]. Recent studies have identified Northwest China as one of the most groundwater-stressed regions globally, with annual depletion rates reaching 10–20 mm [18]. The consequences include ecosystem degradation, agricultural yield reduction, intensified water conflicts, and land subsidence [19,20]. Despite the evident crisis, a comprehensive understanding of the relative roles of climatic and anthropogenic drivers in groundwater depletion remains limited, particularly at the basin scale in arid Northwest China. Accurately identifying these drivers is essential for designing effective groundwater management strategies [9].

Conventional groundwater monitoring approaches, such as observation wells and hydrogeological models, provide valuable local information but are constrained by high costs, sparse spatial coverage, and difficulty in representing regional trends, especially in economically underdeveloped arid regions [21,22,23,24]. Furthermore, spatial heterogeneity in hydrogeological conditions introduces substantial uncertainty into basin-scale assessments [25]. In contrast, satellite-based remote sensing, particularly through the Gravity Recovery and Climate Experiment (GRACE) mission, is a powerful tool for monitoring terrestrial water storage anomaly (TWSA) changes over large spatial and temporal scales [26]. GRACE-derived data, combined with surface water and soil moisture estimates, enable the indirect inference of groundwater storage anomalies (GWSAs) and have been widely applied in data-scarce, water-stressed regions such as northern India, Central Asia, and the North China Plain [27,28,29,30]. These datasets have significantly enhanced our capacity to detect long-term groundwater trends and assess their driving mechanisms [31,32].

Groundwater dynamics are driven by a complex interplay of climatic factors (e.g., precipitation variability, snowmelt) and anthropogenic influences (e.g., irrigation, land use change, population growth) [33,34]. While it is widely acknowledged that human activities are a major contributor to groundwater depletion [35,36], effectively distinguishing between natural and anthropogenic contributions remains a methodological challenge [9]. Recent studies have attempted to resolve this issue using model-based water balance approaches. These approaches typically simulate groundwater storage under natural conditions using land surface or hydrological models and then compare the results with GRACE-derived GWSAs to isolate anthropogenic influences [28,37]. For instance, Liu et al. applied this approach across multiple global basins and effectively differentiated climate-induced and human-induced anomalies [9]. Such integrative methods provide a quantitative foundation for assessing the relative importance of different drivers and supporting targeted water management strategies.

The Shule River Basin, located in the westernmost part of the Hexi Corridor in Northwest China, represents a typical arid inland basin facing severe groundwater stress [38,39]. It originates in the Qilian Mountains and extends northwestward, covering a drainage area of approximately 102,300 km² [40]. The basin is primarily fed by glacial and snowmelt runoff, making it highly sensitive to climatic variability [41,42]. With limited surface water availability, groundwater serves as the main water source for agriculture, industry, and domestic use. Over the past two decades, the basin has experienced rapid agricultural intensification, urban expansion, and population growth, partly due to policy-driven ecological resettlement programs [43]. These transformations have substantially increased water demand, leading to continuous groundwater depletion, reduced natural recharge, and an increased risk of groundwater overdraft [44,45]. While previous research has examined aspects of surface water change and climate impacts [41], few studies have conducted comprehensive assessments of groundwater dynamics that incorporate both natural and anthropogenic drivers.

Despite its ecological and socioeconomic importance, the Shule River Basin remains under-represented in large-scale hydrological assessments. Existing studies tend to focus on local-scale observations or surface water processes, lacking a holistic basin-wide perspective [38,39,40,41,42,43,44,45,46]. For the influencing mechanism of groundwater change, scholars pay more attention to precipitation and temperature, actual evapotranspiration, water extraction activities, and farmland change, while less attention is paid to other influencing factors (such as potential evapotranspiration, transpiration, GDP, population migration, and other land types), and there is typically a weak emphasis on the relationship between groundwater consumption and compensation in the context of a natural background. Moreover, there is a lack of discussion on the comprehensive perspective of natural change and human development. To address this gap, this study employs an integrated approach combining GRACE satellite observations, land surface modeling, land use/land cover (LULC) data, and socioeconomic statistics to analyze groundwater storage dynamics in the Shule River Basin from 2003 to 2023. The specific objectives are as follows: (1) to quantify the spatiotemporal variations in GWSA using GRACE data and land surface modeling; (2) to separate the respective contributions of natural variability and human activities to groundwater changes via a water balance approach; (3) to explore the interactions between LULC changes, socioeconomic development, and groundwater dynamics, revealing the mechanisms through which human interventions affect groundwater systems. By integrating remote sensing, hydrological modeling, and socioeconomic analysis, this study provides empirical evidence of the relative roles of climate and human activity in shaping groundwater trends. The results are expected to inform the development of evidence-based water resource management strategies tailored to the unique challenges of arid inland basins. Furthermore, the methodological framework adopted here can be applied to other data-scarce, water-stressed regions globally, contributing to the broader aim of achieving sustainable groundwater management in the context of global change.

2. Study Area and Data

2.1. Study Area

The Shule River Basin is located in northwestern China and constitutes the westernmost section of the Hexi Corridor in Gansu Province (Figure 1). Originating from the Shaguolinamujimu Mountains in the western Qilian Mountains, the river flows northwest through several gorges into the Changmabao Basin. It then forms the expansive Changma alluvial fan—one of the largest in the Hexi Corridor—before continuing westward through Yumen City and Guazhou County, ultimately ending at Hala Lake in the Lop Nor Desert. Historically, the Shule River discharged seasonally into Lop Nur; however, extensive irrigation and water diversion have significantly reduced its flow, confining its terminus to Hala Lake in recent decades. The basin covers an area of approximately 102,300 km², with a main river length of about 861 km [40]. The basin is characterized by an arid continental climate, with low annual precipitation and a high evaporation rate. Precipitation is concentrated during the summer months, while winters are dry and cold. Meltwater from glaciers and seasonal snowpack in the Qilian Mountains provides a critical source of runoff, making the basin highly sensitive to climatic fluctuations [39,40,41]. Recent studies have reported rising temperatures and altered precipitation patterns in the upper basin, leading to significant changes in runoff dynamics and water availability [42]. The Shule River system includes several tributaries, such as the Changma, Shiyou, Baiyang, Tashi, and Dang Rivers. These tributaries contribute glacial meltwater and seasonal flows to the main channel, which traverses multiple basins and feeds a network of reservoirs and irrigation systems. The structural framework of the Shule River Basin has been shaped by multiple tectonic events since the Mesozoic era. A central depression formed during the Triassic era and subsided rapidly following the Yanshan Movement, accumulating substantial Cenozoic sediments [47]. Uplifted structural hills such as the Nanjie, Beijie, and Hanxia Mountains segmented the depression. From the late Tertiary era, especially after the Pliocene, the rapid uplift of the Tibetan Plateau intensified erosion, delivering large quantities of clastic material into the basin. Over time, the Shule River and its tributaries deposited gravels and cobbles, forming landforms such as alluvial fans, floodplains, and lacustrine plains.

The Yumen-Tashi Basin hosts both single-layer and phreatic-confined aquifers. The single-layer aquifer, composed of coarse, highly permeable materials, extends across the Changma–Yulin proluvial fan, reaching depths of up to 600 m near the Qilian Mountains (Figure 2). The well yields exceed 5000 m³/day in the Changma fan and range from 3000 to 5000 m³/day in the Yulin fan, decreasing to under 1000 m³/day at the northern edge [38]. Rivers are the main source of recharge. The phreatic-confined system, found in the fine-soil plain from the Gobi foreland to Beishan, consists of interbedded sands and clays. Its water table is shallow (1–5 m), with yields generally below 1000 m³/day. Thicker gravel layers (20–80 m) offer higher productivity (up to 5000 m³/day), while thinner sandy zones (3–10 m) yield much less productivity [38]. In the fine-soil plain, groundwater discharges in the form of springs and is lost through evaporation and transpiration. In the Guazhou Basin, aquifers transition from single to multilayer systems westward. The eastern piedmont hosts a 70 m thick single-layer aquifer of gravel, pebbles and sand, with water tables over 10 m deep. In contrast, the western alluvial plain contains multilayer phreatic-confined aquifers with declining yields. The aquifers comprise gravel, sand, fine-grained sand, and silty clay. The thickness is 25–30 m in the eastern plain, 30–40 m in middle of the Guazhou countryside, and below 20 m in Xihu Township. The water table in the aquifer is below 5 m, and the confined water head is 24 m above the surface. Well yields decrease from 4000 m³/day in the east to 700 m³/day in the west, as well as recording complex flow patterns [48].

The Changma alluvial fan serves as a key hydrological feature, promoting infiltration and groundwater recharge. Groundwater recharge primarily occurs through riverbed infiltration and return flow from irrigation, particularly in the middle and upper reaches of the basin. Groundwater is a crucial resource for agricultural, industrial, and domestic use throughout the Shule River Basin. Over the past few decades, increased extraction—driven by agricultural expansion and urban development—has led to substantial declines in groundwater levels, especially in the middle and lower reaches. Although localized management interventions have led to temporary stabilization in some areas, overexploitation remains a persistent challenge. The overall sustainability index of groundwater in the basin remains low, reflecting a growing imbalance between extraction and natural recharge rates [49].

2.2. Data

2.2.1. GRACE Data

Changes in TWSA can be retrieved from GRACE (Gravity Recovery and Climate Experiment) and GRACE-FO satellite observations using two main approaches: the spherical harmonic coefficient method and the Mascon (mass concentration block) method. Compared with the former, the Mascon method offers higher spatial resolution without requiring complex filtering, making it more effective for detecting localized hydrological signals [50]. Among the three agencies that provide Mascon solutions—CSR (the University of Texas Center for Space Research), JPL (the Jet Propulsion Laboratory), and GSFC (the Goddard Space Flight Center)—CSR products are widely used in hydrological studies due to their superior resolution and performance [8].

This study used CSR’s latest GRACE/GRACE-FO RL06 Mascon dataset to estimate GWSAs in the Shule River Basin. The dataset has a spatial resolution of 0.25° × 0.25° and covers the period from April 2002 to May 2024. To maintain the consistency of the spatial resolution of each data value involved in GWSA calculation and reduce its calculation error, we resampled the GRACE data using bilinear interpolation and adjusted its spatial resolution to 0.1° × 0.1°. Furthermore, missing data for between July 2017 and May 2018 were interpolated using singular spectrum analysis (SSA), supplemented with linear interpolation for short gaps [51].

2.2.2. GLDAS NOAH Land Surface Data

The Global Land Data Assimilation System (GLDAS), developed by NASA/GSFC and NCEP, integrates satellite- and ground-based observations through land surface models. Among the four available models (NOAH, CLM, VIC, and Mosaic), NOAH exhibits strong agreement with in situ observations and GRACE-derived estimates, especially for soil moisture [8]. We selected the GLDAS v2.1 NOAH model to extract land water components, including soil moisture, the snow water equivalent, and canopy water. The data have a spatial resolution of 0.25° × 0.25° and cover the period from April 2002 to May 2024. We resampled the GLDAS data using the same method (bilinear interpolation) and adjusted its spatial resolution to 0.1° × 0.1°. To ensure consistency with GRACE anomalies (referencing the 2004–2009 mean), the GLDAS variables were similarly normalized before use. Data were obtained from NASA’s Earthdata portal (https://disc.gsfc.nasa.gov (accessed on 24 September 2024)).

2.2.3. Groundwater Monitoring Data

The Observation well data provide reliable, localized measurements of groundwater levels and are commonly used to validate GRACE-based GWSA estimates through correlation analysis [52]. In this study, groundwater-level data from 16 observation wells in the Hexi Corridor were collected from the Gansu Provincial Department of Water Resources. These included four wells in the Shule River Basin. The selected records span the period from 2002 to 2024, matching the GRACE’s time range.

2.2.4. Other Data

To analyze the driving forces behind groundwater variability, we incorporated both climatic and anthropogenic indicators. Natural factors include precipitation, temperature, evapotranspiration (actual and potential), transpiration, runoff depth, and the NDVI. Human activity indicators include GDP, urban and rural population, cropland area, forest/grassland area, and the extent of impervious surfaces. Due to the limited availability of direct data on water consumption, GDP was used as a proxy for economic water use. Precipitation, temperature, and NDVI data were sourced from the Tibetan Plateau Data Center, with spatial resolutions of 1 km (temperature and precipitation) and 250 m (the NDVI). Evapotranspiration was derived from the GLEAM dataset at a resolution of 0.1° (https://www.gleam.eu/ (accessed on 28 September 2024)). Land use data were obtained from the China Land Cover Dataset (30 m resolution) published by Wuhan University [53]. The runoff depth data were sourced from the ERA5 Land dataset at a resolution of 0.1° resolution (https://cds.climate.copernicus.eu/ (accessed on 28 April 2025)). The population and GDP data were collected from the Gansu Water Resources Bulletin. The auxiliary data ranged from 2003 to 2023.

3. Research Methods

3.1. Methods for Estimating Changes in Groundwater Storage

GWSAs were estimated based on a water balance approach using GRACE-derived terrestrial water storage anomalies (TWSA). By subtracting other water components—canopy water (CWSA), soil moisture (SMSA), snow water equivalent (SWEA), and surface water storage (SWSA)—from TWSA, the GWSA could be calculated as

$$GWSA=TWSA-CWSA-SMSA-SWEA-SWSA$$

(1)

where all components are measured in centimeters. In this study, we replaced SWSA with runoff depth data from ERA5-Land data, and lakes and wetlands were excluded from the calculation due to their minimal presence in the arid Shule River Basin.

3.2. Trend Analysis Using Slope Estimation

We analyzed the temporal trends regarding GWSA using linear regression and spatial trends using slope estimation on each grid cell. The slope reflects the rate of change over time and was calculated as [54]

$$slope={\displaystyle \frac{n{\sum }_{i=1}^n\left(i\times {GWSA}_i\right)-{\sum }_{i=1}^{n}i\times {\sum }_{i=1}^n{GWSA}_i}{n{\sum }_{i=1}^n{i}^2-{\left({\sum }_{i=1}^{n}i\right)}^2}}$$

(2)

where n indicates an increasing trend in groundwater storage, while a negative slope suggests depletion.

3.3. Relative Contribution

The GRACE-derived GWSA reflect the combined effects of both natural processes and human activities. To separate their respective contributions, a natural-state GWSA was first estimated using precipitation (PREA), evapotranspiration (ETA), and other natural water balance components. Human-induced changes were then derived by subtracting the natural component from total GWSA [9,33]:

$${GWSA}_{natural}=PREA-ETA-CWSA-SMSA-SWEA-SWSA$$

(3)

$${GWSA}_{human}=GWSA-{GWSA}_{natural}$$

(4)

where PREA is the annual precipitation (cm), ETA is the annual evaporation (cm), CWSA is the annual canopy water (cm), SMSA is the annual soil water (cm), SWEA is the annual snow deep water (cm), and SWSA is the annual runoff depth (cm).

The relative contribution rates of natural factors and human disturbance to GWSA changes in Hexi Corridor were calculated as follows:

$$\begin{array}{c}{C}_{natural}={\displaystyle \frac{{GWSA}_{natural}}{\left|{GWSA}_{natural}\right|+\left|{GWSA}_{human}\right|}}\end{array}$$

(5)

$$\begin{array}{c}{C}_{human}={\displaystyle \frac{{GWSA}_{human}}{\left|{GWSA}_{natural}\right|+\left|{GWSA}_{human}\right|}}\end{array}$$

(6)

3.4. Land Use Transition Matrix

Land use change is a key indicator of human activity. To examine spatial transitions and quantify land conversion processes in the Shule River Basin from 2003 to 2023, a land use transition matrix was constructed. This matrix captured the area of land converted from one type to another:

$$\begin{array}{c}{Q}_{ab}=\left(\begin{array}{ccc}{Q}_{11}& \dots & Q_{1n}\\ ⋮& \ddots & ⋮\\ Q_{n1}& \dots & Q_{mn}\end{array}\right)\end{array}$$

(7)

where $Q$ is the area, $a$ is the initial land use type at the start of the study period, and $b$ is the land use type at the end of the study period. Moreover, $Q_{ab}$ is the area in which the initial land use type was changed from type $a$ to land use type $b$, and $n$ is the number of land use types. This method facilitated the assessment of land use dynamics and human-induced surface disturbances in the basin [32].

4. Results

4.1. Reliability of Groundwater Storage Evolution

We utilized data from groundwater monitoring wells that coincided both spatially and temporally with GRACE observations in the Shule River Basin to assess the accuracy of the inversion outcomes (

Table 1. Correlation between measured groundwater-level data and GRACE estimates.

Well	Location	Longitude	Latitude	Elevation/m	Monitoring Depth/m	Correlation/r
W1	Zhonggou	95.91	40.51	1199.12	35	0.54
W2	Jiulian	95.82	40.50	1181.32	35	0.69
W3	Shuangta	96.36	40.47	1339.35	15	0.74
W4	Yinma	97.01	40.43	1410.64	50	0.54

). Correlation analysis revealed a moderate but statistically significant positive relationship (r > 0.5) between observed groundwater levels and GRACE-derived estimates across four monitoring sites. The validation results suggest that while GRACE data can capture general groundwater trends in the Shule River Basin, local-scale variations may be more difficult to detect due to the relatively low spatial resolution of GRACE and the limited magnitude of groundwater fluctuations in the basin. This finding lays a solid foundation for future hydrological research in the region. Figure 3 presents the temporal evolution of GWSAs in conjunction with water-level changes from representative monitoring wells, offering a clearer visual interpretation of the positive associations between GWSAs and groundwater levels.

Moreover, the Shule River Basin exhibited a small degree of groundwater variation, as illustrated in Figure 3. This is consistent with the basin’s lower population density, reduced urban footprint, and relatively limited surface water recharge sources, all of which contribute to a more stable but slowly declining groundwater regime. The modest correlation may also reflect the fact that groundwater dynamics in the Shule River Basin are more influenced by slow and diffuse recharge processes, which are harder to resolve with GRACE’s coarse resolution (0.25° × 0.25°). Nevertheless, the statistically significant correlation and consistent trend direction support the applicability of GRACE data for long-term groundwater monitoring in the Shule River Basin. While GRACE may not fully capture localized fluctuations, its ability to reflect basin-scale changes confirms its utility as a complementary data source for hydrological assessments and water resource management in this arid inland region.

4.2. Temporal Variability in Groundwater Storage

Between 2003 and 2023, GWSAs in the Shule River Basin exhibited a sustained decreasing trend, characterized by moderate interannual fluctuations. The linear regression slope of annual GWSA change was approximately −0.31 cm·yr⁻¹, indicating persistent groundwater depletion throughout the two-decade period (Figure 4).

The rate of change varied across different sub-periods. From 2003 to 2006, groundwater depletion was relatively moderate, averaging −0.23 cm·yr⁻¹. During 2007–2017, the rate of depletion accelerated significantly to approximately −0.45 cm·yr⁻¹, coinciding with the basin’s most intensive phase of agricultural expansion and urban development. A notable slowdown in depletion was observed between 2018 and 2020, with an almost negligible rate of −0.02 cm·yr⁻¹. This short period of stabilization was likely due to reduced water demand and slightly more favorable hydroclimatic conditions. However, from 2021 to 2023, the decline resumed, with a rate of approximately −0.19 cm·yr⁻¹.

Seasonal variability in GWSAs was also evident. Groundwater levels typically declined from April to October, reaching their annual minimums in the latter month, a pattern corresponding to the period of intensive agricultural irrigation and peak vegetation water consumption. Recharge generally occurred between November and February, driven by reduced evapotranspiration and delayed infiltration from snowmelt and residual precipitation. However, the degree of recovery was limited, reflecting the arid climate and relatively low recharge potential in the region.

Intra-annual analysis revealed that groundwater storage typically increased slightly from January to March, before declining steadily through the summer and early autumn, followed by minor recovery during winter (Figure 5a). These dynamics highlight the imbalance between groundwater consumption and recharge, particularly in a region where natural recharge processes are slow and seasonally delayed. A consistent lag of approximately 4–6 months was observed between peak precipitation and peak groundwater levels, indicating a delayed recharge response to meteorological inputs due to slow infiltration through unsaturated zones and long travel times from mountainous recharge areas.

Monthly GWSA trends revealed that sharp declines typically occurred during March–April, July–September, and December (Figure 5b). These months correspond to periods of high irrigation demand and strong evapotranspiration, especially in the context of rising temperatures. Conversely, minor increases or stabilizations in groundwater were observed in January–February and October–November, coinciding with reduced vegetation activity and minimal irrigation.

4.3. Spatial Variation in Groundwater Storage

The spatial distribution of GWSA trends across the Shule River Basin exhibited significant heterogeneity. Based on pixel-level slope analysis, the central and southeastern regions of the basin experienced the most pronounced groundwater depletion, with localized rates of up to −0.52 cm·yr⁻¹ due to intensive agricultural practices, groundwater-dependent irrigation systems, and increasing urban water demand (Figure 6). In contrast, parts of the southern mountainous areas showed slightly positive GWSA trends, with slopes of up to +0.09 cm·yr⁻¹. These regions serve as important recharge zones due to snowmelt and limited water abstraction, and the observed increase in groundwater storage may reflect effective localized recharge mechanisms. These positive anomalies, although limited in extent, suggest that natural recharge still occurs in some headwater areas, particularly during wet years or winters with heavy snowfall. The spatial distribution of groundwater trends correlates strongly with land use intensity and hydrogeological features. Areas with concentrated human activities, particularly irrigated agriculture and expanding urban centers, demonstrated the most significant groundwater losses. Meanwhile, more remote or ecologically protected zones showed stable or slightly improving groundwater conditions.

4.4. Attribution of Natural and Anthropogenic Contributions

To disentangle the influences of natural processes and human activities on groundwater dynamics, total GWSAs were separated into natural and anthropogenic components using the water balance approach. The results showed that natural factors, such as precipitation, evapotranspiration, and runoff, were the dominant drivers of groundwater change in 80.95% of the study years. However, in the remaining 19.05% of study years, human activities played a greater role (Figure 7). During 2007–2017, despite relatively high annual precipitation (average 91.6 mm), intensive groundwater withdrawal for agriculture and increasing urban demand led to continued and rapid depletion. In contrast, the 2018–2020 period experienced slightly lower precipitation (88.9 mm), yet the slowdown in anthropogenic water use helped to stabilize GWSAs. From 2021 to 2023, both a significant decrease in precipitation (57.9 mm) and increasing human pressure led to renewed and accelerated groundwater loss (Figure 7).

Since 2016, in several years, the relative contributions of human activities have exceeded 50%, reflecting a shift in the groundwater regime. Several factors contributed to this change. Urban centers such as Guazhou and Yumen experienced significant population growth, while irrigated agriculture that heavily relied on groundwater continued to expand. At the same time, rapid economic development, particularly in industry and services, fueled further land use changes. These developments also led to an increase in the extent of impervious surfaces, limiting water infiltration. These trends highlight the growing dominance of anthropogenic controls over the basin’s groundwater system, particularly under conditions of climatic stress.

Additionally, climatic analysis revealed warming and drying trends over the study period (Figure 8a). While potential evapotranspiration increased, actual evapotranspiration declined, indicating reduced moisture availability (Figure 8b,c). Although ecological restoration programs improved vegetation cover, as reflected by increases in the NDVI and transpiration, these efforts also enhanced water consumption from the subsurface, further straining groundwater reserves (Figure 8d). Surface runoff depth declined continuously, limiting natural recharge opportunities (Figure 8e). Under natural conditions, the imbalance between decreasing recharge and increasing consumption has become more pronounced, even without considering direct anthropogenic abstraction.

In terms of human activities, while the total population remained stable, the urban population increased significantly at the expense of the rural population (Figure 9a), reflecting a trend of urbanization. The increase in urban water demand changed the original pattern of water resource distribution (overuse of surface water will reduce the supply of groundwater, and the exploitation of groundwater will exacerbate the groundwater deficit). This increased reliance on groundwater and aggravated the burden on water resources in dry areas. Economic indicators showed consistent GDP growth, particularly in agriculture (large scale, low output) and developing industrial and service sectors (small scale, high output, high potential) (Figure 9b). These trends reflect increasing water demand due to economic development, while also posing a serious challenge to local sustainable development.

4.5. Land Use Changes and Socioeconomic Drivers

Over the two-decade study period, the Shule River Basin experienced substantial land use changes (Figure 10a–d). Between 2003 and 2023, the areas of cultivated land increased by 50.5%, forest cover by 67.8%, grassland by 4.5%, and impervious surfaces by 104.6%. Tillage activities in arid regions, such as leveling, compaction, and deep tillage, can increase soil density and reduce its permeability, allowing more precipitation to form surface water and increase evaporation. With the expansion of forest and grassland, on one hand, the canopy and litter trap water increased precipitation evaporation. On the other hand, the transpiration process caused by plant roots also reduces groundwater recharge. The impervious area can reduce the percolation rate and prolong the confluence process, and its expansion will increase the surface water and reduce groundwater recharge. The change in land use reshapes the groundwater recharge process, resulting in a reduction in groundwater reserves. These transitions reflect both government-driven land development policies and spontaneous socioeconomic shifts.

The most significant land conversions involved the transformation of grassland and bare land into cropland. Specifically, 22.6% of the 2023 cropland area was converted from former grassland, while 16.3% was converted from bare land. Forest expansion primarily occurred through the conversion of grassland, accounting for 31.9% of the total forest area in 2023. Similarly, 20.9% of grassland gains came at the expense of bare land. Impervious surface expansion was also notable, with 10.6% derived from cropland, 25.2% from grassland, and 15.3% from bare land (Figure 10e). These changes reflect accelerating urbanization and infrastructure development, especially around major towns and transportation corridors. Simultaneously, surface water bodies shrank by an average of 8.17 km² per year, representing a 23.7% reduction over the study period. Many of these water bodies were replaced by bare or developed land, indicating reduced surface recharge potential and intensified hydrological stress. Overall, the expansion of cultivated land and impervious surfaces reduced infiltration zones and altered surface runoff patterns, thereby weakening the basin’s natural groundwater recharge capacity.

4.6. Correlations with Environmental and Human Drivers

Pearson’s correlation analysis revealed that GWSAs had significant relationships with both environmental and socioeconomic variables (Figure 11a). Negative correlations were observed between GWSAs and several human-related factors. The correlation coefficient between GWSAs and GDP was −0.91, suggesting that economic expansion, particularly in water-intensive sectors, strongly contributes to groundwater depletion. Impervious surface area also had a strong negative correlation (r = −0.86) with GWSAs, indicating that urbanization directly limits groundwater recharge. Cropland area (r = −0.95) and potential evapotranspiration (r = −0.84) similarly displayed negative relationships, highlighting the dual pressures of irrigation and climatic water stress. The NDVI, though commonly associated with ecological recovery, was also negatively correlated with GWSAs (r = −0.85), likely due to increased transpiration from denser vegetation cover. Conversely, positive correlations were found with actual evapotranspiration (r = +0.78), runoff depth (r = +0.59), and rural population (r = +0.83). These factors reflect the more sustainable water use patterns and greater recharge potential present in areas with less infrastructure development.

A contribution analysis showed that urban population growth alone accounted for 31.3% of the observed groundwater decline, making it the most significant human-related driver (Figure 11b). This was followed by transpiration (12.3%), actual evapotranspiration (12.0%), and potential evapotranspiration (10.7%). These results underscore the multifaceted nature of groundwater stress in the basin, shaped by both direct withdrawals and indirect land surface processes. In summary, groundwater depletion in the Shule River Basin has been driven by a combination of natural and human factors. While climate variability continues to shape the broader recharge potential, intensified human activities, particularly agricultural expansion, industrialization, and urban growth, have become the dominant forces behind long-term depletion trends. Land use modifications, economic development, and changing water consumption patterns are reshaping the hydrological regime and will likely continue doing so in the coming decades.

5. Discussion

5.1. Impacts of Human Activities and Climate Change on Groundwater Storage

Groundwater depletion in the Shule River Basin is driven by the combined effects of long-term climatic trends and intensifying human activities. While natural hydroclimatic variability provides the environmental backdrop, it alone cannot fully explain the observed trends in GWSAs. Our decomposition analysis indicates that since 2016, anthropogenic factors, particularly land use changes and groundwater abstraction, have surpassed climatic influences as the dominant driver of groundwater decline across much of the basin. Climatic trends in the basin align with the broader warming and drying patterns observed in arid inland regions of Northwest China [8,55]. Although average annual precipitation has remained stable or declined slightly, rising temperatures have led to increased potential evapotranspiration, thereby widening the gap between water input and atmospheric demand. This has significantly reduced natural recharge potential, especially in summer. Under this climate change model, the characteristics of the imbalance in the groundwater consumption–recharge relationship are very similar to those in Central Asia [56]. Nevertheless, natural processes account for only about 61% of the total groundwater decline during the study period, meaning that nearly half (39%) of the loss is attributable to human activities.

Among anthropogenic drivers, the expansion of cropland and impervious surfaces has substantially altered surface hydrology and land–atmosphere interactions. Increased water demand for irrigation and urban use has intensified groundwater extraction, exacerbating aquifer imbalances. These findings are consistent with other studies in arid inland basins [57,58]. For instance, ref. [59] identified human interference as the main cause of groundwater decline in the Weiku Oasis (Tarim Basin), while ref. [60] found that over 80% of groundwater depletion in the Yaoba Oasis was human-induced. Similar conclusions were drawn in the Ebinur Lake Basin, where cropland and oasis expansion significantly altered local hydrology [61]. In addition, similar studies in Central Asia have shown that increased water use for irrigation and industry has exacerbated regional water shortages [56,62]. In the Sahara region of North Africa, which is at a similar latitude to our study area, increases in irrigation area and population density were the key factors affecting GWSAs [63,64].

5.2. Spatial Heterogeneity of Depletion and Localized High-Risk Zones

Groundwater depletion in the Shule River Basin displays marked spatial heterogeneity, with high-risk zones concentrated in specific regions. The most severe depletion is observed in the central basin, particularly in the Changma alluvial fan. This area is characterized by a high dependency on groundwater for irrigation, rapidly expanding urbanization, and limited recharge potential due to surface sealing and reduced soil permeability. Such spatial concentration of groundwater stress mirrors global patterns observed in other arid basins [65,66], where unconfined aquifers in alluvial fans are especially vulnerable to overexploitation. In contrast, the southern high-altitude regions of the basin exhibit slightly positive GWSA trends, mainly due to periodic recharge from glacial and snowmelt sources in the Qilian Mountains [42]. However, these recharge events are neither spatially extensive nor temporally reliable, and they are insufficient to compensate for deficits in more intensively exploited sub-basins. These patterns underscore the need for sub-basin-scale management strategies tailored to local hydrological conditions and land use dynamics.

5.3. Recharge Inhibition and Overextraction Due to Land Use Change

The interplay between land surface modification and increasing water demand has profoundly disrupted the groundwater balance in the basin. One of the most significant anthropogenic impacts is the inhibition of natural recharge caused by the spread of impervious surfaces [67]. Urban expansion, infrastructure development, and industrialization have increasingly sealed the land surface, blocking infiltration pathways and increasing runoff [68]. Additionally, the conversion of natural grasslands and barren lands into croplands has altered soil structure, increased compaction, and changed vegetation cover, further reducing infiltration and increasing evapotranspiration [69,70]. On the demand side, groundwater withdrawals have steadily increased. Urban centers such as Guazhou and Yumen have seen population growth and industrial development, which together have increased domestic and industrial water consumption. Even though rural populations are declining, agricultural water use continues to increase due to expanded planting areas, double-cropping systems, and longer irrigation seasons [12,71]. Key agricultural zones such as Changma (in the Yumen-Tashi sub-basin), Huahai, and Shuangta (in the Guazhou sub-basin) rely heavily on groundwater for irrigation [42]. By 2020, the local population reached 540,000, largely due to the influx of agricultural labor. These trends highlight the urgent need for both structural reforms (e.g., land use planning, irrigation infrastructure) and behavioral changes (e.g., water-saving practices) to restore balance between groundwater supply and demand.

5.4. Policy Implications and Sustainable Management Strategies

The Shule River Basin, long recognized for its sensitivity to climate change, faces increasing water scarcity and ecological vulnerability [72]. The evidence presented in this study emphasizes the need for comprehensive and coordinated groundwater governance. Lessons from the Shiyang River Basin offer valuable insights: there, a set of management interventions, including reductions in cultivated land, limits on groundwater abstraction, inter-basin water transfers, water rights trading, and crop structure adjustments, has helped to stabilize groundwater levels in recent years [73,74]. Based on this context, we propose a set of integrated strategies for sustainable groundwater management in the Shule River Basin. First, enhancing groundwater recharge is essential, particularly in alluvial fan areas with favorable hydrogeological conditions. Runoff from snowmelt or rare storm events should be captured via infiltration basins or recharge wells. Second, stricter regulation of groundwater extraction is needed, especially in high-risk zones. A permit-based system combined with real-time monitoring could effectively control illegal or excessive withdrawals. Third, agricultural water use efficiency must be improved through adopting water-saving irrigation technologies (e.g., drip and subsurface irrigation), soil moisture monitoring, and optimized irrigation scheduling.

In parallel, urban planning and land use control should aim to curb impervious surface expansion by encouraging the construction of green infrastructure and permeable pavements. Zoning policies should also mandate the construction of stormwater management systems to improve infiltration [75]. Meanwhile, groundwater monitoring and modeling capacity must be strengthened by expanding the observation network and integrating remote sensing tools such as GRACE, which can enhance spatial resolution and data accuracy [25,26,27,28]. Furthermore, integrated basin-scale water management should be adopted, drawing on cross-sector collaboration and adaptive planning under climate uncertainty. Finally, inter-basin water transfers, particularly the western route of China’s South-to-North Water Diversion Project, are expected to ease regional water stress. In sum, achieving sustainable groundwater use in the Shule River Basin will require joint efforts from governments, water users, and researchers. Without timely, coordinated, and science-based interventions, continued groundwater depletion could jeopardize long-term water security, agricultural productivity, and ecosystem resilience in this fragile arid environment.

6. Conclusions

This study investigated the temporal and spatial dynamics of GWSAs in the Shule River Basin of China’s Hexi Corridor from 2003 to 2023, using GRACE satellite data combined with land surface modeling and socioeconomic indicators. A contribution separation method was applied to distinguish between natural and anthropogenic drivers of groundwater change. Based on our analysis, the following key conclusions can be drawn:

The Shule River Basin has experienced a sustained decrease in groundwater storage over the past two decades, with an average annual GWSA decrease of approximately −0.31 cm·yr⁻¹. The seasonal pattern reveals that the lowest groundwater levels occur in late summer and early autumn, aligned with peak irrigation demand. The most severe depletion was observed in the central alluvial fan regions, especially near Guazhou, where groundwater extraction is intense and land surface modification is extensive. In contrast, southern mountainous recharge areas showed relatively stable or slightly positive trends, highlighting localized recharge potential.

While natural factors (e.g., reduced precipitation, increasing evapotranspiration) accounted for the majority of early groundwater loss, the relative contribution of human activities has steadily increased, particularly since 2016. Land use transformation, urbanization, and economic expansion have significantly amplified groundwater stress. Under natural conditions, the recharge–consumption relationship was already imbalanced. Human activities intensified groundwater depletion, exacerbating this imbalance. The expansion of impervious surfaces and conversion of grasslands into croplands have impaired natural recharge pathways. Simultaneously, increased industrial and urban water demand has led to increased groundwater withdrawals, placing a double burden on aquifer systems.

The Shule River Basin is approaching a critical threshold in groundwater sustainability. Proactive measures are urgently needed, including artificial aquifer recharge, improved water use efficiency, urban land use control, and the development of integrated water management frameworks tailored to arid inland basins. Among natural factors, potential evapotranspiration and the NDVI had significant negative impacts on groundwater. Among anthropogenic factors, population mobility, land use changes, and industrial development were closely related to groundwater loss.

This study is limited by the lack of high-resolution groundwater observation data and glacier meltwater estimates in the upper basin. Future research should incorporate glacier hydrology, unsaturated zone modeling, and socioeconomic scenario simulations to better understand long-term groundwater evolution and inform adaptive management strategies.