Characteristics and causes of groundwater level dynamics in the Changji oasis on the northern slopes of the Tianshan Mountains, China Open Access

Groundwater is a critical component of the hydrological cycle and plays a vital role in sustaining ecosystems and supporting human activities (Zhang et al. 2014; Yue et al. 2020). Groundwater resources serve as a major water source for agriculture and industry in most countries around the world (Salem et al. 2018; Sohoulande Djebou et al. 2021), particularly in arid and semiarid areas (Zamanirad et al. 2018). Compared to limited surface water resources, groundwater is vital for sustaining local agriculture, and for domestic and industrial use (Dinka et al. 2014; Lv et al. 2019). Climate change, human activities, and other factors cause notable spatial and temporal changes in groundwater recharge, runoff, and discharge relationships (Yue et al. 2020). Groundwater overexploitation occurs when groundwater extraction exceeds natural and other recharge for a prolonged period, leading to alterations in the hydrogeology of aquifers (Bagheri et al. 2019). The amount of groundwater resources is closely related to the spatial and temporal variations in the groundwater levels, and changes in the groundwater level reflect the relationship between groundwater recharge and discharge (Pan et al. 2020). Therefore, the rise and fall of the groundwater level is the most visible indicator of changes in groundwater resources. Excessive groundwater exploitation can lead to a rapid decline in groundwater levels, causing ecological and environmental problems such as grassland degradation, lake and wetland atrophy in various locations, especially in arid inland river basins with large irrigation areas (Bagheri et al. 2019; Akbari et al. 2020). This paper focuses on the spatiotemporal variations and driving factors of the groundwater levels, which can aid in identifying ecological and environmental problems associated with groundwater and accurately evaluating the regional groundwater resources.

Despite the significance of groundwater resources, they are currently facing various crises due to declining groundwater levels in several regions, including northwestern Bangladesh (Salem et al. 2018), northwestern China and the North China Plain (Fan et al. 2008; Long et al. 2020), Iran (Zamanirad et al. 2018), and the southwestern Australia (Le Brocque et al. 2018). Groundwater levels are a critical control indicator in groundwater management (Ma et al. 2015), influenced by natural factors such as precipitation, surface runoff, evaporation, tidal forces, and earthquakes caused by tectonic compression (Mcmillan et al. 2019; Wilopo et al. 2021). Nonetheless, it is the high-intensity human activities that are the primary drivers of groundwater level fluctuations (Zhang et al. 2013; Lin et al. 2020). For example, the rapid increase in urbanization and built-up areas can lead to a reduction in vegetation and agricultural areas, while increasing impervious ground coverage areas. As a result, the recharge rate of groundwater is reduced, and the groundwater level in the area is greatly reduced (Haq et al. 2021; Nath et al. 2021). In areas where irrigated agriculture is the primary economic activity, agricultural practices consume a significant amount of water resources (Du et al. 2020; Pan et al. 2020). This is particularly significant in arid regions, and Liu et al. (2018a) noted that the rapid expansion of irrigated agriculture has further exacerbated the water use conflict between irrigated agriculture and the ecological environment. Additionally, socioeconomic development has prompted local farmers to shift from food crops to cash crops with high water consumption (Zhang et al. 2014). In arid regions where surface water is scarce, massive groundwater extraction to meet irrigation needs has led to a significant drop in the groundwater level (Kang & Kaur 2018; Guo et al. 2021). However, some studies have found that the implementation of water transfer projects (Chen et al. 2016), the improvement of channel lining (Xu et al. 2010; Altafi Dadgar et al. 2018), and the popularization of efficient water conservation facilities (Zhang et al. 2013) have led to a relative decrease in channel leakage and irrigation infiltration, resulting in a decrease in groundwater recharge (Mi et al. 2020). In addition to human activities, climate change can alter temperature and precipitation patterns, impacting groundwater recharge (Salem et al. 2018). However, in some areas, groundwater level fluctuations do not respond significantly to climate change. Therefore, the effect of climate change on groundwater varies in different regions (Guo et al. 2021). The North China Plain has a long history of water problems, and Chinese scholars have determined the partition of groundwater influencing factors by spatially superimposing topography, groundwater level, groundwater exploitation degree, groundwater funnel, rivers and lakes, concluding that groundwater level dynamics are comprehensively affected by different influencing factors (Wang et al. 2009). Others have used backpropagation-artificial neural networks to simulate the response of the groundwater level to natural factors and human activities and carried out a sensitivity analysis. The results showed that the irrigation area and water-saving irrigation area were the most important factors affecting groundwater level dynamics (Sun et al. 2020). Therefore, gaining an in-depth understanding of the factors that affect the long-term dynamics of groundwater levels in arid irrigated agricultural areas is crucial, and it is the foundation for maintaining sustainable use of groundwater resources and promoting healthy agricultural development.

In the past few decades, the geostatistics method (Hu et al. 2019), Gravity Recovery and Climate Experiment (GRACE) (Lin et al. 2020), isotope tracer method (Qiao et al. 2020), and numerical simulation method (Zhou et al. 2020) have been used by many scholars to reveal the spatiotemporal variations in groundwater fluctuations. Correlation analysis (Du et al. 2020) and gray correlation analysis (Liu et al. 2018b) have been used to analyze the relationship between groundwater level dynamics and its factors, and good results have been achieved (Lin et al. 2020). Due to the complexity of geological structures and the uncertainty of groundwater systems (Qiao et al. 2020), the relationship between groundwater dynamics and climate change, human activities, and groundwater circulation is extremely complex. The spatiotemporal variations in groundwater and changes in the hydrological cycle have not been fully studied in arid inland river basins (Wang et al. 2020). At present, research on groundwater resources mostly focuses on dynamic monitoring, pollution control, water quality evaluation, and prediction models (Huang et al. 2021). However, the analysis of groundwater level dynamics has always been a hot research topic, but in general the number of groundwater monitoring wells can limit the research work. For groundwater level declines that are highly localized, geostatistical methods are a very useful tool that differs from traditional statistical methods but is not limited by the low spatial resolution of the GRACE technique (Döll et al. 2014) and helps to reveal the spatial patterns of regional variables. Coupled with the fact that Spearman correlation analysis does not require a normal distribution for the data and has a wide range of applicability, this method can be used to analyze the groundwater level drivers more accurately.

In Changji Hui Autonomous Prefecture (CJHP), due to the excessive groundwater overdraft in the plains for agriculture, production and domestic consumption, the groundwater level has continued to drop from 2000 to 2020. As the groundwater level drops and the aquifer is drained, the storage capacity of groundwater decreases, and the cost of water exploitation also increases. At the same time, due to groundwater overdraft, the flow of springs is greatly reduced in overexploited areas. The spring flow rate of the phreatic overflow zone is less than 10–30% of the original, and sometimes even disappears. In recent years, the Chinese government has attached great importance to the problem of groundwater overexploitation in the CJHP. There are relatively complete long-series groundwater level monitoring data in the CJHP. Many Chinese scholars have also analyzed the trend of groundwater level changes in this area, but the root cause of the decline in groundwater level has not yet been identified (Xiong et al. 2012; Jia et al. 2014; Wei et al. 2017). However, the groundwater level after treatment is still too low. So, this research comprehensively analyzes various activity factors and the reasons for the declining groundwater, which is an urgent need for efficient treatment of overexploitation areas in the CJHP.

Large Quaternary unconsolidated accumulations occur on the northern slopes of the Tianshan Mountains. Groundwater is mainly stored in loosely deposited sediments in the southern margin of the Junggar Basin. In the upper part of the alluvial–diluvial fan, Quaternary unconsolidated accumulation is mainly gravel and has a singular structure, coarse particles, large aquifer thickness, strong water content, and excellent water quality. In the middle of the alluvial–diluvial fan, the aquifer is mainly gravels and pebbles, the sand content is relatively high and the aquifer abundance is greater than that in the submerged zone. From the lower part of the alluvial sector to the margin of the alluvial sector, the slope of the terrain becomes shallow, the particles in the aquifer become fine, phreatic runoff is blocked, and phreatic water spills over the surface in the form of spring water. Phreatic water and multilayer confined water in the alluvial plain and lacustrine plain reside below the alluvial sector margin. The strata are fine sand, silty soil, silty clay, and interbedded clay. The abundance of the aquifer changes from strong to weak from south to north or from southwest to northeast. Due to decades of large-scale exploitation and utilization, most artesian wells have reduced the water head.

Groundwater recharge in the CJHP comes from surface water conversion, lateral inflow, and a small amount of precipitation infiltration. The mountainous area is the water resource formation area, and the precipitation in the mountainous area forms surface runoff. After the river flows through the mountain pass, on the one hand, it can recharge groundwater through seepage along the river course, and on the other hand, it can recharge groundwater through seepage along the canal system and field irrigation infiltration. The plains area has less precipitation and little direct infiltration. Abundant groundwater is stored in the thick and loose deposits. However, the groundwater runoff direction is from the piedmont sloping plain, from south to north, to the downstream fine soil plain and desert area, and the hydraulic gradient is 2.2–16.6‰. At the margin of the alluvial–diluvial fan, groundwater runoff is blocked, a spring overflow zone is formed, or groundwater runoff flows down through the middle and lower aquifers. The main methods of groundwater drainage are artificial exploitation and slight phreatic evaporation and lateral outflow.

In this article, monthly DGL data from 2000 to 2020 were collected from the Changji Water Conservancy Bureau. After screening of groundwater level monitoring data, some monitoring wells collapsed or ceased to be used, so the number of available monitoring wells was 126 (Figure 2 for the monitoring well distribution). In addition, the data on the amounts of groundwater exploitation, and surface water diversion from 2000 to 2020 were collected from the local Water Resources Center. Runoff data (2000–2020) within Changji Prefecture were obtained from the Changji Prefecture Hydrological Bureau, and included Kenshiwat hydrological station, Shimen hydrological station, Santun River hydrological station, Toutun River hydrological station, Baiyang River hydrological station, Kaiken River hydrological station, and Yuejin Reservoir inlet flow, with monthly scale data of runoff from seven stations.

Annual precipitation data from 2000 to 2020 were obtained from the China Meteorological Data Network (http://data.cma.cn), and remote sensing image data from 2000, 2010 and 2020 were obtained from the Data Center for Resources and Environmental Sciences, Chinese Academy of Sciences (http://www.resdc.cn) using Landsat Thematic Mapper (TM), Enhanced Thematic Mapper (ETM) and Landsat 8 as image data sources, with a resolution of 30 m. Remote sensing image data were preprocessed by geometric correction and radiation distortion correction. The land use types in the study area can be divided into six categories: forestland, cropland, grassland, water area, and urban and rural residential land. The consistency of land use data processing and the reliability of interpretation accuracy is guaranteed by the human–computer interaction visual interpretation method and field investigation verification.

Due to the large number of groundwater monitoring wells in the study area the different locations and aquifer media cause great differences in the DGL; thus, the average value of all DGL data in the study period is calculated, the average is used to represent the change in the DGL in the region with great error, and some data monitoring times are covered by only 1–3 years. This paper refers to Hu's pretreatment method for discontinuous groundwater monitoring data (Hu et al. 2019). The data for which the duration of some monitoring data cannot cover the whole study period are filtered and processed, and the annual cycle of all data is changed into the spatial average for comparison with other data (such as runoff and precipitation) in the subsequent study.

Within the study area, the monthly monitored values of groundwater levels exhibit significant intra-annual variations, primarily influenced by the extraction of groundwater for agricultural irrigation. It is well recognized that during the spring season, most crops are in the seedling stage, which requires substantial amounts of water for growth. However, the study area is situated in a northwestern arid region with an annual precipitation of only 231 mm, and rainfall is predominantly concentrated in the summer months. The natural precipitation is insufficient to meet the basic water requirements for crop growth, necessitating extensive groundwater extraction for agricultural irrigation. Consequently, there is an immediate decline in the groundwater levels. Agricultural irrigation can continue until about August, when the irrigation period for high water-demand crops such as cotton comes to an end, and the groundwater level drops to its lowest point of the year. The groundwater level gradually recovered and returned to a higher point around March of the next year. Taking the monitoring data in Fukang irrigation district as an example, the difference between the highest water level and the lowest water level in the year can reach 22.58 m, and agricultural irrigation has brought serious pressure to the local groundwater resources.

The DGL in the CJHP plains area generally decline from 2000 to 2020, as shown in Figure 3. First, the declining DGL of groundwater monitoring wells in HTB, QT, and FK is the most obvious in areas with scarce surface water and extensive cropping, so that groundwater becomes the main source of irrigation water. The groundwater level declined rapidly with values of 29.80, 30.35, and 21.56 m, respectively, with annual decline rates of 1.42, 1.45, and 1.20 m/year, respectively. Second, the plains area is located in the desert near northern ML. Due to the lack of surface water transport, the irrigation area relies on groundwater for agricultural irrigation, and the groundwater level drops significantly, with a depth of 20.36 m and an annual rate of decline of 0.97 m/a. The variation in the groundwater level in the western MNS can be divided into two stages: 2000–2014 and 2014–2020. The groundwater level decreased and fluctuated from 2000 to 2015, with a total decrease of 9.67 m and a decreasing rate of 0.69 m/a over 15 years. The groundwater level showed a rising trend from 2014 to 2019, with a total rise of 6.97 m. However, due to drought, scarce precipitation, and insufficient river inflow, groundwater exploitation increased, resulting in a decreasing trend. Therefore, groundwater in MNS decreased by 4.47 m from 2000 to 2020, with an average annual rate of decline of 0.21 m/a.

The areas with shallower DGL are mainly located in the fine soil plains, and the areas with deeper DGL are mainly located in the piedmont plains with larger aquifer media. In the past 20 years, the area where the DGL is less than 10 m in the WR has decreased from 3,236.20 to 409.57 km², a reduction rate of 87.34%; the area where the DGL is within 10–50 m has decreased from 2,321.31 km² in 2,000 to 4,862.31 km² in 2020, with a rate of increase of 109.46%; and the area where the DGL is greater than 50 m has increased by 28.22%. The groundwater level in the ER changed relatively little. The area where the DGL is less than 10 m decreased from 2,406.23 km² in 2000 to 1,203.43 km² in 2020, a decrease of 50.00%; the area of groundwater buried in the range of 10–50 m is reduced by 3% and basically remains unchanged; the DGL in the ER was relatively shallow compared with that in the WR. In 2000, the DGL did not exceed 50 m, and the area with a DGL greater than 50 m increased to 1,407.91 km² in 2020 in the ER.

The factors affecting the groundwater level can be mainly divided into two categories: The first is to change the groundwater level by affecting the groundwater storage capacity of the aquifer, such as precipitation infiltration, river leakage, groundwater phreatic evaporation, and groundwater exploitation activities. The other is due to the accumulation and release of stress–strain in the aquifer, the thickness of the aquifer, and lithology, such as earthquakes and tidal phenomena.

Since 2000, the WR has been dry in 2004, 2014, and 2020, the precipitation has decreased significantly in 2017, and the ER has been dry in 2001, 2008, 2014 and 2020, and in 2017. After that, the precipitation decreased significantly and continuously, and reached the lowest precipitation in the past in 2020. From 2017 to 2020, the runoff of rivers in the CJHP showed a downward trend, and the decline in runoff was more obvious in the ER than in the WR (Figure 5(b)).

Figure 6(a)–6(c) shows that the land use type in the CJHP is mainly cropland, followed by grassland, and the area of cropland in WR is more densely distributed than in the ER. In 2020, the area of cropland in the CJHP was 8,125.53 km², accounting for 52.27% of the study area; followed by grassland, with an area of 5,347.97 km², accounting for 34.40% of the area of the study area; 1,060.32 km² of unused land, 791.27 km² of settlement, and 208.57 km² of waters, accounting for 6.82, 5.09, and 1.34% of the total area, respectively; and forestland area was the smallest, accounting for only 0.08% of the total area (Table 1).

According to the three-phase LULC, the expansion of cropland mainly occurred from 2000 to 2010, the growth rate of cropland decreased significantly from 2010 to 2020, and the forest area decreased sharply from 2000 to 2010; the change in water area in the WR showed a trend of rising first and then falling, the other land mainly turned to grassland and cropland before 2010, and after 2010, there was a slight upward trend. The settlements in the ER region showed a trend of first decreasing and then increasing, the water area showed a slight upward trend, and the grassland showed a trend of first decreasing and then increasing.

The surface water supply in the ER reached a maximum of 7.95 × 10⁸ m³ in 2010, and then continued to decline. By 2020, the surface water supply reached 4.82 × 10⁸ m³. The groundwater exploitation reached a maximum of 10.64 × 10⁸ m³ in 2014, and then groundwater was exploited. The amount of groundwater has stabilized, but after 2018, the amount of groundwater exploitation has gradually increased, reaching 9.59 × 10⁸ m³ in 2020. The contribution rate of surface water to the total water supply in the ER decreased from 47% in 2000 to 33% in 2020, and the contribution rate of groundwater to the total water supply increased from 53 to 66% in 2000–2020.

The change in water consumption in various industries is affected by socioeconomic development and related policies on water resources (Figure 8(c) and 8(d)). However, the agricultural economy is the main development industry, and agricultural water consumption accounts for more than 92% of the total water consumption in the CJHP. After 2014, the government restricted groundwater exploitation for agricultural water consumption, the total water consumption began to decline, and the proportion of agricultural water consumption showed a downward trend.

As shown in Figure 8(a) and 8(b), the annual precipitation in the irrigation area is scarce, but the surface water supply and groundwater supply do not fluctuate with the increase or decrease of rainfall. Instead, they exhibit their own patterns of change. Before 2010, agricultural irrigation in the WR mainly relied on surface water. After 2010, due to the widespread promotion of drip irrigation technology, the change in the irrigation mode also affected the utilization mode of irrigation water, so groundwater gradually became the main water source of agricultural irrigation. Because there are fewer surface water resources in the ER, agricultural irrigation mainly relied on groundwater resources, and the popularity of drip irrigation technology was relatively short. After 2014, the large-scale popularization of drip irrigation belt technology caused the proportion of groundwater in agricultural water to reach 66% by 2020. According to the statistics of the total water consumption and consumption in the study area(Figure 8(c) and 8(d)), the local area mainly relies on groundwater, and agricultural water occupies a large proportion of groundwater exploitation, so agricultural irrigation water has a great influence on the groundwater level.

Among the natural factors, precipitation and river runoff have no significant correlation（R_W12 = −0.22, R_W13 = −0.12, R_E12 = 0.15, R_E13 = 0.08） with the mean value of groundwater level variation. However, there is a negative correlation between precipitation and groundwater exploitation in the WR (R_W25 = −0.46, p < 0.05). It can be determined that the DGL in the WR is generally large, and precipitation cannot directly recharge groundwater, but the increasing precipitation will reduce artificial irrigation time, during the growing period of crops ((Sohoulande Djebou et al. 2021), which in turn indirectly reduces groundwater extraction. The precipitation in the ER is larger than that in the WR, but the river runoff is much smaller than that in the WR. There is a significant positive correlation between rainfall and surface runoff at the 0.01 level (R_E23 = 0.89), so the change in river runoff is mainly affected by precipitation.

The correlation analysis results of the DGL changes and variables in the WR and ER are shown in Figure 9. Human activities, groundwater exploitation (Z2), domestic water consumption (Z4), industrial water consumption (Z5), agricultural water consumption (Z6), and environmental water consumption (Z7) are highly significantly correlated with the average change in the DGL (Z1) (P < 0.01). The surface water consumption water supply is significantly correlated with the average change in the DGL in three counties and cities in the WR (P < 0.05) and is highly significantly correlated with four counties and cities in the eastern region (P < 0.01). Therefore, groundwater extraction, agricultural irrigation area, agricultural water consumption, domestic water consumption, environmental water consumption and industrial water consumption are the dominant factors related to the change in groundwater level in the CJHP plains.

The effect of LULC on the groundwater level mainly depends on the infiltration capacity of land followed by the groundwater recharge, and the expansion of cropland increases the pressure on agricultural water use. However, due to the lack of continuous land use data, correlation analysis cannot be carried out with other data. In this paper, the distribution of the DGL (Figure 4) and land use type (Figure 6) are spatially superimposed by geographic information system (GIS) technology, and the DGL areas in different ranges of various land types are counted (Table 2). Table 2 shows that from 2000 to 2020, the DGL in the CJHP plains increased, and the land use types in this area were mainly grassland and cropland. The area in which the DGL is less than 10 m in the six land use types decreased over time until 2020, and the study area DGL was less than 10 m in the main grassland area. The DGL of grassland, cropland, and settlement were mainly in the range of 10–50 m. The DGL greater than 50 m in the WR was mainly in the cropland area, and the DGL greater than 50 m in the ER was mainly in grassland and cropland areas. The groundwater levels of QT and ML in the ER were significantly lower than those of FK and JMSR. The agricultural irrigation land in the ER was mainly located in QT and ML, which further shows the effect of agricultural irrigation on the groundwater level.

This research utilizes a 21-year time series of groundwater monitoring data (2000–2020) from the CJHP plain area, complemented by data on natural and anthropogenic factors, to deeply explore the spatial and temporal characteristics of groundwater level fluctuations. Furthermore, it employs advanced statistical methods to analyze in detail the correlations between groundwater level dynamics and both climatic and human activity factors, aiming to provide a comprehensive understanding of the influencing factors on groundwater levels in the region.

The overall groundwater level showed a continuous decline from 2000 to 2020, which was very similar to the trend performance of groundwater level changes in other arid regions (Guo et al. 2021; Wang et al. 2022b). However, the promulgation of water resource management regulations has reduced groundwater extraction efforts (Wang et al. 2022a). The rate of decline in groundwater levels has slowed, and in some cases, groundwater levels have even begun to rebound. For instance, in the WR, the groundwater level is rapidly recovering at a rate of 0.82 m/a. For the CJHP, which relies heavily on groundwater resources, the current situation of groundwater level rise only occurs to a small extent in western Changji City and eastern Mubi County. The spatial distribution of groundwater depth in the western part of the study area is characterized by a gradual shallowing and slow deepening of the GDL from south to north, mainly showing a large depth of burial in the southern alluvial fan, a small depth of burial in the central plain, and a deepening of burial in the northern part near the desert. In the east, the GDL in Mubi and Qitai counties is larger, and the GDL in the northwest direction gradually becomes shallower, except for the small change in the GDL in Jimsar County, and the large change in Fukang City, Qitai County and Mubi County. The phenomenon of rising groundwater levels in some areas after the implementation of the strictest water resource management system in 2014 indicates that the management of groundwater levels is feasible and reasonable (Wang et al. 2022a).

Even though this study covers data from 2000 to 2020, the dynamics of groundwater levels are a long-term process that requires a longer time series of data to verify and expand upon the current findings. Moreover, due to the potential uneven distribution of monitoring wells, the spatial resolution of this study may have limited the capture of small-scale variations in groundwater levels.

Climate change affects groundwater level dynamics and precipitation is one of the sources of groundwater recharge (Kaur et al. 2015; Ma et al. 2015; Mohammadi Ghaleni & Ebrahimi 2015). However, our study reveals only a weak linear correlation (R = 0.15) between precipitation, river runoff, and groundwater levels – a finding attributed to the consistently shallow groundwater depth in our area of study, which remains below 5 m, alongside an average annual rainfall of a mere 231 mm. Additionally, the intensity of individual rainfall events is predominantly less than 10 mm, which is insufficient to directly replenish the groundwater supply (Berghuijs et al. 2024). The infiltration of precipitation is influenced by various factors, including the depth of groundwater level (DGL) and aquifer lithology. In arid inland regions, where the DGL typically exceeds 6 m, the contribution of precipitation to groundwater recharge is notably diminished (Ma et al. 2013; Liu et al. 2018b; Guo et al. 2021). Moreover, evapotranspiration acts as a significant pathway for groundwater discharge, further complicating the balance of the water table in regions with extensive groundwater depths (Fan et al. 2008). While irrigated agriculture in arid areas is less dependent on precipitation for crop growth compared to rainfed agriculture (Sohoulande Djebou et al. 2021), extreme weather conditions with minimal precipitation can still deplete soil moisture, necessitating additional irrigation to counteract the adverse effects of water scarcity on crops. This necessity is particularly evident in areas under strict groundwater management, where the scarcity of precipitation, dry crops, and declining groundwater levels are common (Cao et al. 2021). Groundwater dynamics are also influenced by surface runoff and mountain reservoir regulation and storage (Liu et al. 2020). The southern mountains of the CHJP are neoproterozoic mudstones, so lateral recharge from the mountains is only subsurface flow from the valley. Although the impact of climate change on groundwater levels was considered, this study did not fully assess the potential effects of extreme climate events, such as droughts and floods, on groundwater resources. Subsequent research could focus on analyzing changes in groundwater levels under extreme weather conditions.

Human activities have a profound impact on the fluctuations of groundwater levels, with agricultural, domestic, and industrial water use being the primary pathways for groundwater depletion, particularly in arid inland regions (Akbari et al. 2020). In irrigated agricultural zones, human activities are especially influential, a consensus supported by this and previous studies (Ma et al. 2015; Xia et al. 2019; Wang et al. 2021, 2022a). The research findings underscore the significant effect of LULC on groundwater extraction (Bao et al. 2019; Wilopo et al. 2021). In the case of the CJHP, cropland, which accounts for the largest share of the area, expanded from 40 to 52% of the total area over the study period. This expansion has led to a surge in agricultural water use, exacerbating groundwater extraction in regions where surface water distribution is already uneven (Ma et al. 2015). The 21st-century has witnessed rapid population growth and agricultural expansion, prompting improvements in water transmission and distribution in irrigation areas, alongside advancements in efficient drip irrigation technology (Chen et al. 2018; Wang et al. 2022b). However, the reliance on high-quality groundwater has increased as a substitute for surface water with high sediment and salt content, which can clog irrigation systems (Han et al. 2019; Zhou et al. 2019). Consequently, agriculture, which constitutes over 90% of total water use, has seen a continuous rise in groundwater extraction, contributing to the decline in groundwater levels.

The dynamics of groundwater levels, such as changes in DGL, reflect the interplay of recharge, runoff, and discharge. Decreased recharge and increased discharge can lead to a decline in DGL (Hayashi et al. 2009). The development of water-saving irrigation practices has significantly reduced irrigation infiltration and canal leakage, altering the regional groundwater recharge dynamics (Ma et al. 2015). It was shown by Pang et al. (2009) that the constructed embankment leads to a reduction in groundwater recharge. The construction of infrastructure, such as canals and embankments, and the establishment of mountain reservoirs in the CJHP have further impacted groundwater recharge by reducing water availability in the plain areas (Liu et al. 2018a; Porhemmat et al. 2018).

Although this study attempted to assess the impact of human activities on groundwater levels, it did not provide an exhaustive analysis of all possible human activities, such as changes in agricultural irrigation techniques and the acceleration of urbanization processes. Future research can further investigate these aspects.

The agricultural and domestic water needs of the Changji Plain are largely dependent on groundwater resources. The sporadic distribution of river runoff and precipitation volume necessitates significant groundwater extraction to fulfill irrigation and residential demands, leading to a swift decline in groundwater levels. Our research indicates that between 2000 and 2020, the fluctuations in groundwater levels within our study area were influenced by both human activities and natural factors, with human actions being the most impactful. The region is predominantly allocated to arable land, which coincidentally are the areas where groundwater levels are the lowest. With agricultural water usage constituting over 90% of total water consumption, the critical imbalance between irrigation requirements and the capacity of water resources is the central cause of the falling groundwater levels. Post-2014, stringent implementation of water resource management policies has effectively mitigated the rate of decline in groundwater levels.

In response to the challenges faced in managing groundwater resources, future research and management strategies should concentrate on the following pivotal areas: First, enhancing inter-regional water transfers and reconfiguring the agricultural sector to ease the tension between the supply and demand of groundwater is essential. Second, we must improve and strengthen groundwater management policies to ensure the sustainable use of water resources over time. Third, it is imperative to investigate adaptive management strategies that ensure a reliable water supply during droughts and extreme weather events, by adjusting groundwater extraction practices to equitably satisfy agricultural requirements. Finally, a thorough review of existing policies is warranted to boost their resilience in extreme climatic conditions, making certain they can counteract the decline in groundwater levels without neglecting the regular needs of agricultural irrigation. By implementing these integrated measures, we can more effectively manage groundwater resources, guarantee their long-term sustainability, and bolster agricultural development.

This research was funded by the Natural Science Foundation of Xinjiang (2021D01A99). The author would like to thank the Water Resources Center of Changji Water Resources Bureau for providing water resources statistics.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.