Monitoring the Impacts of Human Activities on Groundwater Storage Changes Using an Integrated Approach of Remote Sensing and Google Earth Engine

Abstract

Groundwater storage refers to the water stored in the pore spaces of underground aquifers, which has been increasingly affected by both climate change and anthropogenic activities in recent decades. Therefore, monitoring their changes and the factors that affect it is of great importance. Although the influence of natural factors on groundwater is well-recognized, the impact of human activities, despite being a major contributor to its change, has been less explored due to the challenges in measuring such effects. To address this gap, our study employed an integrated approach using remote sensing and the Google Earth Engine (GEE) cloud-free platform to analyze the effects of various anthropogenic factors such as built-up areas, cropland, and surface water on groundwater storage in the Lake Urmia Basin (LUB), Iran. Key anthropogenic variables and groundwater data were pre-processed and analyzed in GEE for the period from 2000 to 2022. The processes linking these variables to groundwater storage were considered. Built-up area expansion often increases groundwater extraction and reduces recharge due to impervious surfaces. Cropland growth raises irrigation demand, especially in semi-arid areas like the LUB, leading to higher groundwater use. In contrast, surface water bodies can supplement water supply or enhance recharge. The results were then exported to XLSTAT software2019, and statistical analysis was conducted using the Mann–Kendall (MK) non-parametric trend test on the variables to investigate their potential relationships with groundwater storage. In this study, groundwater storage refers to variations in groundwater storage anomalies, estimated using outputs from the Global Land Data Assimilation System (GLDAS) model. Specifically, these anomalies are derived as the residual component of the terrestrial water budget, after accounting for soil moisture, snow water equivalent, and canopy water storage. The results revealed a strong negative correlation between built-up areas and groundwater storage, with a correlation coefficient of −1.00. Similarly, a notable negative correlation was found between the cropland area and groundwater storage (correlation coefficient: −0.85). Conversely, surface water availability showed a strong positive correlation with groundwater storage, with a correlation coefficient of 0.87, highlighting the direct impact of surface water reduction on groundwater storage. Furthermore, our findings demonstrated a reduction of 168.21 mm (millimeters) in groundwater storage from 2003 to 2022. GLDAS represents storage components, including groundwater storage, in units of water depth (mm) over each grid cell, employing a unit-area, mass balance approach. Although storage is conceptually a volumetric quantity, expressing it as depth allows for spatial comparison and enables conversion to volume by multiplying by the corresponding surface area.

1. Introduction

Groundwater is found almost everywhere beneath the land surface and plays a crucial role in the Earth’s hydrologic cycle, which involves the continuous movement of water across the planet. Its widespread availability makes it a key source of water supply worldwide [1,2]. As the Earth’s largest reservoir of fresh, liquid water, groundwater accounts for over 20% of global water use and 43% of irrigation water [3]. It provides essential freshwater to more than 2 billion people globally. Groundwater is also linked to the energy cycle, serving as both a thermal energy storage system and an energy consumer when extracted, linking it to global food security, climate resilience, and energy stability [4]. Globally, groundwater is a vital resource that is increasingly affected by natural supply variations and growing human demand [5]. In the most recent decades, socioeconomic development and population growth have led to unprecedented impacts on groundwater from human activities. These activities not only cause various levels of pollution through the direct discharge of contaminants but also alter the environmental conditions of groundwater, such as pH, redox state, and microbiological composition [6,7]. The growing demand for food has placed additional pressure on agriculture, increasing groundwater use for irrigation [8]. While surface water was traditionally used for irrigation, the development of large-scale sprinkler systems has driven a greater reliance on groundwater, as it is often more accessible and adaptable to irregular land surfaces without the need for canal systems and in many areas surface water is scarce and already allocated [9,10]. Therefore, it is important to implement efficient planning and management strategies to use it optimally and preserve it for the future [11,12].

Groundwater storage and movement are often poorly monitored because of the difficulty in collecting data on water located below the surface. Our understanding of groundwater usage and availability is typically based on limited or sparse in situ measurements, which are difficult to obtain on a global scale, especially in regions with restricted access [13]. Monitoring techniques, such as groundwater well measurements, are further complicated by diverse geology and the resulting variations in soil type, aquifer structure, and yield. In areas lacking data or facing conflict, information on groundwater depletion is often unavailable [14,15,16]. In recent decades, remote sensing methods for monitoring groundwater or aquifer changes have offered several advantages over traditional techniques. The application of satellite-based monitoring in aquifer studies marks a significant advancement because of its high resolution, low cost, and optimal spatial and temporal coverage [17]. In the field of remote sensing technology, numerous cloud-based platforms, such as Google Earth Engine (GEE), Sentinel Hub, Open Data Cube (ODC), the System for Earth Observation Data Access, Processing, and Analysis for Land Monitoring (SEPAL), openEO, JEODPP, and pipsCloud, have emerged to facilitate the modeling and monitoring of Earth’s features [18,19]. Among these, GEE is one of the most widely used platforms for monitoring Earth’s features [20]. It integrates a multi-petabyte catalog of satellite imagery and geospatial datasets with planetary-scale analysis capabilities. GEE provides free access to various datasets relevant to water management, climate change monitoring, and the modeling of natural and environmental risks [21,22].

For groundwater modeling, GEE offers several key datasets, including the Gravity Recovery and Climate Experiment (GRACE), GRACE Follow-On (GRACE-FO), and the Global Land Data Assimilation System (GLDAS), which are instrumental in monitoring groundwater storage [23,24]. GLDAS consists of three components: GLDAS-2.0, GLDAS-2.1, and GLDAS-2.2. GLDAS-2.0 is driven entirely by the Princeton meteorological forcing data and provides a temporally consistent series from 1948 to 2014 [25]. The daily resolved GLDAS V2.2 dataset begins on 1 February 2003 extends to the present and has a spatial resolution of 0.25° × 0.25° (depending on latitude ≈ 27.83 km × 27.83 km). It includes 26 variables, such as terrestrial water storage, groundwater storage, canopy water, snow depth equivalent, evapotranspiration, rainfall rate, and snowfall rate [26,27]. Additionally, GEE provides free access to anthropogenic variables, such as built-up areas, land use/cover, and consumed surface water resources, which provide an easy and comprehensive analysis of these variables with groundwater storage on a long-term and large-scale [28].

In Iran, groundwater serves as a vital water resource, particularly in arid and semi-arid regions such as the Lake Urmia Basin (LUB). Over the past two decades, the region has experienced increasing pressure on its groundwater resources. Data from the Natural Resources and Watershed Management Organization of Iran indicate that the number of semi-deep wells (typically 30–70 m in depth) rose from 59,161 in 2002 to 95,811 in 2015. During the same period, the number of deep wells (greater than 70 m) increased from 7057 to 7457. This expansion of groundwater extraction infrastructure underscores a growing reliance on subsurface water, particularly for agricultural irrigation and domestic use. These developments, coupled with the reduction in surface water bodies such as Lake Urmia, highlight the urgent need for a systematic assessment of the anthropogenic impacts on groundwater storage in the region (https://en.frw.ir/, accessed on 15 September 2024).

According to previous research, studies have employed Synthetic Aperture Radar (SAR) remote sensing data [29,30,31,32,33,34], GLDAS and GRACE datasets [24,25,26,35,36,37,38], and Geographic Information Systems (GIS) [39,40,41,42] for groundwater storage modeling. Although several studies have explored the impact of climate change on groundwater storage [43,44,45], the contributions of anthropogenic impacts have rarely been explored. Traditional methods have been used to monitor groundwater changes without analyzing the potential impact of external factors such as human activities. However, this study presents an integrated approach using remote sensing and GEE to monitor groundwater storage patterns and assess the effects of anthropogenic factors. Few studies have evaluated the impact of anthropogenic factors on groundwater changes, and those that have typically focused on a limited number of variables such as urban development. In contrast, this study incorporates several key variables, including built-up areas, croplands, and surface water resources to comprehensively investigate their relationship with groundwater storage. The methodology applied in this study offers a simple and efficient alternative to traditional methods for monitoring groundwater patterns and the potential effects of human activities. Furthermore, while most previous research has focused on the effects of anthropogenic factors on groundwater quality [46,47,48], this study uses GLDAS data to assess changes in groundwater storage on a regional scale. In the present study, groundwater storage refers to changes in groundwater storage anomaly, estimated using GLDAS model outputs. Specifically, groundwater storage is calculated on a global scale using regular grid cells as the residual component of the terrestrial water budget after accounting for soil moisture, snow water equivalent, and canopy water storage. While the resulting storage values are expressed as anomalies (in millimeters), they represent relative changes in groundwater over time rather than absolute volumes. Statistical analysis based on the Mann–Kendall (MK) non-parametric trend test was applied to investigate the potential relationships between anthropogenic factors and groundwater storage.

2. Study Area

The LUB is a crucial geographical and ecological area located in northwestern Iran, covering an area of 51,876 km² (Figure 1). It is a closed basin and a total of 60 rivers feed into it, with 21 being permanent or seasonal and 39 classified as temporary. The primary inflows to Lake Urmia include the Zarineh, Simineh, and Aji Chai rivers. Within the basin, contributions come from the Zarineh river (14%), Simineh river (11%), Godar (8%), Barandoz (6%), Shahrchai (2%), and Nazlu Chai (6%), along with seven seasonal rivers, 39 intermittent streams, internal springs, and precipitation from rain and snow [49]. Groundwater extraction via wells has seen a marked increase in both volume and number over the past few decades [50]. This trend, especially during the severe drought between 1999 and 2001, coupled with human water usage, led to the water table dropping by 2003 [51]. From 2003 to 2014, the lake’s surface area was nearly halved, and water levels dropped by an additional 3 m, with greater seasonal fluctuations in the lake’s extent [52]. Post-2015, the lake’s size and storage began to recover to be reached around 1000 to 1200 km², thanks to higher precipitation in 2015 and 2016 [53].

3. Material and Methods

3.1. Materials

To examine the effects of anthropogenic factors on groundwater storage, various human-related variables such as built-up areas, croplands, surface water resources, and groundwater storage data were analyzed from 2000 to 2022. All analyses were conducted within the GEE environment. To validate the accuracy of the employed products, sample points and relevant information were collected from the Ministry of Roads and Urban Development (https://www.mrud.ir, accessed on 20 September 2024) for built-up areas, Ministry of Agriculture Jihad (https://www.maj.ir, accessed on 20 September 2024) for cropland, and Water and Sewage Organization (https://www.nww.ir, accessed on 20 September 2024) for surface water resources and groundwater storage. Figure 2 presents the variables used for several years. Further details regarding the employed anthropogenic variables can be found in the following subsections.

3.1.1. Built-Up Area

This study used the JRC/GHSL/P2023A/GHS_BUILT_S product [54] to assess the impact of built-up area extension on groundwater (Figure 2). Symbolic Machine Learning is a novel supervised classification framework, which is used for producing GHSL layers. The use of this approach allowed for a gradual refinement of thematic information, from the initial MODIS binary data at 500 m resolution available in 2010, to the current Sentinel-2-derived continuous prediction of built surfaces at 10 m resolution. This raster dataset depicts the distribution of built-up surfaces measured in square meters per 100 m grid cell. The dataset measures: (a) the total built-up surface, and (b) the built-up surface allocated to grid cells predominantly used for non-residential (NRES) purposes. The data are spatially and temporally interpolated or extrapolated from 1975 to 2030, in 5-year intervals [54].

3.1.2. Cropland

This study utilized the MODIS/061/MCD12Q1 dataset (available from 2001 to present) to evaluate the influence of land cover, particularly croplands, on groundwater, with croplands represented by Class 12 (Figure 2). The dataset offers annual global land cover data, derived from the supervised classification of MODIS Terra and Aqua satellite reflectance. Land cover types are categorized based on the classification schemes of the International Geosphere-Biosphere Programme (IGBP), the University of Maryland (UMD), the Leaf Area Index (LAI), BIOME-Biogeochemical Cycles (BGC), and Plant Functional Types (PFT) (https://doi.org/10.5067/MODIS/MCD12Q1.061, accessed on 25 June 2025). For the year 2000, we applied a supervised classification based on the Maximum Likelihood method using MODIS data for land use/cover classification.

3.1.3. Surface Water

Surface water patterns play a crucial role in influencing changes in groundwater storage. In this study, the JRC Yearly Water Classification History, version 1.4, was utilized (Figure 2). Expert systems were employed for the yearly water classification. They are non-parametric classifiers that incorporate image interpretation expertise into the classification process and can be used with multiple data sources. This system assigns each pixel to one of three classes, either water, land or non-valid observations. These data were derived from 4,716,475 scenes captured by Landsat satellites 5, 7, and 8, spanning from 16 March 1984 to 31 December 2022. Each pixel was classified as either water or non-water using an expert system, with the results compiled into a monthly history covering the entire period, as well as two specific epochs (1984–1999 and 2000–2021) for change detection analysis. This mapping product consists of a single image with seven bands, capturing various aspects of the spatial and temporal distribution of surface water over the past 38 years. Areas where water was never detected are masked out [55].

3.1.4. Groundwater

To assess the impact of anthropogenic variables on groundwater storage, this study used data from NASA’s Global Land Data Assimilation System (GLDAS) Version 2, specifically the NASA/GLDAS/V022/CLSM/G025/DA1D dataset (Figure 2). GLDAS-2.1 incorporates both model outputs and observational data from 2000 onward, while GLDAS-2.2 (available here: https://disc.gsfc.nasa.gov/datasets/GLDAS_CLSM025_DA1_D_2.2/summary, accessed on 25 June 2025) employs data assimilation (DA), in contrast to the “open-loop” systems of GLDAS-2.0 and GLDAS-2.1, which do not use data assimilation [25]. Since groundwater data from the GLDAS-2.2 product is only available from 2003, we used data collected by the Water and Sewage Organization (https://www.nww.ir, accessed on 10 October 2024) to fill in the missing data from 2000.

3.2. Methodology

Three main steps were involved in analyzing the effects of anthropogenic factors on groundwater storage. First, anthropogenic variables, including built-up areas, croplands, and surface water resources, were retrieved and pre-processed using the cloud-free GEE platform. Second, the time series of these variables were monitored and exported to the XLSTAT environment. Finally, statistical analysis based on the MK non-parametric trend test was conducted to investigate potential relationships between anthropogenic factors and groundwater storage. Figure 3 provides an overview of the methodology applied for assessing the effects of anthropogenic variables on groundwater storage.

In GEE, the two primary geographic data structures are Images and Features, corresponding to raster and vector data types, respectively. Images consist of multiple bands, which represent different spectral or temporal layers, and are associated with a dictionary of properties that describe metadata and attributes. In contrast, features are composed of geometries such as points, lines, or polygons accompanied by a dictionary of properties that includes information about the spatial characteristics and attributes of the features (https://courses.spatialthoughts.com/end-to-end-gee.html, accessed on 30 October 2024). Since our analysis focuses on raster datasets, we utilized the Image data type to efficiently process large-scale temporal changes. The region of interest (LUB) was defined for the study and represented by geometries such as points, lines, polygons, or a collection of these shapes, limiting the analysis to pixels within the polygon representing LUB, which was prepared using ArcGIS 10.8. Covering the period from 2000 to 2022, we extracted variables such as built-up areas, croplands, surface water, and groundwater storage from multiple satellite products available in GEE. The data export was performed in formats such as GeoTIFF, CSV, or directly to Google Drive, providing flexibility for use in various GIS and statistical software. The accuracy of the products was then evaluated using field-based data. ArcGIS 10.8 was employed to map and estimate changes in variable areas during the 2000–2022 period, as shown in Figure 2, providing a spatial perspective on how these factors evolved over time. Additionally, Microsoft Excel was used to create detailed graphs and calculate correlation coefficients between the different variables and groundwater storage, allowing for a quantitative assessment of relationships and trends across the study period.

Statistical Analysis

The MK non-parametric trend test is a widely recognized statistical method employed to detect monotonic trends either increasing or decreasing in time series data, without the need to assume a specific data distribution. In the present study, this method is utilized to assess whether the observed temporal changes are statistically significant. The MK test assesses trends by evaluating the sign of the difference between all possible pairs of data points within a time series. For a time series of length $n$, each data point is compared with all subsequent data points, resulting in a total of $n(n–1)/2$ pairwise comparisons. The test statistic $S$, as defined in Equation (1), is the cumulative sum of the sign function $sgn(x_j-x_i)$ over all pairs $(i, j)$ where $i<j$. The sign function, presented in Equation (2), yields a value of +1, 0, or −1 depending on whether the difference between the two values is positive, zero, or negative, respectively [56].

$$S={\sum }_{i=1}^{n-1}{\sum }_{j=i+1}^{n}sgn(x_j-x_i),$$

(1)

$$sgn\left(x_j-x_i\right)=\left\{\begin{array}{c}+1\\ 0\\ -1\end{array} \begin{array}{c}{if x}_j>x_i\\ {if x}_i=x_j\\ {if x}_j<x_i\end{array}\right.$$

(2)

The value of the test statistic $S$ reflects the overall direction and strength of a trend: a large positive or negative $S$ indicates a strong increasing or decreasing trend, respectively. Under the null hypothesis of no trend and assuming that the data are independent and randomly ordered (a key limitation) the statistic $S$ approximates a normal distribution with a mean of zero when the sample size exceeds 10 (n > 10). This approximation enables the application of standard statistical techniques to assess the significance of the observed trend [56].

The MK test has broad applicability across environmental and scientific disciplines. It is widely used in climate and hydrological research to detect trends in variables such as temperature, precipitation, streamflow, and groundwater levels [57]. Additionally, it has been employed in air quality monitoring, glaciological studies, agricultural evaluations, and public health time series analyses [58,59]. Nevertheless, the MK test is not without limitations. It assumes that data points are serially independent. When time series data exhibit autocorrelation, a common feature in environmental datasets, the test may yield spurious trend detections. Moreover, the MK test is designed to identify monotonic trends and may fail to capture cyclical, periodic, or non-linear variations over time. Consequently, results should be interpreted with an understanding of these underlying assumptions and, where appropriate, supported by complementary analytical methods. It is important to recognize that the MK test is inherently univariate, meaning it does not account for multivariate interactions or time-lagged effects among variables. As a result, it is unable to capture complex interdependencies or delayed responses that may exist, for example, between groundwater dynamics and anthropogenic pressures [60].

4. Results

This study utilized several anthropogenic variables, including built-up areas, croplands, and surface water, to assess their relationship with groundwater storage in the LUB from 2000 to 2022. The findings of this study highlight the significant impact of anthropogenic factors on the reduction in groundwater storage during the study period.

Table 1. Correlation coefficients between various anthropogenic factors and groundwater storage.

	Groundwater Storage (mm)	Cropland Area (m2)	Built-up Areas (m2)	Surface Water Resources (km2)
Groundwater Storage (mm)	1	−0.85	−1.00	0.87
Cropland Area (m2)	−0.85	1	−0.80	−0.89
Built-up Areas (m2)	−1.00	−0.80	1	−0.99
Surface Water Resources (km2)	0.87	−0.89	−0.99	1

and Figure 4 present the results in greater detail. As shown in Table 1, there is a strong negative correlation between built-up areas and groundwater storage, with a correlation coefficient of −1.00. This significant relationship indicates that urban expansion and development have led to increased groundwater consumption, particularly in large cities such as Tabriz, Urmia, and Maragheh. The growth of built-up areas has been a major factor in depleting groundwater resources, especially in urban areas where water demand has surged in tandem with population growth and industrial activity. Additionally, the study found a notable negative correlation between cropland area and groundwater storage, with a correlation coefficient of −0.85, as shown in Table 1. The agricultural sector, the largest consumer of water resources in the LUB, plays a critical role in groundwater depletion. Over the study period, the expansion of agricultural land has significantly increased water demand, further exacerbating groundwater reduction. As agricultural irrigation is heavily dependent on groundwater in this region, the pressure on aquifers has intensified, especially in times of water scarcity. The expansion of agricultural land in the LUB has substantially increased water demand, thereby intensifying groundwater depletion. The region receives an average annual precipitation of approximately 350 mm, though this value has fluctuated significantly in recent decades from peaks near 465 mm in the mid-1990s to lows below 160 mm during severe drought years, such as 2017. Climate projections under high-emission scenarios (SSP5–8.5) indicate a potential decline in mean annual rainfall to around 363 mm by the end of the 21st century. Agricultural irrigation in the basin is predominantly groundwater-dependent, particularly during dry years, placing additional pressure on already-stressed aquifers. Cropping patterns have shifted toward more water-intensive species, notably alfalfa (~7600 m³/ha/year) and sugar beet (~7100 m³/ha/year), which are primarily cultivated for domestic livestock feed and local sugar production, respectively. These crops are especially prevalent in areas such as Miandoab, a major agricultural sub-region. With agricultural withdrawals comprising the majority of water use in the basin, and naturalized runoff estimated at approximately 6.5 billion m³ per year, unsustainable irrigation practices continue to diminish inflows to Lake Urmia and exacerbate groundwater stress [61,62]. Furthermore, Table 1 also reveals a strong positive correlation between surface water availability and groundwater storage, with a correlation coefficient of 0.87. Since 2000, surface water resources have decreased significantly in the region, as indicated by [28]. This reduction in surface water has directly affected groundwater consumption, as decreasing surface water availability has forced a greater reliance on groundwater sources to meet agricultural, industrial, and domestic needs [63]. This suggests that the depletion of surface water resources has been a key driver of groundwater over-extraction across the LUB. In conclusion, the results of this study clearly demonstrate that anthropogenic factors specifically urbanization, agricultural expansion, and the decline in surface water availability have collectively contributed to the significant reduction in groundwater storage in the LUB.

Human activities have substantially affected groundwater storage in the study area over recent decades, as illustrated in Table 1 and Figure 4. Using groundwater data derived from the GLDAS, storage values are reported as depth equivalents in millimeters (mm), representing the thickness of water stored in the soil and subsurface per unit area (i.e., millimeters of water equivalent across the LUB). This unit of measurement enables consistent spatial and temporal comparisons. A significant decline commenced in 2008: prior to this year, the average groundwater storage was 670.54 mm, which dropped to 598.68 mm in the subsequent years. Our analysis reveals that variables such as cropland and urban expansion did not exhibit significant trends or fluctuations between 2000 and 2022, as depicted in Figure 4. In contrast, surface water resources showed a statistically significant decline of 2.2% per decade (p < 0.05), with a sharp decrease visible in Figure 4. Before this change, the average surface water area was 9.0%, which then decreased to 6.8%. This detailed analysis highlights the varied impacts of human-induced factors on groundwater storage, emphasizing the need for integrated management strategies moving forward.

5. Discussion

This research tracks the impact of human activities on groundwater storage in the LUB from 2000 to 2022. The findings reveal strong associations between various human-induced factors and the decline in groundwater storage. While previous studies have focused on the impact of climate change on groundwater storage [43,44,45], only a few studies have concentrated on assessing anthropogenic factors, which are among the most critical drivers of groundwater depletion. Previous studies have mostly applied traditional methods for monitoring groundwater patterns. This study provides a more comprehensive understanding of the influence of human actions on groundwater storage change. The applied methodology is of great importance for groundwater monitoring and analyzing the effects of external contributors such as human on groundwater change since it offers an easy and cost-effective way compared to the traditional methods. The results of our research highlight the significant impact of human activities, particularly in built-up areas, croplands, and surface water, on groundwater storage in the LUB between 2000 and 2022. Our findings also show a significant reduction in groundwater storage from 2000 to 2022. Since Iran is a semi-arid and arid region, groundwater serves as an essential source of freshwater for human consumption and industrial use. The increased density of both semi-deep and deep wells in the LUB over the past two decades serves as a clear indicator of the growing reliance on groundwater resources. As previously discussed, (see Introduction), this intensification of groundwater abstraction has coincided with a significant decline in groundwater storage, particularly after 2008. The substantial rise in the number of semi-deep wells typically ranging from 30 to 70 m in depth suggests that shallow aquifers are experiencing increasing stress. Simultaneously, the gradual increase in deep wells (exceeding 70 m) indicates a shift toward the exploitation of deeper groundwater reserves.

Groundwater storage in this study was estimated using the NASA GLDAS Version 2.2 dataset in GEE. It is important to note that GLDAS does not directly observe groundwater; rather, it infers groundwater storage through land surface modeling, which introduces model-based uncertainties. These uncertainties originate from errors in input data, simplifications in the parameterization of hydrological processes, and the lack of direct assimilation of groundwater observations. Previous studies (e.g., [23]) have demonstrated that groundwater estimates derived from GLDAS can be useful for identifying general trends, though their accuracy may be limited in regions characterized by complex hydrogeological conditions or sparse ground-based observations. It is also essential to account for the influence of precipitation modeling on the estimation of groundwater storage values. As GLDAS relies on atmospheric forcing data particularly precipitation as a primary input, any biases or inaccuracies in precipitation estimates can propagate through the modeling framework and affect the inferred groundwater storage. These uncertainties are especially pronounced in regions characterized by high precipitation variability or limited availability of ground-based rainfall observations, where reliance on satellite-derived or reanalysis products each with inherent limitations is necessary. Therefore, the accuracy of groundwater storage estimates derived from GLDAS is, to a significant extent, dependent on the quality and reliability of the precipitation data incorporated into the model. Compared to the GRACE satellite mission which measures changes in total terrestrial water storage through variations in Earth’s gravity field, GLDAS provides higher spatial and temporal resolution. However, while GRACE offers direct observational constraints, GLDAS relies on modeled assumptions. In regions where in situ groundwater measurements are available, comparisons with both GLDAS and GRACE estimates have shown general agreement, although discrepancies may arise due to differences in spatial resolution, data processing methodologies, and underlying assumptions [64].

The analysis of anthropogenic factors indicates a significant increase in built-up areas within the LUB region between 2000 and 2022, as shown in Figure 2. This urban expansion corresponds with population growth, which, in turn, has driven higher water consumption demands. To exemplify, the population of Tabriz, the largest city in the LUB, increased from 1,264,000 in 2000 to 1,644,000 in 2022. A similar trend is observed in Urmia, where the population grew from 435,200 in 1996 to 736,224 in 2016. Another major city in the region, Maragheh, had a population of 146,405 in 2006, which increased to 175,255 by 2016 [65]. Our findings also show a marked increase in cropland coverage across the LUB during this period. The amount of water used for crop cultivation is influenced by various factors, particularly the type of irrigation technology employed. Since traditional irrigation methods dominate in the LUB, there is substantial reliance on groundwater and surface water resources. Traditional irrigation methods such as surface irrigation, flood irrigation, and furrow irrigation often rely on open canals and unlined ditches. These systems result in significant water loss due to evaporation, seepage, and over-application of water exceeding crop needs. Moreover, they typically lack integration with modern water-saving technologies. To maintain crop yields, especially during droughts, farmers in the LUB region tend to pump more groundwater. This practice has led to noticeable groundwater depletion, particularly in agricultural areas like the Maragheh and Urmia plains. Regarding surface water, our study highlights a concerning trend of water resource depletion in the region from 2000 to 2022, with the regular drying of Lake Urmia being one of the most obvious examples in the LUB (See Appendix A) [66]. Notably, the agricultural sector emerges as the largest consumer of water resources in the LUB. In terms of groundwater, the months of April and May are characterized by high groundwater storage. Specifically, in 2004, April and May recorded groundwater storage of 726.67 mm and 735.19, respectively, the highest levels recorded during the 2000–2022 period. However, by 2022, these values had dropped to 550.93 mm and 548.35 mm for the same months, respectively. According to groundwater data, October and November experienced significant reductions in groundwater storage compared to other months throughout the study period. The groundwater storage for October and November peaked in 2003 at 575.30 mm and 577.43 mm, respectively. By 2022, groundwater storage in these months had decreased to 511.49 mm and 514.00 mm in the LUB as can be seen in Figure 5 (See Appendix A).

While this study focuses on quantifying anthropogenic contributions to groundwater storage variations, we recognize that natural factors such as precipitation variability, evapotranspiration rates, and geological heterogeneity also play critical roles in shaping groundwater dynamics. Temporal shifts in rainfall patterns and the intensity of droughts influence recharge rates, while evapotranspiration contributes to water loss. In the study area, evapotranspiration increased between 2000 and 2021, whereas annual total precipitation declined over the same period [66]. However, it is important to note that the groundwater changes documented in the present study are not primarily driven by the effects of severe drought or ongoing climate change. This distinction helps readers understand that these factors are separate and less significant compared to the impacts of land use change and increased irrigation. Additionally, subsurface geological structures significantly affect aquifer storage and flow capacity, introducing spatial heterogeneity not fully captured in the current dataset. Specifically, the Lake Urmia Basin contains both alluvial and fractured rock aquifers, which vary in their storage capacities and recharge potentials, further contributing to spatial differences in aquifer responses.

As with all scientific studies, this research has several limitations. First, the accuracy of the remote sensing products employed particularly those derived from the GLDAS and land cover datasets was not independently validated within the study area. Although these datasets are widely utilized, regional validation (e.g., through in situ measurements or comparisons with national groundwater monitoring records) remains essential, particularly in regions such as the LUB, where local hydrological dynamics may deviate from the assumptions underlying global models. Second, the groundwater storage data utilized in this study possesses a relatively coarse spatial resolution (approximately 0.25°, or ~25 km), which constrains their ability to capture local-scale variations. For example, in regions characterized by high well density or rapid land use change, small-scale aquifer dynamics may go undetected. This limitation reduces the practical applicability of the findings for fine-scale groundwater management tasks, such as permit allocation or the design of local recharge initiatives. Future research should consider integrating additional variables, including soil texture, groundwater abstraction volumes, irrigation practices, and aquifer geology, to produce more nuanced insights. Furthermore, policies that influence groundwater use such as well licensing regulations and irrigation subsidies should be incorporated into comprehensive, integrated assessments.

6. Conclusions

The present study employed an integrated approach using remote sensing and the cloud-free GEE platform to monitor groundwater change and analyze the effects of anthropogenic variables, including built-up areas, cropland, and surface water, on it. Our findings indicate that urban and cropland expansion have the greatest influence on groundwater changes across the study area from 2000 to 2022. The results highlight the high efficiency of the cloud-free GEE platform in monitoring groundwater storage patterns and the anthropogenic variables affecting its changes on a large scale. Additionally, our findings emphasize the effectiveness of combining GEE with remote sensing data to simulate and monitor anthropogenic impacts on groundwater storage in a fast and cost-effective manner. This methodology proves particularly efficient in semi-arid and arid regions that suffer from water scarcity and rely on groundwater storage for effective planning and management. Overall, the results of this study offer valuable insights for researchers working in the fields of water and energy management.