Investigating the coherence of groundwater quality of a coastal aquifer and the hydroclimatic factors via remote sensing and wavelet analysis

GWQ

groundwater quality

TWS

total water storage

GRACE-FO

Gravity Recovery and Climate Experiment Follow-On

NDVI

normalized difference vegetation index

WC

wavelet coherence

PWC

partial wavelet coherence

MWC

multi wavelet coherence

ENSO

El Niño Southern Oscillation

NAO

North Atlantic Oscillation

PDO

Pacific Decadal Oscillation

AMO

Atlantic Multidecadal Oscillation

SOI

Southern Oscillation Index

Assessing groundwater quality (GWQ) is of paramount importance since groundwater serves as a vital source of drinking water for a significant portion of the global population. As such, ensuring its quality is essential to safeguarding public health and preventing the spread of waterborne diseases. Moreover, groundwater plays a crucial role in sustaining ecosystems and supporting various ecological processes, including the health of surface water bodies. Understanding GWQ is crucial for preserving biodiversity, protecting sensitive habitats, and maintaining the overall ecological balance. GWQ assessment helps identify potential contaminants or pollutants that could impact crops, livestock, industrial processes, and overall economic activities. Furthermore, groundwater resources are often limited and sensitive to changes in usage and environmental conditions. Investigating GWQ facilitates proper management and sustainable utilization of this valuable resource, aiding in long-term water resource planning and conservation efforts. Therefore, it is essential to assess the impact of the different factors on GWQ.

Various factors have the potential to alter the GWQ. Climate change can have profound effects by influencing precipitation patterns, temperature fluctuations, and evaporation rates (Earman & Dettinger 2011; Jafarzadeh et al. 2021; Prayag et al. 2023). Rising temperatures can enhance evaporation and decrease soil moisture, thereby reducing groundwater recharge. Anthropogenic activities, such as industrial wastewater discharge, chemical-intensive agriculture, improper water usage, and pollution from mining activities, can also contribute to groundwater contamination and degradation. Thus, the interaction of climate change and the impact of anthropogenic activities on GWQ should be considered thoroughly, which may have been overlooked in prior studies.

Conversely, teleconnection parameters can impact rainfall patterns and groundwater availability in different regions worldwide. These indices, which represent large-scale atmospheric patterns, can influence climate conditions such as precipitation and temperature in distant regions. Teleconnection refers to the noteworthy association and interrelation between changes over time in two circulation systems or patterns that are geographically distant from each other. It serves as a means to summarize climate models and better comprehend the exchange of heat, energy, moisture, and momentum in the Earth's atmosphere, thereby gaining insights into the functioning of climate change and regional ecosystems (Ahmadi et al. 2019). During El Niño events, teleconnection indices like the Southern Oscillation Index (SOI) can lead to altered rainfall patterns, potentially influencing the quality of groundwater resources. Some studies investigated the influence of teleconnection pattern on groundwater (GW) (e.g. see, Najafi et al. 2022). To analyze and quantify the effects of climate oscillations on various hydroclimate variables in the Lake Urmia watershed, Rezaei & Gurdak (2020) employed wavelet coherence (WC) to investigate the impacts of the El Niño Southern Oscillation (ENSO), North Atlantic Oscillation (NAO), Pacific Decadal Oscillation (PDO), and Atlantic Multidecadal Oscillation (AMO) on different parameters. The results highlight the significant impact of ENSO, PDO, NAO, and AMO on the hydroclimate dynamics of the Lake Urmia watershed, providing valuable insights into the dominant drivers of variability in the water resources of the region and the associated environmental factors. By understanding and monitoring teleconnection indices, the potential impacts on GWQ can be identified and proactive measures will be obtained to manage and protect this vital resource. While studies have delved into the effects of GW and teleconnection indices, the influence of these indices on GWQ has been ignored and it is essential to consider the interaction of GWQ and teleconnection indices.

WC is a valuable tool used to investigate the relationship between different parameters and the time series data. It is especially valuable in situations where the method involves identifying potential links between two processes by examining the frequency ranges and temporal periods. WC combines the concepts of wavelet analysis and coherence analysis to provide insights into the time-varying coherence between two signals (Nourani et al. 2019). This technique helps in uncovering hidden patterns and coherency using traditional correlation analysis. There are some studies that used WC to assess hydroclimatic factors (e.g. see, Rezaei & Gurdak 2020; Nourani et al. 2021; Roushangar et al. 2023). Nevertheless, bivariate relationships can be deceptive due to the interdependence between the predictor variable and other factors, particularly in groundwater flow systems that encompass intricate recharge and discharge components. Therefore, the multi WC (MWC) and partial WC (PWC) were used in recent studies (Gu et al. 2022). PWC can identify localized and scale-specific bivariate relationships between predictor and response variables by eliminating the influence of other variables. MWC has been used to reveal the multivariate relationships between variables (Ng & Chan 2012). Utilizing PWC and MWC can serve as suitable tools for evaluating the coherence between GWQ and various factors, which is lacking in previous studies.

The objective of this study is to examine the factors influencing GWQ. To achieve this, various factors, including climate change indicators (temperature), anthropogenic activities, the normalized difference vegetation index (NDVI), lake level, lake area, total water storage (TWS) fluctuations, Gravity Recovery and Climate Experiment Follow-on (GRACE-FO), and teleconnection patterns (AMO, PDO, and SOI), are analyzed via WC to determine their coherence with GWQ. PWC is utilized to understand the individual effects of each factor, while MWC is employed to investigate the combined effects of multiple factors. To the authors' best knowledge, in the current paper, the first application of PWC and MWC for assessing the impact of various parameters and also considering the teleconnection indices on GWQ were presented in different zones of the aquifer using ensemble clustering. In addition, the coherence of the GWQ and GRACE-FO data, as a useful measure of water storage, was investigated in this study, which is neglected in the previous studies. Although, the WC analysis was employed in coherency assessment of groundwater level in previous studies, GWQ analysis has been overlooked, especially at the zonal scale.

The primary goals of the paper are as follows:

• i. Applying clustering ensemble for clustering the piezometers and zoning.

• ii. Identification of the trends of parameters via the Mann–Kendall test.

• iii. Coherence assessment of teleconnection indices and GWQ.

• iv. Assessment of the relationship between GRACE-FO and GWQ.

The contribution of the current article toward the advancement of specialized knowledge in this field is to provide valuable information regarding GWQ and various influencing factors, which is a crucial aspect in the groundwater management, especially in arid and semi-arid regions. This information offers decision-makers valuable insights for basin management and controlling the agricultural activities, water allocation, and other relevant issues.

The rest of the paper was structured as follows: Section 2 presents methods and materials, which include the proposed methodology and model setup, explanation of the case study, applied data, and employed methods; Section 3 presents the results and discussion, which consists of the obtained results of the trend test and WC analysis; and Section 4 presents conclusions. The detailed description of the sections is presented in the following sections and sub sections.

In this study, among 200 data items from 33 piezometers, specifically measuring GWQ in the aquifer, were utilized. The choice of these piezometers was determined by the presence of perfect data and their consistency in the analyzing period. The missing data of the time series are filled with the Modified Akima cubic Hermite interpolation method in the MATLAB software.

In this study, the GRACE-FO data are used for TWS fluctuation impact on GWQ consideration, and Landsat data served as a proxy for human activities in the modeling procedure. The data of GWQ and temperature from 2002 to 2020 were obtained from the regional water company of Iran.

In addition, teleconnection indices (AMO, PDO, and NAO) were employed and obtained from https://psl.noaa.gov/. The AMO represents fluctuations in sea surface temperatures (SST) across the North Atlantic Ocean spanning from 0° to 70° N, exhibiting multidecadal variability (Abiy et al. 2019). In general, AMO has 65–70 years cycle with a 0.4 °C range between extremes. It is considered that SST change in the Atlantic is a feedback of the transport of the top layer of warm and salty oceanic water from the equatorial to the North Atlantic region.

The SST in the Pacific Ocean from 20°N to poleward has an ENSO-like pattern with a significant impact on wind pattern in the Northern Pacific Ocean. The impact of changes in the PDO regime is associated with global scale, changes in temperature, precipitation, wind pattern, and speed. During a PDO warm phase, the central and western Pacific SST becomes less than normal and the eastern portion is greater than normal. During a PDO cold phase, the central and western Pacific SST becomes greater than normal and the eastern portion is less than normal.

The NAO is a prominent mode of variability with important impacts from the polar to subtropical Atlantic and surrounding landmasses. It is most pronounced during boreal winter. The NAO is associated with changes in wind strength and direction, heat and moisture transport, and the frequency and strength of storms. It also plays an important role in modulating ocean properties, such as SST, mixed layer depth, and, on long timescales, and basin-wide changes in circulation including modulation of North Atlantic overturning (Gupta & McNeil 2012).

The utilization of the GRACE dataset has brought about a revolutionary comprehension of Earth's water resources through the measurement of gravity field changes (Humphrey et al. 2023). This dataset offers valuable information regarding monthly variations in terrestrial water storage anomalies, which can be extracted from https://grace.jpl.nasa.gov. The study utilized the GRACE and GRACE-FO datasets acquired from the Jet Propulsion Laboratory (JPL) center. Initially launched in 2002, the GRACE satellite mission operated until its conclusion in 2017. The GRACE-FO mission was launched in May 2018. However, a notable temporal gap between the two missions was observed, alongside an increase in missing observations toward the conclusion of the GRACE mission. The substantial number of missing observations toward the mission's end and the 11-month observational hiatus (from June 2017 to May 2018) between missions restrict the optimal utilization of GRACE and GRACE-FO data (Huang et al. 2022; Gauer et al. 2023). Therefore, the modified Akima cubic Hermite Interpolation method was implemented to fill in the missing data. Prior to applying the GRACE and GRACE-FO datasets for geophysical analyses, preprocessing steps are necessary. Preprocessing was conducted in this study with solutions derived from Satellite Laser Ranging as outlined in Technical Note TN-14 (Loomis et al. 2020).

To evaluate the anthropogenic activities, the NDVI is considered as a reliable indicator employed to monitor changes in vegetation and the environment, extensively utilized for assessing vegetation dynamics across various scales. It is calculated using the red (R) and near-infrared (NIR) reflectance bands (NDVI = (NIR − R)/(NIR + R)) of Landsat images, for Landsat 7 and Landsat 8 (Guha et al. 2020; Ashok et al. 2021; Erasmi et al. 2021; Fokeng & Fogwe 2022). In addition, for calculation of Urmia Lake area, the normalized difference water index is calculated using Landsat data by taking the difference between the green and shortwave infrared bands, divided by their sum (Kshetri 2018). The Landsat data were obtained from: https://developers.google.com/earth-engine/datasets/catalog/landsat-7 and https://developers.google.com/earth-engine/datasets/catalog/landsat-8

Because there are many piezometers within the region, incorporating each piezometer individually in the analyses is a time-consuming task. To tackle the issue of redundant data, clustering methods can be utilized to select a single representative piezometer for every zone of the aquifer. The ensemble, Linkclue clustering method, was employed in this study to group the piezometers, due to its more accurate performance presented in the previous literature (e.g. see, Sharghi et al. 2022). Traditional clustering methods have demonstrated acceptable performance, but they may exhibit different levels of effectiveness when applied to the same clustering task. Therefore, employing an ensemble method can enhance the efficiency of the clustering. Linkclue clustering, developed by Iam-on & Garrett (2010), is a clustering ensemble method and the methodology of the Linkclue clustering is explained in the diagram depicted in Figure S4. In the proposed methodology, the initial process was conducted by creating the cluster ensemble ⁠, as shown in Figure S4. The objective of the Linkclue technique is using link-based similarity metrics, which enhances the efficacy of the traditional pairwise similarity method. In the Linkclue method, the similarity matrix is computed and it is utilized to obtain final clustering results. For more information about the Linkclue method, see Iam-on & Garrett (2010).

The mentioned technique, similar to the simple WC, ranges between 0 and 1.

In this case, a low PWC squared shown at where a high WC squared was found implies that the time series x₁ does not have significant influence on the time series y at that particular time–frequency space, and time series x₂ dominates the effect on the variance of y, and vice versa for the opposite case.

MWC is an extension from the PWC to the multivariate case. In this case, the correlation of the variables with each other is taken into account when calculating coherence and phase differences.

The Mann–Kendall test is a statistical test commonly used in time series analysis to assess the presence of trends or patterns in data. The test plays a crucial role in understanding the underlying changes and making informed decisions based on obtained results. The outputs of the Mann–Kendall test for applied time series are presented in Table 1.

The GRACE-FO data have indicated a declining trend in the Miandoab aquifer based on the Mann–Kendall test. Previous studies, such as those by Jeihouni et al. (2018), Dehghanipour et al. (2019), and Javadzadeh et al. (2020), have highlighted a declining trend in the groundwater level of this region according to the field data. Since GRACE-FO data encompass groundwater, the findings align well with the outcomes of these earlier investigations.

The spatial distribution of pumping wells and pumping rate are illustrated in Figure S5. The Miandoab aquifer faces significant water extraction for various purposes, for instance, agriculture, manufacturing, and residential consumption. Excessive pumping from the aquifer leads to a depletion of groundwater resources, causing the groundwater level to decline. The GRACE satellite data capture these changes in the gravitational field associated with the depletion of groundwater storage in the aquifer. Conversely, the recharge rate can be insufficient to restore the water being withdrawn. Reasons such as changes in farmlands and inefficient water monitoring can hinder recharge process. Consequently, the rate of groundwater extraction exceeds the natural replenishment rate, leading to a declining trend in the aquifer. Climate variability, including precipitation and temperature patterns variation, can influence water resources. Decrease of precipitation or increase in evapotranspiration because of shifting climate conditions can worsen the downward trend of the groundwater level.

The Urmia Lake level and area exhibit a decreasing trend, which is influenced by multiple and intricate reasons, consisting of climate change and human activities. Table 1 presents the data, indicating that the rate of decrease in the lake area is greater than that of the lake level. Urmia Lake is located in a region with a dry and arid climate. The high evaporation rates in the area cause the water surface to shrink more rapidly, leading to a faster decline in the area of the lake. The excessive extraction of water from the rivers and streams that feed into Urmia Lake for agriculture, industry, and domestic use has reduced the inflow of water into the lake. This imbalance between water withdrawal and replenishment has a more significant impact on the lake area compared to the lake level. Based on Table 1, the NDVI showed an increasing trend in the aquifer. The NDVI is a widely used remote-sensing indicator that measures the health and vitality of vegetation. The Miandoab aquifer is an agricultural area, and increasing the NDVI values could be related to enhanced agricultural techniques, including better irrigating methods, efficient water management, and the use of fertilizers. These practices can enhance crop health and productivity, leading to higher vegetation density and consequently higher the NDVI values. Alterations in land utilization trends, such as transforming regions into farmland, may lead to increase in the NDVI. The enlargement of farming regions can lead to higher vegetation density and subsequently higher the NDVI readings. Among teleconnection indices the AMO only showed increasing trend while there was no trend in other teleconnection indices. The AMO is a natural climate pattern. It can interact with and be modulated by long-term climate change. Global warming caused by human activities, such as the increase in greenhouse gas concentrations, can affect the behavior and intensity of climate patterns, including the AMO teleconnection. Changes in atmospheric circulation, oceanic processes, and other climate factors under the influence of climate change can lead to altered responses to the AMO, potentially resulting in increasing trends in certain variables.

Given that GWQ is influenced by various factors, it is important to investigate the coherence between GWQ and each individual factor separately. This approach can yield valuable insights into the specific effects of each factor on GWQ. WC can be used to reveal patterns of coherence, phase relationships, and time-varying associations between the parameters. Additionally, WC can indicate changes in the coherence patterns over time, providing insights into the dynamic relationships between the parameters. In this study, PWC and MWC were employed for investigating coherency between GWQ and different variables influencing GWQ.

In this section, WC, PWC, and MWC were calculated between GWQ (TDS) and GRACE-FO data, the NDVI, climate change, lake area, lake level, and teleconnection indices. Since the GWQ data are of every 6 months, the parameters that are on a monthly scale are upscaled to every 6 months. The WC analysis was employed for coherence analysis of the four zones obtained from the ensemble clustering method. The obtained clusters are distributed in the north, east, south, and west of the aquifer, which are respectively denoted as Clusters 1, 2, 3, and 4 (see Figure S6). In the following subsections, the results of WC, PWC, and MWC are illustrated.

Cluster 3 exhibits a significant antiphase correlation between the lake area and GWQ during the periods of 2018–2022 and scales 4–8. This cluster is located farther away from the shoreline, suggesting that their GWQ may be indirectly influenced by the decline in lake area. Lakes play a vital role in the hydrological balance of an aquifer system. They act as a source of recharge, replenishing groundwater through direct infiltration and seepage. When the area of the lake decreases, the recharge rates from the lake to the aquifer may decline. This reduction in recharge can disrupt the natural balance and increase the concentration of dissolved solids, contributing to higher TDS levels in the far parts of the aquifer. The shrinking of the lake may influence the flow patterns of groundwater within connected aquifer. A decrease in the lake area can change the hydraulic gradient and alter the direction of groundwater flow. As a result, groundwater from the surrounding areas may flow toward the far parts of the aquifer connected to the lake. This inflow of groundwater can introduce higher TDS water into the far parts, leading to an increase in TDS levels. The time series of Urmia Lake area is shown in Figure S9. As shown in Figure S9, the area of the lake reached its lowest value in 2018 and after that there is a slight increase in area. The impact of the increase in area is clear in Clusters 3 and 4, as shown in Figure 6(b), where the coherence has risen to 1. Cluster 2 showed significant coherence with lake level decline in scale 8. The arrows showed in phase 90°. Lakes and groundwater systems are often interconnected through hydrological processes. In coastal areas, lake level decline can increase the risk of saltwater intrusion into underlying aquifers. When a lake's water level decreases, the freshwater–saltwater interface can shift landward. This movement can cause saltwater from the ocean to infiltrate into the underlying groundwater, leading to increased TDS levels. Moreover, Vaheddoost & Aksoy (2018) presented a strong correlation between groundwater fluctuations on the eastern coast and changes in the Urmia Lake level. This agreement further strengthens the understanding that variations in the lake water level have a significant influence on the groundwater dynamics in the region.

Also, investigation of Figure 7 indicated that PDO showed shorter phase lag compared to AMO and SOI. Among the climate indices mentioned, the PDO would be more relevant to Urmia Lake compared to the SOI and the AMO. Urmia Lake is located in northwest Iran, which is geographically closer to the Pacific Ocean than the Atlantic Ocean. The PDO primarily focuses on SST variations in the North Pacific Ocean, making it more geographically proximate to Urmia Lake compared to the AMO, which pertains to the North Atlantic Ocean. While the SOI is associated with the ENSO phenomenon in the tropical Pacific, it may have a more indirect influence on Urmia Lake compared to the PDO. ENSO-related climate impacts, including precipitation patterns, tend to be more pronounced in regions closer to the Equator and along the Pacific coast, rather than in inland areas like Urmia Lake. Therefore, in terms of geographical proximity and potential climatic influence, the PDO would have a closer connection to Urmia Lake compared to the SOI and the AMO. Based on Figure 7, high coherency is noted between teleconnection parameters and GWQ in Cluster 3. The southeast of the aquifer has a higher elevation compared to other parts. The AMO and PDO are climate indices that primarily focus on SST variability in the North Atlantic and North Pacific oceans, respectively. While these indices are not directly associated with precipitation patterns like ENSO, they can indirectly influence atmospheric circulation and, consequently, precipitation in certain regions. The AMO is characterized by multidecadal fluctuations in the North Atlantic SSTs. During the positive phase of the AMO, the Atlantic Ocean's tropical and extratropical regions experience warmer-than-average SSTs, while the negative phase is associated with cooler SST anomalies. The AMO has been linked to changes in atmospheric circulation patterns, which can influence precipitation patterns in surrounding regions. Similarly, the PDO is a long-term climate pattern in the North Pacific characterized by shifts in SSTs. It has a more localized influence compared to the AMO. The PDO can influence atmospheric circulation patterns, which, in turn, can impact precipitation patterns. Mountains play a vital role in replenishing groundwater through diverse mechanisms. Orographic effects often result in higher levels of precipitation in mountainous regions compared to the surrounding low-lying areas. This abundant rainfall enhances recharge rates and increases the availability of water for groundwater replenishment. Moreover, during warmer months, the melting of snow contributes significantly to groundwater recharge. As the melted water infiltrates the subsurface by percolating through mountain slopes, it effectively replenishes the underlying aquifers. The WCs depicted in Figure 7 provide evidence of non-stationary responses observed in the GWQ in relation to large-scale climate oscillations. These non-stationary responses are evident as the WC, measured at specific periodicity bands, exhibits variability for each teleconnection index. The fluctuating WC values suggest that the interactions between teleconnection indices and GWQ are not constant over time, indicating a dynamic relationship between these factors. This non-stationarity underscores the complex and evolving nature of the impacts of climate oscillations on the hydroclimate dynamics of the lake watershed. The analysis of PWC between GWQ and SOI showed that coherence increases after 2010 in majority of the clusters, especially in Clusters 1 and 3 in which the coherence increased from the range of 0.6–0.7 to the range 0.9–1. Also, coherency increased in Clusters 2 and 4 from the range of 0.4–0.5 to 0.6–0.8. About analysis of PWC between GWQ and PDO, again increase in coherency was clear in all clusters, especially in Cluster 3, ranging from 0.6–0.7 to 0.9–1. The reason for this may be the changes in global climate patterns that have influenced the behavior of the SOI, leading to increased coherence in the Urmia Lake Basin. Also, alterations in atmospheric circulation patterns, such as changes in pressure systems or wind patterns, could have contributed to the enhanced coherence between the indices. Moreover, modifications in land-use practices in the region, such as deforestation or urbanization, may have impacted local climate dynamics, potentially influencing the SOI. In contrast to the SOI, the analysis of the PWC between GWQ and AMO indicated that the coherence declined during the study period in Clusters 1 and 4. However, there was an increase from 0.6–0.7 to 0.9–1 in recent years in Clusters 2 and 3.

The comparison between the simultaneous consideration of climate change and anthropogenic activities and combination of the teleconnection patterns effect showed that GWQ approximately has higher coherence with anthropogenic activities and climate change. Moreover, the comparison between PWC and MWC highlights that GWQ demonstrates significant coherence when considering the effects of other parameters simultaneously.

Comparing the MWC values for teleconnection indexes and for a combination of climate change and anthropogenic activities revealed that in the majority of clusters, the MWC values of teleconnections on a 0–4 scale displayed notable consistency with GWQ. In contrast, for the combination of climate change and anthropogenic activities, significant coherence was observed over periods ranging from 4 to 16. Teleconnection patterns have the potential to impact GWQ earlier than climate change and anthropogenic activities. Teleconnection patterns represent large-scale atmospheric circulation patterns. These patterns can influence weather and climate conditions over large geographic regions. Unlike localized anthropogenic activities, which may take time to accumulate and affect GWQ, teleconnection patterns can cause rapid and widespread changes in climate variables such as precipitation and temperature that directly impact groundwater systems. Teleconnection patterns are part of the Earth's natural climate variability, which can exhibit multi-year to decadal oscillations. These natural variations can induce fluctuations in climate parameters that influence groundwater systems. Changes in precipitation patterns, for instance, can alter recharge rates and the dilution or flushing of contaminants in groundwater. Teleconnection patterns can lead to abrupt shifts in weather patterns and climate conditions. For example, the onset of an El Niño event can trigger intense rainfall or drought conditions within a relatively short period. These rapid changes in weather can directly impact GWQ by altering the transport and fate of contaminants in the subsurface. Changes in GWQ can also interact with teleconnection patterns through feedback mechanisms. For instance, alterations in groundwater levels and quality can influence local climate conditions by modifying surface water availability, evaporation rates, and land surface processes. These feedback mechanisms can further enhance the influence of teleconnection patterns on GWQ.

This study aimed to investigate the influencing factors on GWQ, thus different factors such as climate change (temperature), anthropogenic activities (NDVI), lake level, lake area, TWS fluctuation (GRACE-FO), and teleconnection patterns (AMO, PDO, and SOI) were considered to investigate the coherence of these factors with GWQ. To evaluate the coherence across different periods and times, the WC was employed. For investigating the individual effects of multiple parameters, the PWC was utilized. Furthermore, to explore the combined effects of the factors, the MWC was employed in analysis. This approach allows for a comprehensive analysis of the impact of various factors on the study's objectives. Although some studies already applied WC analysis for assessment of groundwater level, investigation of coherence of GWQ in different zones of the aquifer with remote-sensing and teleconnection data was the gap in previous studies that was tried to be filled in by the current research. As TDS had higher coherence with other GWQ parameters, it was used in investigation. There were multiple piezometers in the study region, therefore clustering was used and finally four zones were considered and the central piezometers of each zone were used in assessments. In addition, the Mann–Kendall test was used for investigation of the applied factors’ trends, among which the factors of GRACE-FO data, lake level, and lake area have a decreasing trend, while NDVI and PDO show increasing trend. Among the teleconnection patterns, the PDO showed higher coherence between 0.9 and 1 with GWQ compared to the AMO and SOI. Comparison of the PWC showed the lake area has higher coherence with GWQ than lake level. Among the investigated factors, the coherence of GWQ with GRACE-FO data was more significant than other factors. Specifically, during the 8–10 period, most of the clusters exhibited an increase in coherence 0.9–1 after 2010, coinciding with a negative fluctuation in GRACE-FO. Although the climate change coherence with GWQ (0.9–1) was more than that for the anthropogenic activities (0.6–0.8), comparison of the coherence results between the anthropogenic and climate change impacts showed almost similar pattern in the 8–16 scales.

As a suggestion for future studies, the WC, PWC, and MWC can be applied in identification of the coherence between GWQ and different factors in future, and help decision-makers to better manage the GW sources.

This study is jointly supported by the Iran National Science Foundation (Grant No. 4021444) and the National Natural Science Foundation of China (Grant No. 42361144709).

Data cannot be made publicly available; readers should contact the corresponding author for details.

The authors declare there is no conflict.