Next Article in Journal
Transient Responses of Freshwater Lens Development and Seawater Intrusion Mitigation to Freshwater Injection in Unconfined Island Aquifers
Previous Article in Journal
Leveraging Artificial Intelligence in Hydrology to Process Citizen Science Photos of Water Levels
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Hydrology
Volume 13
Issue 5
10.3390/hydrology13050135
hydrology-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Review

GRACE Downscaling and Machine Learning Models for Groundwater Prediction: A Systematic Review

by
Mohammed S. Al Nadabi
,
Mohammed El-Diasty
*,
Talal Etri
and
Mohammad Reza Nikoo
Department of Civil and Architectural Engineering, Sultan Qaboos University, Muscat 123, Oman
*
Author to whom correspondence should be addressed.
Hydrology 2026, 13(5), 135; https://doi.org/10.3390/hydrology13050135
Submission received: 10 April 2026 / Revised: 7 May 2026 / Accepted: 10 May 2026 / Published: 14 May 2026
(This article belongs to the Section Hydrological and Hydrodynamic Processes and Modelling)
Browse Figures
Versions Notes

Abstract

Gravity Recovery and Climate Experiment (GRACE) satellites primarily monitor changes in land water storage, including groundwater, soil moisture, lake and river surface water, and canopy and snow water. However, its coarse spatial resolution of 0.25 degrees limits its ability to observe smaller basins. To assess aquifer depletion and evaluate a long-term water resource management framework, GRACE data are crucial. It remains rare for GRACE-focused studies to be conducted in great depth. A comprehensive review of 80 articles published between 2011 and 2025 was conducted using the Scopus and Web of Science databases. These articles focused on downscaling GRACE data using machine learning (ML) methods. The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) reporting guidelines were used in this review. This study highlights the attributes of ML models, the input variables used, the evaluation metrics, and the output resolution. Based on the analysis of the articles, random forest (RF) methods were used in the majority of the papers. Gradient boosting (GB), artificial neural networks (ANN), support vector machines (SVM), support vector regression (SVR), and long short-term memory (LSTM) were the most widely used ML methods. As input variables, rainfall (Pr), soil moisture (SM), and runoff (Qs) are essential. In 2011, there were very few journal articles; since 2021, the number has increased. The number of published studies from China was the highest (24), followed by the USA (12) and Iran (9). A total of 38 journals published reviewed articles. In terms of articles, Remote Sensing generates 19%, Journal of Hydrology has 10%, and Journal of Hydrology: Regional Studies has 8%. The paper also discusses limitations, challenges, recommendations, and potential future directions for improving the accuracy of the GWS change prediction model.

1. Introduction

Groundwater resources have a significant impact on human life across various aspects, including drinking water supply, agriculture, and economic development [1]. Furthermore, the indirect impact of groundwater resources is clear on the environment and ecosystems. Approximately 50% of the global population depends on groundwater for drinking water, and groundwater accounts for 43% of total agricultural water consumption [2]. Water covers nearly 71% of Earth’s surface, of which 97% is seawater and only 3% is available as freshwater, as reported in the global water distribution shown in Figure 1 [3].
Groundwater storage (GWS) change prediction refers to the estimation of temporal variations in the amount of water stored in an aquifer or groundwater system. It involves using remote sensing or well-observed data and/or modeling approaches to determine how groundwater levels and volumes change over time due to natural processes, such as climate variability, or human activities, such as excessive groundwater extraction. For sustainable socioeconomic development, the long-term outlook of the GWS anomaly must be clearly understood to inform policy and decision-making in the water sector, enabling more comprehensive strategies for better management of this valuable resource.
The standard GWS monitoring approach relies on monitoring wells that directly measure GWLs [4]. Nevertheless, this approach has explicit shortcomings, including limited spatial coverage, unevenly distributed wells, less accessible areas, and the challenges and costs of maintaining these observation wells. Furthermore, the temporal resolution of data collection is often insufficient, as measurements are taken at intervals that fail to capture rapid changes in GWS during critical events such as droughts or heavy rainfall. Collectively, these limitations emphasize the need for alternative monitoring methods.
Advancements in satellite technology have significantly enhanced the estimation of groundwater storage (GWS) through remote sensing [5]. The Gravity Recovery and Climate Experiment (GRACE) mission, a joint initiative between the US National Aeronautics and Space Administration and the German Aerospace Centre (NASA and DLR), was launched on March 17, 2002. This mission comprises two satellites that provided comprehensive measurements of variations in Earth’s gravity field for over 15 years [6]. The GRACE Follow-On (GRACE-FO) mission, which launched twin satellites in May 2018, continues the work of the previous mission. The twin satellites are about 220 km apart at an altitude of 500 km, from which they will measure Earth’s gravity field. Changes in Earth’s gravitational field cause a change in the separation between these two satellites. This change is measured using onboard K-Band microwave ranging systems and additional data. This information is necessary for understanding variations in Earth’s gravity due to changes in TWS [7,8,9], including groundwater, surface water, soil moisture, and snow water storage. If one subtracts the anomalies in soil moisture, surface water, and snow water from the GRACE data, the resulting anomaly can be interpreted as a groundwater storage anomaly. However, due to GRACE’s coarse 0.25° resolution, the data cannot be used to study changes in water storage in smaller basins and must first be reconstructed to a more readily studied spatial scale.
Downscaling is a method for enhancing the spatial resolution of an observed quantity by combining higher-resolution data from multiple sources. In general, GRACE downscaling approaches can be grouped into dynamic downscaling (DD) and statistical downscaling (SD). DD methods rely on assimilating GRACE data into physically based land-surface or hydrological models, using atmospheric forcing and processed representations to generate high-resolution estimates of water storage [10]. In contrast, SD methodologies create empirical links between observations gathered from GRACE and detailed environmental variables, including precipitation, evapotranspiration, land surface temperature, vegetation indices, and topography, that affect GW storage. SD methods are easier to implement, computationally efficient, and generally data-driven [11]. Moreover, SD is increasingly reliant on machine learning (ML) systems, which can model complex nonlinear relationships between predictors and target variables without requiring explicit physical assumptions. ML-based SD can be categorized into shallow learning and deep learning frameworks. Shallow learning methods, such as artificial neural networks (ANNs) and traditional regression models, generally rely on manual feature engineering and have relatively simple architectures. In comparison, deep learning models employ multilayer architectures that are particularly effective for modeling complex, noisy, and nonlinear time series, making them well-suited for hydroclimatic applications [12].
The objectives of this review encompass a systematic identification of peer-reviewed studies that apply machine learning (ML) to groundwater modeling or to the estimation of groundwater storage anomalies (GWAS) using GRACE-derived data. These studies will be classified according to their primary aims: downscaling, reconstruction, prediction or forecasting, and hybrid modeling. The review will further summarize key elements such as the ML algorithms employed, such as artificial neural networks (ANN), support vector machines (SVM), random forest (RF), and others, alongside data inputs, spatial and temporal scales, validation strategies, and major findings. The challenges and limitations in these applications will be highlighted, followed by proposed recommendations for future research directions and best practices.
While there have been recent reviews of prior work in this area [13,14], many gaps remain. This literature review has gone beyond previous work by not only adding to the number of new methods developed in 2024 and 2025 but also by providing a more application-oriented approach to classifying the new methods researchers used to study groundwater resource availability with GRACE. Furthermore, this review addresses key challenges, such as uncertainty propagation and spatial resolution limitations, more critically than previous reviews, providing a clearer understanding of how those issues are not being addressed adequately in earlier reviews. Overall, this review provides the most up-to-date and analytically sound synthesis of the current state of research in this field.

2. Review Methodology

Systematic reviews are crucial for understanding the relations within a specific field and combining existing knowledge. They make a valuable contribution by highlighting current research gaps and guiding the direction of future research. Therefore, utilizing an evident, organized, and properly designed methodology is essential for achieving reliable results. There are a variety of systematic review methodologies, which vary by goal, such as Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA), Scoping Review, Qualitative Systematic Review, Mixed Methods Reviews, and Meta-analysis. We applied the PRISMA framework in this study. It provides a reliable framework for evaluating whether literature is relevant to our study subject and for establishing precise inclusion and exclusion criteria. Following the PRISMA methodology, our study comprises four main steps: identification, screening, eligibility assessment, and inclusion (see Figure 2) [13]. Initially, it determines the full scope of potential data and optimizes the search filters accordingly. Subsequently, predefined criteria for including or excluding samples are established to align with the objectives of the reviews, leading to the selection of acceptable research. The first stage involved defining the screening and acceptability criteria, focusing on eligible studies that met the review’s inclusion criteria. The second stage involved extracting information from the selected final studies [14]. Lastly, we reported and analyzed the results. Our workflow chart for PRISMA review is shown in Figure 3.
The systematic review was conducted on 25 December 2025, yielding 190 records from the selected databases that met the search criteria. The reference, study location, native and downscaled spatial resolution, machine learning model, input variables, evaluation metrics, and related information were recorded in an Excel spreadsheet. After removing duplicate records (n = 47), 143 records remained and were screened for titles and abstracts (Figure 3). In addition, only peer-reviewed articles published in English were considered during the screening process, and non-English and non-peer-reviewed records were excluded at this stage. This screening was based on a set of inclusion criteria. The following criteria were employed to evaluate whether the article had been excluded (1) or included (2).
  • Exclusion criteria: The following criteria were employed to exclude any article:
    • The title did not align with the study’s objectives (which focused on ML for downscaling GRACE-derived TW and GW storage), leading to the rejection of 28 papers (Figure 3).
    • The abstract failed to convey the original research, leading to the removal of 15 papers (Figure 3).
  • Inclusion criteria: The study included 80 papers that fulfilled the following criteria and were considered in the review process:
    • Studies applying ML algorithms, such as GB, RF, SVM, and deep learning, to spatial downscale GRACE data.
    • Publications using GRACE/GRACE-FO data to downscale to a finer spatial resolution using auxiliary datasets such as precipitation, MODIS, or land surface models.
    • Peer-reviewed journal articles in English, published up to December 2025, presenting original methods, results, and validation metrics (e.g., RMSE, MAE, NSE, R2).
Using specific keywords for search from Web of Science and Scopus as follows; Using specific keywords for search from Web of Science and Scopus as follows; (“GRACE” OR “GRACE-FO” OR “GRACE Follow-On”) AND (“groundwater” OR “ground water” OR “groundwater storage” OR “GWS”) AND (“machine learning” OR “random forest” OR “deep learning” OR “ensemble” OR “downscaling” OR “Artificial Intelligence”). The selected studies were categorized along multiple dimensions. These included the machine learning algorithms employed (e.g., RF, SVM, ANN, and GB), the region and spatial resolution (e.g., global, basin, aquifer, or country scales at coarse or fine resolutions), and data inputs or predictors like GRACE TWSA, Global Land Data Assimilation System (GLDAS) outputs, meteorological and land-surface variables, well data, or teleconnection indices. Classifications also covered validation metrics such as R2, RMSE, NSE, and correlation coefficient, along with key findings and limitations.

3. Results and Discussion

Since the launch of the GRACE satellite, many studies have estimated large-scale GWS anomalies using GRACE data. Variations in groundwater storage can be estimated by incorporating auxiliary data, e.g., soil moisture and snow. In their studies [15] separated GWS from water storage data in GRACE via GLDAS. The GRACE-Derived Total Water Storage Anomalies (TWSA) consists of components of water, including soil moisture (SM), snow water equivalent (SWE), surface water (SW), and canopy water (CW), as well as groundwater (GW) [16]. The vertical water model (Equation (1)) is used to estimate the groundwater storage anomaly.
GWSA = TWSA − (SMSA + SWEA + SWSA)
In arid regions, some parameters of Equation (1) can be excluded under certain conditions, so GW storage anomalies and soil moisture storage anomalies determine the TWSA.

3.1. Downscaling GRACE Data Utilizing ML-Based Models

The spatial resolution of GRACE (0.25°) is considered coarse, limiting its suitability for analyzing changes in water storage, especially in small-scale basins. Spatial downscaling is required to achieve the desired resolution of observed data by combining finer-scale information from various sources. Investigations within the framework of downscaling GRACE are presented in Table 1. It summarizes approximately 40 selected papers published between 2011 and 2025 that met the inclusion criteria, highlighting their classifications and main findings. While the full table is not included here due to space limitations and is provided in the Supplementary Materials (Table S1), we discuss the trends and patterns of the complete data extracted below. As indicated in the extracted data, some papers achieved a spatial resolution of about 1 km using GRACE data. However, because most of the articles examined relied on GLDAS data for their analyses, the achieved resolution is 0.25° (~27.8 km). Without exact physical modeling, machine learning (ML), which uses data-driven algorithms, can learn patterns from datasets. ML has been widely used in hydrology, including rainfall-runoff modeling, water quality monitoring, estimation of groundwater storage changes, prediction of groundwater infiltration, and groundwater contamination. It has various advantages, including handling nonlinear relationships, automating feature extraction, especially in deep learning, and integrating multi-source data. There are different ML models, such as deep neural networks (DNNs), random forests (RF), support vector machines (SVMs), and gradient boosting (GB). Using remote sensing data, such as GRACE, alongside ML is a recent and growing trend.
ML can handle data gaps and downscale GRACE’s coarse spatial resolution to finer scales for deeper regional understanding by integrating GRACE-derived anomalies with input variables such as precipitation, temperature, runoff, soil moisture, and more. Based on the reviewed articles, the most frequently applied ML algorithms were Random Forest (RF), Gradient Boosting (GB), Artificial Neural Network (ANN), Support Vector Machine (SVM), Support Vector Regression (SVR), and Long Short-Term Memory (LSTM), as shown in Figure 4.
Cross-study evaluations of model performance reveal patterns in the effectiveness of various machine-learning model types. Random forests (RF) and gradient boosting (GB) ensemble models generally achieve better prediction accuracy than other model types when modeling nonlinear relationships among hydroclimate variables [56]. ANN models capture temporal and spatial dependence effectively; therefore, they perform well in complex, dynamic physical settings [57]. However, linear regression models generally produce more readable descriptions of the spatial patterns being modeled than more complex models, but they typically yield lower overall prediction accuracy. Finally, model performance variability will be influenced by the methods used to select input data, the amount of input data used to train the models, and the regional hydroclimate conditions at the study site(s) [58]. As such, these analysis results indicate that choosing a downscaling model for GRACE must be based on the application’s purpose, the characteristics of the data to be modeled, and the objectives of the analysis.

3.2. Input Variables

Commonly used input variables for the downscaling of groundwater storage (GWS) change include soil moisture (SM), precipitation (Pr), evapotranspiration (ET), and temperature (T). In addition to these variables, others such as the Normalized Difference Vegetation Index (NDVI), canopy water storage (CWS), land surface temperature (LST), Digital Elevation Model (DEM), slope, and aspect, as well as numerous other topographic features, may also be widely integrated to represent changes in vegetation dynamics or terrain-controlled hydrological processes. Seasonal and surface water dynamics can often be represented using snow water equivalent (SWE) and surface runoff (Qs), and some studies include specialized parameters such as drainage density, aquifer thickness, and lithology to improve model representation [11].
An analysis of multiple studies indicates that, despite differences in cross-study methods and in how groundwater recharge and depletion are predicted, there are consistent trends in how various input parameters are selected across regions based on their dominant predictive capability. Studies conducted in arid and semi-arid areas (e.g., the Middle East and South Asia) have shown that precipitation, evapotranspiration, and temperature are the primary predictors, as they directly affect groundwater recharge and depletion [59]. On the other hand, studies conducted in humid areas often use vegetation indices, soil moisture, and topographic variables as core input variables because they provide more accurate representations of land-atmosphere interactions than other variables. The variation and/or choice of input variables reflect hydroclimatic conditions and data availability, underscoring the critical need for locally specific designs to build models that achieve improved downscaling accuracy (Figure 5).

3.3. Evaluation Metrics

Evaluation metrics are key indicators of a downscaled model’s accuracy and reliability. Common metric types used in performance evaluation include Nash–Sutcliffe Efficiency (NSE), Root Mean Square Error (RMSE), Mean Absolute Error (MAE), Pearson’s Correlation Coefficient (R), and Coefficient of Determination (R2) [60]. Each of these metric types provides insight into how well the model performed relative to the others. By reviewing the various studies presented, it is possible to understand how evaluation metrics are selected based on the model’s purpose. Error-based metrics, such as RMSE and MAE, are more commonly used to evaluate the accuracy of predictions from local datasets (R, R2) than correlation metrics [61].
Additionally, when evaluating hydrologic models, the NSE metric is preferred. Studies looking at downscaling (spatial) will use correlation metrics more often than error metrics, whereas studies examining prediction modeling will do the reverse. These contrasts tell us the importance of choosing appropriate metrics to evaluate models based on their objectives (Figure 6).

3.4. Noted Patterns in the Reviewed Studies

Publications in GRACE-derived TWSA and GWSA have shown an increasing trend over the last few years. Figure 7 shows the trend from 2011 to 2025. The journal articles were limited to 2011–2020; since 2021, the number of articles has trended upward. According to this review, most articles focused on specific countries (Figure 8 and Figure 9). For instance, among 80 publications, China (24), followed by the USA (12), Iran (9), and India (6), are the top 4 countries with the highest number of publications. The reviewed articles were published in 38 various journals. A larger portion of the articles is attributed to Remote Sensing (19%), Journal of Hydrology (10%), Journal of Hydrology: Regional Studies (8%), and Science of the Total Environment (6%) (see Figure 10).
There are several reasons for the increase in publications since 2021, including greater availability of long-term GRACE/GRACE-FO datasets, advancements in machine learning techniques, and improved computing capacity to manage large geospatial datasets [6]. There is also a growing concern about global groundwater depletion, particularly in arid and semi-arid regions, which has led to increased interest in researching high-resolution monitoring and predictive modeling. The concentration of publications from countries such as China, the United States, Iran, and India is likely the result of a combination of high groundwater stress, significant research funding, and greater access to remote sensing data and open-source analytical tools. The correlation between both research activities and environmental pressures laid the foundation for discussions of the drivers of regional groundwater change. In particular, advances in downscaling and modeling methods have been developed by communities currently experiencing high levels of agricultural water use or groundwater overexploitation [58].

3.5. Drivers of Groundwater Storage Decline

Declines in groundwater storage identified in the reviewed papers are primarily driven by the intensive human abstraction and climate-related reduction in recharge [51]. On the North China Plain, the downscaled GWSA decreased by −0.51 ± 7.25 mm/yr [62]. A decline in GW storage was observed in the Indus basin irrigation system at rates of 3.39 ± 0.49 mm/yr and 4.16 ± 0.26 mm/yr in the lower Indus basin from 2003 to 2016 [41]. The prediction of GWSA during the years 2005–2017 showed a steady linear rise of about 0.77 ± 0.30 mm/yr in the Middle and Lower Yangtze River Basin of China, excluding the Han River Basin, where GWSA declined by −1.18 ± 0.38 mm/yr, resulting from decreased precipitation [63]. Regarding GWS in Iran, a decline of up to 15 mm/yr was reported [64].
These regions arise from the effects of human activity and climate on groundwater storage changes worldwide. Some of these areas experience ongoing groundwater depletion from intensive irrigation and high water demand, whereas many others are showing localized recoveries driven by both precipitation variations and basin hydrologic conditions. Collectively, the evidence of increasing global groundwater stress supports the need for integrated monitoring systems and advanced, data-driven modeling frameworks to ensure sustainable management of global groundwater supplies [65].
Groundwater storage trend data typically come from both GRACE (satellite observations) and machine learning (ML)-based downscaling, depending on the study. GRACE provides reliable large-scale trends but also has a large pixel size, which can obscure local differences [66]. Conversely, ML-based downscaling provides finer spatial representations and may help accurately identify regional differences in groundwater dynamics. Several studies suggest that downscaled products show more spatially refined depletion and/or hotspots that were not observable in GRACE data with large pixel sizes. However, the reliability of these trends depends heavily on model performance, input variable selection, and validation techniques. Downscaling can help interpret groundwater decline; however, in some situations, it can also introduce additional uncertainty if improperly constrained (i.e., under-constrained). Therefore, to derive strong, physically meaningful conclusions, it is necessary to combine trend analysis with downscaling model evaluation.

4. Challenges and Limitations

Several persistent challenges remain despite the progress shown above. GRACE grid cells are large (hundreds of kilometers). When applied to local aquifer wells or pumping zones, representativeness errors can be introduced. Downscaling may help, but it may exacerbate uncertainties. Groundwater response may also be missed at a monthly resolution (e.g., during pumping events and seasonal irrigation). Many studies rely on relatively sparse well networks for validation, often in regions with good monitoring, due to spatial heterogeneity in aquifers, pumping, recharge, and land use. This leads to weak or absent validation in data-poor regions.
Additionally, converting between a groundwater storage anomaly and groundwater level requires knowledge of aquifer parameters (specific yield, etc.), which are often lacking. ML models frequently ingest many predictor variables (climate, land cover, pumping proxies, teleconnections). The risks of misspecification, collinearity, and overfitting remain. ML offers correlations but not causality; therefore, interpretability is lower than that of physical models. Although many ML studies report point estimates (R2, RMSE), they rarely quantify prediction uncertainty, model sensitivity, or error propagation (from GRACE, predictor inputs, and model parameters). Several models are tailored for specific regions (e.g., Iran, Central Valley) and may not translate well to other hydrogeological settings. Downscaling studies may obscure the uncertainty amplification from coarse GRACE inputs. The ability to generalize across regions is rarely tested. The use of ensemble and stacking methods mitigates some of these issues, but larger transfer-learning studies are rare.
Several researchers have recognized that one of the major shortcomings of many of these studies is the potential transfer of uncertainty from the coarse-resolution input datasets (GRACE, GLDAS) to the downscaled groundwater products [67]. Often, GLDAS variables are used at spatial scales similar to those of GRACE, thus limiting the extent to which spatial detail can be truly enhanced. There are also greater concerns about how well the downscaled outputs are independent of one another and how reliable they are. In addition, without clearly defining or quantifying the uncertainty of these inputs and validating against high-resolution in situ observations, it becomes increasingly difficult to assess the resulting products accurately. Given that uncertainty propagation analysis and multi-source validation frameworks are necessary for assessing the remaining uncertainty and robustness of information derived from downscaling groundwater estimates, future studies must implement these techniques.

5. Future Research Directions

Groundwater investigations using the GRACE satellites should be improved by applying machine learning methods to more robust analyses of datasets that leverage statistical and physically based methods. The use of machine learning methods, such as transformers and other advanced architectures, is also expected to improve model accuracy by improving the models’ ability to recognize dynamic groundwater storage patterns. Additionally, developing hybrid models that combine data-driven and physically based approaches will increase overall confidence in hydrological models.
There are many ways to deal with the spatial resolution limitations of GRACE. Examples include advanced downscaling and data fusion techniques for different types of data from various sources (e.g., remote sensing products, complex climate reanalysis datasets, or in situ observations). Uncertainty quantification methods are also necessary to improve confidence in models’ ability to predict future conditions and to aid decision-making.
Finally, future research should focus on applying these techniques in areas with limited raw data and increasingly scarce water resources, enabling improved groundwater resource management through enhanced monitoring. Furthermore, future studies may benefit from integrating geostatistical approaches, such as kriging-based spatial interpolation, with machine learning frameworks to improve the spatial representation and uncertainty characterization of downscaled GRACE products. Future studies should also investigate model generalizability across hydroclimatic regions using transfer-learning and cross-regional ensemble approaches.

6. Conclusions

In this systematic review, GRACE data were used to map the recent advances in machine-learning applications for groundwater modeling. Our study categorized studies into downscaling, reconstruction/hindcasting, prediction/forecasting, and hybrid modeling; we found that random forests and gradient boosting methods dominate, ensemble models and stacking models are on the rise, and deep learning and hybrid physics–ML approaches are gaining popularity. A key strength of the field is its ability to use GRACE data to determine groundwater storage changes, increase spatial resolution through machine-learning downscaling, and predict or extend groundwater fluctuations when data are sparse. Scale mismatch, validation scarcity, interpretation, uncertainty quantification, and transferability across a variety of hydrogeologic contexts remain significant challenges. To improve uncertainty management, deep learning, multi-sensor fusion, high-resolution modeling, and operational water management integration should be embraced in the field in the future. With groundwater stress intensifying globally, ML-enabled GRACE-based modeling offers a promising solution for improved monitoring, understanding, and management of aquifers.

Supplementary Materials

The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/hydrology13050135/s1, Table S1: Complete data extraction of reviewed studies including model type, input variables, spatial resolution, and performance metrics (Additional references are included [4,11,63,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98]).

Author Contributions

Conceptualization, M.S.A.N., M.E.-D., T.E. and M.R.N.; methodology, M.S.A.N.; software, M.S.A.N.; validation M.S.A.N.; formal analysis, M.S.A.N.; investigation, M.S.A.N., M.E.-D., T.E. and M.R.N.; resources, M.S.A.N., M.E.-D., T.E. and M.R.N.; data curation, M.S.A.N., M.E.-D., T.E. and M.R.N.; writing—original draft preparation, M.S.A.N.; writing—review and editing, M.S.A.N., M.E.-D., T.E. and M.R.N.; visualization, M.S.A.N.; supervision, M.E.-D., T.E. and M.R.N. All authors have read and agreed to the published version of the manuscript.

Funding

This research was funded by the Ministry of Higher Education, Research and Innovation (grant no. RC/GRG/ENG/C&A/25/460) and the Article Processing Fees (APF) by Sultan Qaboos University.

Data Availability Statement

Data is contained within the article.

Conflicts of Interest

The authors declare no conflicts of interest.

Abbreviations

The following abbreviations are used in this manuscript:
GRACEGravity Recovery and Climate Experiment
PRISMAPreferred Reporting Items for Systematic Reviews and Meta-Analyses
NASANational Aeronautics and Space Administration
DLRDeutsches Zentrum für Luft- und Raumfahrt (German Aerospace Center)
TWSTerrestrial Water Storage
GWSGroundwater Storage
SMSoil Moisture
SWESnow Water Equivalent
SWSurface Water
CWCanopy Water
ETEvapotranspiration
PrPrecipitation
QsSurface Runoff
GLDASGlobal Land Data Assimilation System
NDVINormalized Difference Vegetation Index
LSTLand Surface Temperature
DEMDigital Elevation Model
TTemperature
MLMachine Learning
RFRandom Forest
GBGradient Boosting
ANNArtificial Neural Network
SVMSupport Vector Machine
SVRSupport Vector Regression
LSTMLong Short-Term Memory
DNNDeep Neural Network
SDStatistical Downscaling
DDDynamic Downscaling
RMSERoot Mean Square Error
MAEMean Absolute Error
NSENash–Sutcliffe Efficiency
R2Coefficient of Determination

References

  1. Singh, L.; Saravanan, S. Satellite-Derived GRACE Groundwater Storage Variation in Complex Aquifer System in India. Sustain. Water Resour. Manag. 2020, 6, 43. [Google Scholar] [CrossRef]
  2. Li, L.; Barry, D.A.; Pattiaratchi, C.B.; Masselink, G. BeachWin: Modelling Groundwater Effects on Swash Sediment Transport and Beach Profile Changes. Environ. Model. Softw. 2002, 17, 313–320. [Google Scholar] [CrossRef]
  3. Ahuja, S. Water Quality Worldwide. In Handbook of Water Purity and Quality; Elsevier: Amsterdam, The Netherlands, 2021; pp. 19–33. [Google Scholar]
  4. Miro, M.E.; Famiglietti, J.S. Downscaling GRACE Remote Sensing Datasets to High-Resolution Groundwater Storage Change Maps of California’s Central Valley. Remote Sens. 2018, 10, 143. [Google Scholar] [CrossRef]
  5. Chen, Z.; Zheng, W.; Yin, W.; Li, X.; Ma, M. Improving Spatial Resolution of GRACE-Derived Water Storage Changes Based on Geographically Weighted Regression Downscaled Model. IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens. 2023, 16, 4261–4275. [Google Scholar] [CrossRef]
  6. Tapley, B.D.; Bettadpur, S.; Watkins, M.; Reigber, C. The Gravity Recovery and Climate Experiment: Mission Overview and Early Results. Geophys. Res. Lett. 2004, 31. [Google Scholar] [CrossRef]
  7. Mukherjee, A.; Ramachandran, P. Prediction of GWL with the Help of GRACE TWS for Unevenly Spaced Time Series Data in India: Analysis of Comparative Performances of SVR, ANN and LRM. J. Hydrol. 2018, 558, 647–658. [Google Scholar] [CrossRef]
  8. Scanlon, B.R.; Longuevergne, L.; Long, D. Ground Referencing GRACE Satellite Estimates of Groundwater Storage Changes in the California Central Valley, USA. Water Resour. Res. 2012, 48. [Google Scholar] [CrossRef]
  9. Syed, T.H.; Famiglietti, J.S.; Rodell, M.; Chen, J.; Wilson, C.R. Analysis of Terrestrial Water Storage Changes from GRACE and GLDAS. Water Resour. Res. 2008, 44. [Google Scholar] [CrossRef]
  10. Jyolsna, P.J.; Kambhammettu, B.V.N.P.; Gorugantula, S. Application of Random Forest and Multi-Linear Regression Methods in Downscaling GRACE Derived Groundwater Storage Changes. Hydrol. Sci. J. 2021, 66, 874–887. [Google Scholar] [CrossRef]
  11. Seyoum, W.M.; Kwon, D.; Milewski, A.M. Downscaling GRACE TWSA Data into High-Resolution Groundwater Level Anomaly Using Machine Learning-Based Models in a Glacial Aquifer System. Remote Sens. 2019, 11, 824. [Google Scholar] [CrossRef]
  12. Nourani, V.; Ghaneei, P.; Kantoush, S.A. Robust Clustering for Assessing the Spatiotemporal Variability of Groundwater Quantity and Quality. J. Hydrol. 2022, 604, 127272. [Google Scholar] [CrossRef]
  13. Nourani, V.; Jabbarian Paknezhad, N.; Ng, A.; Wen, Z.; Dabrowska, D.; Üzelaltınbulat, S. Application of the Machine Learning Methods for GRACE Data Based Groundwater Modeling, a Systematic Review. Groundw. Sustain. Dev. 2024, 25, 101113. [Google Scholar] [CrossRef]
  14. Afrifa, S.; Zhang, T.; Appiahene, P.; Varadarajan, V. Mathematical and Machine Learning Models for Groundwater Level Changes: A Systematic Review and Bibliographic Analysis. Futur. Internet 2022, 14, 259. [Google Scholar] [CrossRef]
  15. Rodell, M.; Chen, J.; Kato, H.; Famiglietti, J.S.; Nigro, J.; Wilson, C.R. Estimating Groundwater Storage Changes in the Mississippi River Basin (USA) Using GRACE. Hydrogeol. J. 2007, 15, 159–166. [Google Scholar] [CrossRef]
  16. Foroumandi, E.; Nourani, V.; Huang, J.J.; Moradkhani, H.; Jeanne Huang, J.; Moradkhani, H. Drought Monitoring by Downscaling GRACE-Derived Terrestrial Water Storage Anomalies: A Deep Learning Approach. J. Hydrol. 2023, 616, 128838. [Google Scholar] [CrossRef]
  17. Yirdaw, S.Z.; Snelgrove, K.R. Regional Groundwater Storage from GRACE over the Assiniboine Delta Aquifer (ADA) of Manitoba. Atmos. Ocean. 2011, 49, 396–407. [Google Scholar] [CrossRef]
  18. Verma, K.; Katpatal, Y.B. Groundwater Monitoring Using GRACE and GLDAS Data after Downscaling Within Basaltic Aquifer System. Groundwater 2020, 58, 143–151. [Google Scholar] [CrossRef]
  19. Zhang, G.; Zheng, W.; Yin, W.; Lei, W. Improving the Resolution and Accuracy of Groundwater Level Anomalies Using the Machine Learning-Based Fusion Model in the North China Plain. Sensors 2020, 21, 46. [Google Scholar] [CrossRef]
  20. Ali, S.; Liu, D.; Fu, Q.; Cheema, M.J.M.; Pham, Q.B.; Rahaman, M.M.; Dang, T.D.; Anh, D.T. Improving the Resolution of GRACE Data for Spatio-Temporal Groundwater Storage Assessment. Remote Sens. 2021, 13, 3513. [Google Scholar] [CrossRef]
  21. Vishwakarma, B.D.; Zhang, J.; Sneeuw, N. Downscaling GRACE Total Water Storage Change Using Partial Least Squares Regression. Sci. Data 2021, 8, 95. [Google Scholar] [CrossRef]
  22. Zuo, J.; Xu, J.; Chen, Y.; Li, W. Downscaling Simulation of Groundwater Storage in the Tarim River Basin in Northwest China Based on GRACE Data. Phys. Chem. Earth Parts A/B/C 2021, 123, 103042. [Google Scholar] [CrossRef]
  23. Ali, S.; Liu, D.; Fu, Q.; Cheema, M.J.M.; Pal, S.C.; Arshad, A.; Pham, Q.B.; Zhang, L. Constructing High-Resolution Groundwater Drought at Spatio-Temporal Scale Using GRACE Satellite Data Based on Machine Learning in the Indus Basin. J. Hydrol. 2022, 612, 128295. [Google Scholar] [CrossRef]
  24. Yin, W.; Zhang, G.; Liu, F.; Zhang, D.; Zhang, X.; Chen, S. Improving the Spatial Resolution of GRACE-Based Groundwater Storage Estimates Using a Machine Learning Algorithm and Hydrological Model. Hydrogeol. J. 2022, 30, 947–963. [Google Scholar] [CrossRef]
  25. Pascal, C.; Ferrant, S.; Selles, A.; Maréchal, J.-C.; Paswan, A.; Merlin, O. Evaluating Downscaling Methods of GRACE Data: A Case Study over a Fractured Crystalline Aquifer in South India. Hydrol. Earth Syst. Sci. Discuss. 2022, 2022, 1–25. [Google Scholar]
  26. Yin, W.; Zhang, G.; Han, S.; Yeo, I.; Zhang, M. Improving the Resolution of GRACE-Based Water Storage Estimates Based on Machine Learning Downscaling Schemes. J. Hydrol. 2022, 613, 128447. [Google Scholar] [CrossRef]
  27. Agarwal, V.; Akyilmaz, O.; Shum, C.K.; Feng, W.; Haritashya, U.; Chen, W. Machine Learning Application for Modeling High-Resolution Groundwater Storage Variations in North China Plain. Res. Sq. 2022, preprint. [Google Scholar] [CrossRef]
  28. Pulla, S.T.; Yasarer, H.; Yarbrough, L.D. GRACE Downscaler: A Framework to Develop and Evaluate Downscaling Models for GRACE. Remote Sens. 2023, 15, 2247. [Google Scholar] [CrossRef]
  29. Khorrami, B. Satellite-Based Investigation of Water Stress at the Basin Scale: An Integrated Analysis of Downscaled GRACE Estimates and Remotely Sensed Data. J. Hydroinformatics 2023, 25, 1501–1512. [Google Scholar] [CrossRef]
  30. Agarwal, V.; Akyilmaz, O.; Shum, C.K.K.; Feng, W.; Yang, T.-Y.T.; Forootan, E.; Syed, T.H.; Haritashya, U.K.; Uz, M. Machine Learning Based Downscaling of GRACE-Estimated Groundwater in Central Valley, California. Sci. Total Environ. 2023, 865, 161138. [Google Scholar] [CrossRef]
  31. Sabzehee, F.; Amiri-Simkooei, A.R.; Iran-Pour, S.; Vishwakarma, B.D.; Kerachian, R. Enhancing Spatial Resolution of GRACE-Derived Groundwater Storage Anomalies in Urmia Catchment Using Machine Learning Downscaling Methods. J. Environ. Manag. 2023, 330, 117180. [Google Scholar] [CrossRef] [PubMed]
  32. Ghaffari, Z.; Easson, G.; Yarbrough, L.D.; Awawdeh, A.R.; Jahan, M.N.; Ellepola, A. Using Downscaled GRACE Mascon Data to Assess Total Water Storage in Mississippi Alluvial Plain Aquifer. Sensors 2023, 23, 6428. [Google Scholar] [CrossRef]
  33. Khorrami, B.; Ali, S.; Gündüz, O. Investigating the Local-Scale Fluctuations of Groundwater Storage by Using Downscaled GRACE/GRACE-FO JPL Mascon Product Based on Machine Learning (ML) Algorithm. Water Resour. Manag. 2023, 37, 3439–3456. [Google Scholar] [CrossRef]
  34. Tao, H.; Al-Sulttani, A.H.; Salih, S.Q.; Mohammed, M.K.A.A.; Khan, M.A.; Beyaztas, B.H.; Ali, M.; Elsayed, S.; Shahid, S.; Yaseen, Z.M. Development of High-Resolution Gridded Data for Water Availability Identification through GRACE Data Downscaling: Development of Machine Learning Models. Atmos. Res. 2023, 291, 106815. [Google Scholar] [CrossRef]
  35. Yazdian, H.; Salmani-Dehaghi, N.; Alijanian, M. A Spatially Promoted SVM Model for GRACE Downscaling: Using Ground and Satellite-Based Datasets. J. Hydrol. 2023, 626, 130214. [Google Scholar] [CrossRef]
  36. Rafik, A.; Ait Brahim, Y.; Amazirh, A.; Ouarani, M.; Bargam, B.; Ouatiki, H.; Bouslihim, Y.; Bouchaou, L.; Chehbouni, A. Groundwater Level Forecasting in a Data-Scarce Region through Remote Sensing Data Downscaling, Hydrological Modeling, and Machine Learning: A Case Study from Morocco. J. Hydrol. Reg. Stud. 2023, 50, 101569. [Google Scholar] [CrossRef]
  37. Shilengwe, C.; Banda, K.; Nyambe, I. Machine Learning Downscaling of GRACE/GRACE-FO Data to Capture Spatial-Temporal Drought Effects on Groundwater Storage at a Local Scale under Data-Scarcity. Environ. Syst. Res. 2024, 13, 38. [Google Scholar] [CrossRef]
  38. Wang, Y.; Li, C.; Cui, Y.Y.; Cui, Y.Y.; Xu, Y.; Hora, T.; Zaveri, E.; Rodella, A.-S.; Bai, L.; Long, D. Spatial Downscaling of GRACE-Derived Groundwater Storage Changes across Diverse Climates and Human Interventions with Random Forests. J. Hydrol. 2024, 640, 131708. [Google Scholar] [CrossRef]
  39. Kalu, I.; Ndehedehe, C.E.; Ferreira, V.G.; Janardhanan, S.; Currell, M.; Kennard, M.J. Statistical Downscaling of GRACE Terrestrial Water Storage Changes Based on the Australian Water Outlook Model. Sci. Rep. 2024, 14, 10113. [Google Scholar] [CrossRef]
  40. Arshad, A.; Mirchi, A.; Taghvaeian, S.; AghaKouchak, A. Downscaled-GRACE Data Reveal Anthropogenic and Climate-Induced Water Storage Decline Across the Indus Basin. Water Resour. Res. 2024, 60, e2023WR035882. [Google Scholar] [CrossRef]
  41. Ali, S.; Ran, J.; Khorrami, B.; Wu, H.; Tariq, A.; Jehanzaib, M.; Khan, M.M.; Faisal, M. Downscaled GRACE/GRACE-FO Observations for Spatial and Temporal Monitoring of Groundwater Storage Variations at the Local Scale Using Machine Learning. Groundw. Sustain. Dev. 2024, 25, 101100. [Google Scholar] [CrossRef]
  42. Liu, C.; Zhang, Z.; Xu, C.; Zhang, W. Reconstructing Long-Term, High-Resolution Groundwater Storage Changes in the Songhua River Basin Using Supplemented GRACE and GRACE-FO Data. Remote Sens. 2024, 16, 4566. [Google Scholar] [CrossRef]
  43. Nannawo, A.S.; Lohani, T.K.; Eshete, A.A. Analyzing Groundwater Storage Anomalies in Data-Scarce Areas of Ethiopia’s Rift Valley Basin Using Artificial Neural Network. World Water Policy 2024, 10, 909–939. [Google Scholar] [CrossRef]
  44. Yin, G.; Park, J.; Yoshimura, K. Spatial Downscaling of GRACE Terrestrial Water Storage Anomalies for Drought and Flood Potential Assessment. J. Hydrol. 2025, 658, 133144. [Google Scholar] [CrossRef]
  45. Youssefi, F.; Soltani, S.S.; Ali, S.; Khorrami, B. Integrating Fully-Coupled Hydrological Modeling and Random Forest to Enhance Spatial Resolution of GRACE-Observed Water Storage Across the Rhine Basin. Nat. Resour. Res. 2025, 34, 2667–2684. [Google Scholar] [CrossRef]
  46. Hamou-Ali, Y.; Karmouda, N.; Mohsine, I.; Bouramtane, T.; Kacimi, I.; Tweed, S.; Tahiri, M.; Kassou, N.; El Bilali, A.; Chafki, O. Downscaling GRACE Total Water Storage Data Using Random Forest: A Three-Round Validation Approach under Drought Conditions. Front. Water 2025, 7, 1545821. [Google Scholar] [CrossRef]
  47. Cui, L.; Li, Y.; Zhong, B.; An, J.; Meng, J.; Guo, H.; Xu, C. Assessing the Impact of 2022 Extreme Drought on the Yangtze River Basin Using Downscaled GRACE/GRACE-FO Data Obtained by Partitioned Random Forest Algorithm. Int. J. Remote Sens. 2025, 46, 1219–1247. [Google Scholar] [CrossRef]
  48. Mikaeeli Ahamadabad, F.; Dastjerdi, P.A.; Nasseri, M.; Ahamadabad, F.M.; Dastjerdi, P.A.; Nasseri, M. Spatiotemporal GRACE TWS Downscaling Using Statistical and Machine Learning Methods: Random Forest, Area-to-Area Kriging, and Hybrid Methods. J. Hydrol. Reg. Stud. 2025, 62, 102885. [Google Scholar] [CrossRef]
  49. Ghaffari, Z.; Awawdeh, A.R.; Easson, G.; Yarbrough, L.D.; Heintzman, L.J. Evaluating an Ensemble-Based Machine Learning Approach for Groundwater Dynamics by Downscaling GRACE Data. Limnol. Rev. 2025, 25, 39. [Google Scholar] [CrossRef]
  50. Mohtaram, A.; Shafizadeh-Moghadam, H.; Ketabchi, H. A Flexible Multi-Scale Approach for Downscaling GRACE-Derived Groundwater Storage Anomaly Using LightGBM and Random Forest in the Tashk-Bakhtegan Basin, Iran. J. Hydrol. Reg. Stud. 2025, 57, 102086. [Google Scholar] [CrossRef]
  51. Kashani, A.; Safavi, H.R. Assessing Groundwater Drought in Iran Using GRACE Data and Machine Learning. Sci. Rep. 2025, 15, 14671. [Google Scholar] [CrossRef]
  52. Avila, L.; de Lavenne, A.; Ramos, M.-H.; Kollet, S. Estimation of Monthly Water Table Depth Anomalies Based on the Integration of GRACE and ERA5-Land with Large-Scale Simulations Using Random Forest and LSTM Networks. Water Resour. Manag. 2025, 39, 2899–2918. [Google Scholar] [CrossRef]
  53. Tian, B.; Yang, T.; Xu, S.; Wang, W.; Dou, B.; Tan, X.-C. Spatiotemporal Variation Characteristics of Groundwater Storage and Groundwater Resource Assessment in Jinan City Based on MGWR Model. China Rural. Water Hydropower 2025, 60–67. [Google Scholar] [CrossRef]
  54. Wu, C.; Lu, C.; Hu, J.; Liu, B.; Shu, L.; Zhang, Y. Machine Learning-Based Downscaling of GRACE Data to Enhance Assessment of Spatiotemporal Evolution of Coastal Plain Groundwater Storage. Water Resour. Manag. 2025, 39, 6377–6397. [Google Scholar] [CrossRef]
  55. Solovey, T.; Śliwińska-Bronowicz, J.; Janica, R.; Stradczuk, A.; Brzezińska, A. Temporal and Spatial Variability of Groundwater Storage Derived from Downscaled GRACE Data in the Transboundary Bug River Basin (Poland–Ukraine–Belarus Border Region). Sci. Total Environ. 2025, 1009, 181023. [Google Scholar] [CrossRef]
  56. Breiman, L. Random Forests. Mach. Learn. 2001, 45, 5–32. [Google Scholar] [CrossRef]
  57. Kratzert, F.; Klotz, D.; Brenner, C.; Schulz, K.; Herrnegger, M. Rainfall–Runoff Modelling Using Long Short-Term Memory (LSTM) Networks. Hydrol. Earth Syst. Sci. 2018, 22, 6005–6022. [Google Scholar] [CrossRef]
  58. Shen, C. A Transdisciplinary Review of Deep Learning Research and Its Relevance for Water Resources Scientists. Water Resour. Res. 2018, 54, 8558–8593. [Google Scholar] [CrossRef]
  59. Rodell, M.; Famiglietti, J.S.; Wiese, D.N.; Reager, J.T.; Beaudoing, H.K.; Landerer, F.W.; Lo, M.-H. Emerging Trends in Global Freshwater Availability. Nature 2018, 557, 651–659. [Google Scholar] [CrossRef]
  60. Nash, J.E.; Sutcliffe, J. V River Flow Forecasting through Conceptual Models Part I—A Discussion of Principles. J. Hydrol. 1970, 10, 282–290. [Google Scholar] [CrossRef]
  61. Chai, T.; Draxler, R.R. Root Mean Square Error (RMSE) or Mean Absolute Error (MAE)?—Arguments against Avoiding RMSE in the Literature. Geosci. Model Dev. 2014, 7, 1247–1250. [Google Scholar] [CrossRef]
  62. Ali, S.; Ran, J.; Luan, Y.; Khorrami, B.; Xiao, Y.; Tangdamrongsub, N. The GWR Model-Based Regional Downscaling of GRACE/GRACE-FO Derived Groundwater Storage to Investigate Local-Scale Variations in the North China Plain. Sci. Total Environ. 2024, 908, 168239. [Google Scholar] [CrossRef]
  63. Hu, L.; Wang, L.; Peng, Z.; Zhang, H. High-Resolution Groundwater Storage Anomalies in the Middle and Lower Yangtze River Basin of China Using Machine Learning Fusion of in-Situ Wells, Satellite Gravity and Hydrological Model. J. Environ. Manag. 2025, 375, 124322. [Google Scholar] [CrossRef]
  64. Memarian Sorkhabi, O.; Asgari, J.; Randhir, T.O. Monitoring Groundwater Storage Based on Satellite Gravimetry and Deep Learning. Nat. Resour. Res. 2023, 32, 1007–1020. [Google Scholar] [CrossRef]
  65. Wada, Y.; Van Beek, L.P.H.; Van Kempen, C.M.; Reckman, J.W.T.M.; Vasak, S.; Bierkens, M.F.P. Global Depletion of Groundwater Resources. Geophys. Res. Lett. 2010, 37. [Google Scholar] [CrossRef]
  66. Tapley, B.D.; Watkins, M.M.; Flechtner, F.; Reigber, C.; Bettadpur, S.; Rodell, M.; Sasgen, I.; Famiglietti, J.S.; Landerer, F.W.; Chambers, D.P.; et al. Contributions of GRACE to Understanding Climate Change. Nat. Clim. Change 2019, 5, 358–369. [Google Scholar] [CrossRef] [PubMed]
  67. Schumacher, M.; Forootan, E.; van Dijk, A.I.J.M.; Müller Schmied, H.; Crosbie, R.S.; Kusche, J.; Döll, P. Improving Drought Simulations within the Murray-Darling Basin by Combined Calibration/Assimilation of GRACE Data into the WaterGAP Global Hydrology Model. Remote Sens. Environ. 2018, 204, 212–228. [Google Scholar] [CrossRef]
  68. Al-Abadi, A.M.; Hassan, A.A.; Al-Moosawi, N.M.; Handhal, A.M.; Alzahrani, H.; Jabbar, F.K.; Anderson, N.L. Drought Susceptibility Mapping in Iraq Using GRACE/GRACE-FO, GLDAS, and Machine Learning Algorithms. Phys. Chem. Earth Parts A/B/C 2024, 134, 103583. [Google Scholar] [CrossRef]
  69. Ali, S.; Khorrami, B.; Jehanzaib, M.; Tariq, A.; Ajmal, M.; Arshad, A.; Shafeeque, M.; Dilawar, A.; Basit, I.; Zhang, L.; et al. Spatial Downscaling of GRACE Data Based on XGBoost Model for Improved Understanding of Hydrological Droughts in the Indus Basin Irrigation System (IBIS). Remote Sens. 2023, 15, 873. [Google Scholar] [CrossRef]
  70. Arshad, A.; Mirchi, A.; Samimi, M.; Ahmad, B. Combining Downscaled-GRACE Data with SWAT to Improve the Estimation of Groundwater Storage and Depletion Variations in the Irrigated Indus Basin (IIB). Sci. Total Environ. 2022, 838, 156044. [Google Scholar] [CrossRef] [PubMed]
  71. Arshad, A.; Shafeeque, M.; Tran, T.N.D.; Mirchi, A.; Xiang, Z.; He, C.; AghaKouchak, A.; Besnier, J.; Rahman, M.M. Multi-Model Ensemble Machine Learning-Based Downscaling and Projection of GRACE Data Reveals Groundwater Decline in Saudi Arabia throughout the 21st Century. J. Hydrol. Reg. Stud. 2025, 60, 102552. [Google Scholar] [CrossRef]
  72. Awawdeh, A.R.M.; Yasarer, H.; Ghaffari, Z.; Yarbrough, L.D. Downscaling GRACE Data for Improved Groundwater Forecasting Using Artificial Neural Networks. Civ. Eng. J. 2025, 11, 406–419. [Google Scholar] [CrossRef]
  73. Chen, L.; He, Q.; Liu, K.; Li, J.; Jing, C. Downscaling of GRACE-Derived Groundwater Storage Based on the Random Forest Model. Remote Sens. 2019, 11, 2979. [Google Scholar] [CrossRef]
  74. Chen, Z.; Zheng, W.; Yin, W.; Li, X.; Zhang, G.; Zhang, J. Improving the Spatial Resolution of Grace-Derived Terrestrial Water Storage Changes in Small Areas Using the Machine Learning Spatial Downscaling Method. Remote Sens. 2021, 13, 4760. [Google Scholar] [CrossRef]
  75. Cui, B.; Xue, D.; Gui, D.; Liu, Q.; Abd-Elmabod, S.K.; Chen, X.; Goethals, P.; De Maeyer, P. Downscaled GRACE Data Reveals Anthropogenic Dominance in Groundwater Storage Decline across China’s Oases. Ecol. Indic. 2025, 179, 114209. [Google Scholar] [CrossRef]
  76. Djima, R.; Wan, X.; Fiifi, R.; Sidi, A.B.; Bouko, D. The Egyptian Journal of Remote Sensing and Space Sciences Assessing Groundwater Storage Variations in the Volta River Basin Combining Remote Sensing Tools and Machine Learning Downscaling Techniques. Egypt. J. Remote Sens. Sp. Sci. 2025, 28, 383–394. [Google Scholar] [CrossRef]
  77. Gaffoor, Z.; Gritzman, A.; Pietersen, K.; Jovanovic, N.; Bagula, A.; Kanyerere, T. An Autoregressive Machine Learning Approach to Forecast High-Resolution Groundwater-Level Anomalies in the Ramotswa/North West/Gauteng Dolomite Aquifers of Southern Africa. Hydrogeol. J. 2022, 30, 575–600. [Google Scholar] [CrossRef]
  78. Gao, G.; Zhao, J.; Wang, J.; Zhao, G.; Chen, J.; Li, Z. Spatiotemporal Variation and Driving Analysis of Groundwater in the Tibetan Plateau Based on GRACE Downscaling Data. Water 2022, 14, 3302. [Google Scholar] [CrossRef]
  79. Gorugantula, S.S.; Kambhammettu, B.P. Sequential Downscaling of GRACE Products to Map Groundwater Level Changes in Krishna River Basin. Hydrol. Sci. J. 2022, 67, 1846–1859. [Google Scholar] [CrossRef]
  80. He, H.; Yang, K.; Wang, S.; Petrosians, H.A.; Liu, M.; Li, J.J.; Marcato Junior, J.J.; Goncalves, W.N.; Wang, L.; Li, J.J.; et al. Deep Learning Approaches to Spatial Downscaling of GRACE Terrestrial Water Storage Products Using EALCO Model Over Canada. Can. J. Remote Sens. 2021, 47, 657–675. [Google Scholar] [CrossRef]
  81. Kalu, I.; Ndehedehe, C.E.; Ferreira, V.G.; Kennard, M.J. Machine Learning Assessment of Hydrological Model Performance under Localized Water Storage Changes through Downscaling. J. Hydrol. 2024, 628, 130597. [Google Scholar] [CrossRef]
  82. Khorrami, B.; Pirasteh, S.; Ali, S.; Gungor, O.; Vaheddoost, B. Statistical Downscaling of GRACE TWSA Estimates to a 1-Km Spatial Resolution for a Local-Scale Surveillance of Flooding Potential. J. Hydrol. 2023, 624, 129929. [Google Scholar] [CrossRef]
  83. Kumar, R.; Gaur, S.; Soni, P.; Maurya, P.; Ohri, A. HRU-Based Downscaling of GRACE-TWS to Quantify the Hydrogeological Fluxes and Specific Yield in the Lower Middle Ganga Basin. J. Hydrol. 2024, 639, 131591. [Google Scholar] [CrossRef]
  84. Li, P.; Yu, H.; Zhou, P.; Zhang, P.; Wang, R. Downscaling Inversion of GRACE-Derived Groundwater Storage Changes Based on Ensemble Learning Downscaling Inversion of GRACE-Derived Groundwater Storage Changes Based on Ensemble Learning. Int. J. Digit. Earth 2023, 16, 2998–3022. [Google Scholar] [CrossRef]
  85. Liang, Q.; Hao, X.; Ci, M.; Yuan, M.; Di, Y.; Sun, F.; Wang, C.; Zhang, J.; Fan, X.; Xiong, H. Disentangling and Integrating Spatiotemporal Features: Deep Learning-Based Downscaling of Groundwater Storage Anomalies from GRACE and GRACE-FO Satellites. J. Hydrol. Reg. Stud. 2025, 62, 102982. [Google Scholar] [CrossRef]
  86. Milewski, A.M.; Thomas, M.B.; Seyoum, W.M.; Rasmussen, T.C. Spatial Downscaling of GRACE TWSA Data to Identify Spatiotemporal Groundwater Level Trends in the Upper Floridan Aquifer, Georgia, USA. Remote Sens. 2019, 11, 2756. [Google Scholar] [CrossRef]
  87. Mojtaba, S.; Mohammad, M.; Kavianpour, R. A Non–Stationary Downscaling and Gap–Filling Approach for GRACE / GRACE–FO Data under Climatic and Anthropogenic Influences. Appl. Water Sci. 2025, 15, 1–15. [Google Scholar] [CrossRef]
  88. Nourani, V.; Paknezhad, N.J.; Mohammadisepasi, S.; Zhang, Y. Regionalization of GRACE Data in Shorelines by Ensemble of Artificial Intelligence Methods. J. Hydrol. 2024, 636, 131268. [Google Scholar] [CrossRef]
  89. Rahaman, M.M.; Thakur, B.; Kalra, A.; Li, R.; Maheshwari, P. Estimating High-Resolution Groundwater Storage from GRACE: A Random Forest Approach. Environments 2019, 6, 63. [Google Scholar] [CrossRef]
  90. Raza, A.; Ryo, M.; Ghazaryan, G.; Baatz, R.; Main-knorn, M.; Inforsato, L.; Nendel, C. International Journal of Applied Earth Observation and Geoinformation Predicting Regional-Scale Groundwater Levels at High Spatial Resolution Using Spatial Random Forest Models. Int. J. Appl. Earth Obs. Geoinf. 2025, 144, 104918. [Google Scholar] [CrossRef]
  91. Sahour, H.; Sultan, M.; Vazifedan, M.; Abdelmohsen, K.; Karki, S.; Yellich, J.A.; Gebremichael, E.; Alshehri, F.; Elbayoumi, T.M. Statistical Applications to Downscale GRACE-Derived Terrestrial Water Storage Data and to Fill Temporal Gaps. Remote Sens. 2020, 12, 533. [Google Scholar] [CrossRef]
  92. Satizábal-Alarcón, D.A.; Suhogusoff, A.; Ferrari, L.C. Characterization of Groundwater Storage Changes in the Amazon River Basin Based on Downscaling of GRACE/GRACE-FO Data with Machine Learning Models. Sci. Total Environ. 2024, 912, 168958. [Google Scholar] [CrossRef] [PubMed]
  93. Wang, J.; Xu, D.; Li, H. Constructing GRACE-Based 1 Km Resolution Groundwater Storage Anomalies in Arid Regions Using an Improved Machine Learning Downscaling Method: A Case Study in Alxa League, China. Remote Sens. 2023, 15, 2913. [Google Scholar] [CrossRef]
  94. Xue, H.; Wang, H.; Dong, G.; Li, Z. Spatial Downscaling of GRACE Groundwater Storage Based on DTW Distance Clustering and an Analysis of Its Driving Factors. Remote Sens. 2025, 17, 2526. [Google Scholar] [CrossRef]
  95. Yang, R.; Zhong, Y.; Zhang, X.; Maimaitituersun, A.; Ju, X. A Comparative Study of Downscaling Methods for Groundwater Based on GRACE Data Using RFR and GWR Models in Jiangsu Province, China. Remote Sens. 2025, 17, 493. [Google Scholar] [CrossRef]
  96. Yuan, Y.; Shen, D.; Cao, Y.; Wang, X.; Zhang, B.; Dong, H. An Ensemble Machine Learning Approach for High-Resolution Estimation of Groundwater Storage Anomalies. Water 2025, 17, 1445. [Google Scholar] [CrossRef]
  97. Zhang, G.; Xu, T.; Yin, W.; Bateni, S.M.; Jun, C.; Kim, D.; Liu, S.; Xu, Z.; Ming, W.; Wang, J. A Machine Learning Downscaling Framework Based on a Physically Constrained Sliding Window Technique for Improving Resolution of Global Water Storage Anomaly. Remote Sens. Environ. 2024, 313, 114359. [Google Scholar] [CrossRef]
  98. Zhang, J.; Liu, K.; Wang, M. Downscaling Groundwater Storage Data in China to a 1-Km Resolution Using Machine Learning Methods. Remote Sens. 2021, 13, 523. [Google Scholar] [CrossRef]
Hydrology 13 00135 g001
Figure 1. Worldwide distribution of water resources.
Figure 1. Worldwide distribution of water resources.
Hydrology 13 00135 g001
Hydrology 13 00135 g002
Figure 2. PRISMA methodological framework.
Figure 2. PRISMA methodological framework.
Hydrology 13 00135 g002
Hydrology 13 00135 g003
Figure 3. The PRISMA flow diagram summarizes the study selection procedure, including duplicate removal, title and abstract cleaning, and the final inclusion of relevant studies based on established eligibility criteria.
Figure 3. The PRISMA flow diagram summarizes the study selection procedure, including duplicate removal, title and abstract cleaning, and the final inclusion of relevant studies based on established eligibility criteria.
Hydrology 13 00135 g003
Hydrology 13 00135 g004
Figure 4. The numbers (occurrences) of evaluation metrics used in the reviewed studies: (a) Frequency of ML methods used; (b) Percentage of ML used. Note: RF = Random Forest; GB = Gradient Boosting; ANN = Artificial Neural Network; SVM = Support Vector Machine; SVR = Support Vector Regression; LSTM = Long Short-Term Memory; MLP = Multilayer Perceptron; MLR = Multiple Linear Regression.
Figure 4. The numbers (occurrences) of evaluation metrics used in the reviewed studies: (a) Frequency of ML methods used; (b) Percentage of ML used. Note: RF = Random Forest; GB = Gradient Boosting; ANN = Artificial Neural Network; SVM = Support Vector Machine; SVR = Support Vector Regression; LSTM = Long Short-Term Memory; MLP = Multilayer Perceptron; MLR = Multiple Linear Regression.
Hydrology 13 00135 g004
Hydrology 13 00135 g005
Figure 5. The numbers (occurrences) of input variables used in the reviewed studies.
Figure 5. The numbers (occurrences) of input variables used in the reviewed studies.
Hydrology 13 00135 g005
Hydrology 13 00135 g006
Figure 6. The applied performance measures (PMs): (a) An analysis of individual PMs; (b) Combination summary of PMs used across studies.
Figure 6. The applied performance measures (PMs): (a) An analysis of individual PMs; (b) Combination summary of PMs used across studies.
Hydrology 13 00135 g006
Hydrology 13 00135 g007
Figure 7. Numbers of published papers in TWSA and GWSA downscaling over time.
Figure 7. Numbers of published papers in TWSA and GWSA downscaling over time.
Hydrology 13 00135 g007
Hydrology 13 00135 g008
Figure 8. Overview showing number of publications per country in reviewed publications.
Figure 8. Overview showing number of publications per country in reviewed publications.
Hydrology 13 00135 g008
Hydrology 13 00135 g009
Figure 9. Geographical distribution of reviewed publications.
Figure 9. Geographical distribution of reviewed publications.
Hydrology 13 00135 g009
Hydrology 13 00135 g010
Figure 10. Journals in which reviewed papers were published.
Figure 10. Journals in which reviewed papers were published.
Hydrology 13 00135 g010
Table 1. Publications employed GRACE data downscaling frameworks, including implemented predictors and evaluation metrics.
Table 1. Publications employed GRACE data downscaling frameworks, including implemented predictors and evaluation metrics.
ReferenceStudy AreaNative ResolutionDownscaled ResolutionML MethodInputsEvaluation Metrics
[17]Assiniboine Delta Aquifer, Canada1°2.5 kmANN GW level CC, RMSE
[18]Maharashtra State, India1°0.125°ANN SM, CWSCC
[19]North China Plain, China1°0.25°GB Pr, Qs, ET, SM, SWE, TWSA Pearson correlation, RMSE, MAE, NSE
[20]Indus Basin, Pakistan1°0.25°ANN, RFSM, Pr, ET, DEM, slope, aspect, Qs, CWS, TCC, RMSE, MAE, NSE
[10]Four Hydrogeological basins, India1°0.25°MLR, RF bare soil evaporation, CWS, canopy water evaporation, ET, Pr, LST, SM, Qs, QssRMSE, CC
[21]160 global river basins3°0.5° PLSRPr, ET, QsRMS
[22]Tarim River Basin, China1°1 km RFLST, NDVICC, R2, RMSE, MAE, NSE
[23]Indus Basin, Pakistan1°0.25°GB, RF, SVM, ANN Pr, DEM, Slope, Aspect, SM, T, ET, Qs, CWS Pearson correlation, NSE, RMSE, MAE
[24]Haila River Basin, China1°0.25°MLR, RF ET, LST, NDVI, T, SM, SWE Pearson correlation, NSE, RMSE
[25]Fractured crystalline aquifer, India3°0.5°RF NDVI, Pr, surface SM Pearson correlation, R2, RMSE
[26]Haiha River Basin, China0.5°0.05°RF T, ET, NDVI, Pr, SM, LST, SWE, Qs, PWC Pearson correlation, RMSE
[27]North China Plain, China0.25°5 kmRFSM, Pr, T, slope, wells data, Saturated hydraulic conductivity (K), CC, NSE, RMSE
[28]State of Mississippi, USA0.5°1 km Keras dense NN, GB, MLP, KNN GRACE, TerraClimate grid cells, CHIRPS grid cells R2, NSE, RMSE, Pearson correlation, Spearman correlation
[29]Western Mediterranean Basin, Turkey0.5°10 km RF SM, SWS, T, ET, Pr, Qs, elevationCC, RMSE
[30]Central Valley, California, USA3°0.05°RF Pr, T, SM, ET, Slope, Texture, Saturated hydraulic conductivityNSE, CC, RMSE
[31]Lake Urmia catchment, Iran0.25°0.1°RF, SVR, MLPSM, Pr, NDVI, Qs, ET, LST, SWE NSE, Pearson correlation, RMSE
[32]Mississippi Alluvial Plain, USA0.5°5 km RF T, DEM, Pr, ET, NDVI, soil type, land cover, Aquifer thickness MAE, RMSE, R2
[33]Western Anatolian Basin, Türkiye 50 km10 km RFDEM, SM, snow water, rainfall, Qs, ETCC, R2, MAE, RMSE
[34]National scale, Iraq1.0° 0.1° RF, SVM, ANN Pr, ET, Qs, Qss, surface water thickness, SMNRMSE, NSE, md, R2, KGE
[35]Semirom Basin and its neighbors, Iran0.5°0.25°SP-SVMIn situ Pr, in situ E, Eta, Pr, NDVI, EVI, SM, SWE, RZSM, CSWR2, RMSE, MAE, Bias
[36]Iguerounzar basin, Morocco0.25°9 kmRF, SVRPr, ET, SM, Qs, NDVI, LST, DEM, Slope, Soil map, LULCNSE, RMSE, R2, PBIAS
[37]Barotse catchment, Zambia0.25°5 kmXGBoost and RFPr, ET, SM, LST, NDVI, EVINSE, R2, MAE, RMSE
[38]Central Yunnan, China1°0.1°RFPr, LST, NDVI, EVI, ET, SMCC, NSE, RMSE, MAE
[39]Cambrian Limestone Aquifer (CLA), Australia1.0°0.05°SVMPr, ET, Qs, GWLRMSE, NSE, MAE
[40]Transboundary Indus Basin (Pakistan, India, China, Afghanistan)0.25°1 kmEF, RFgwelevation, ET, SM, Pr, NDVI, population density, CWS, SWSNSE, R2, RMSE, KGE, CC
[41]Upper Indus Plain Aquifer (Pakistan)0.5°0.1°XGBoostSMS, ET, T, Qs, rainfall, DEM, slope, aspect, GWLPearson correlation, NSE, RMSE, PBIAS
[42]Songhua River Basin, NE China0.25°1 kmGWR, RFPr, NDVI, ETa, SM, TNSE, RMSE, CC
[43]Rift Valley Basin, Ethiopia1°0.25°ANNPr, ET, T, SM, CWS, SWSRMSE, CC, NSE
[44]Texas–Gulf Basin, USA3°0.5°LSTM, RFSM, SWE, canopy water, Qs, latent heat flux, GW, Pr, TR, MAE, ubRMSE,
[45]Rhine Basin, Central Europe (9 countries; focus on German part)0.25°0.1°RFPr, T, SMS, ET, Qs, SWE, DEM, ParFlow–CLM (PFC) fully coupled hydrological model outputsPearson correlation, RMSE, NSE
[46]Northern Morocco1°1 kmRFPr, NDVI, LST, ET, DEM, NDSINSE, RMSE, MAE, R2, CC
[47]Yangtze River Basin, China1°0.1°RFPRE, ET, SM, SWE, PCW, Qs, STNSE, RMSE, CC, LCCC
[48]Karkheh & Karoon basins, Iran0.5°0.1°RFSM, T, air pressure, longwave radiation, Qs, ET, streamflow, Pr, snow coverCC, NSE, KGE, RMSE, MAE
[49]Mississippi Delta, USA0.5°1 kmRFPr, T, NDVI, SM, Qs, aquifer thicknessR2, MAE, RMSE
[50]Tashk–Bakhtegan–Maharlo basin, Iran1°1 kmLightGBM, RFSMS, SWE, CWS, Qs, ET, LST, NDVI, DEM, PrNSE, CC, RMSE
[51]Six major basins of Iran, Iran0.5°0.25°XGBoostSM, SWE, CWS, Qs, T, ET, Pr, DEM, Teleconnections data, Canadian Earth System Model (CanESM5) dataRMSE, CC, MAE
[52]Seine River Basin, France0.25°0.11°RF and LSTMPr, T, EKGE, RMSE, Pearson correlation
[53]North China Plain, China3°0.25°RFPr, ET, SM, T, CWS, Qs, SWE, SAR dataCC, NSE, RMSE
[54]Hang-Jia-Hu coastal plain, Zhejiang, China0.5°4 kmRF, XGBoost, LightGBMTerraClimate vars, DEM, SM, CW, Qs, snow, GWCC, R2, NRMSE
[55]Bug River Basin, at the Poland–Ukraine–Belarus border0.25°0.1°RFSM, SWS, CWS, Pr, ET, Qs, GW levelRMSE, CC
[35]Semirom Basin and its neighbors, Iran0.5°0.25°SP-SVMIn situ Pr, in situ E, Eta, Pr, NDVI, EVI, SM, SWE, RZSM, CSWR2, RMSE, MAE, Bias
[36]Iguerounzar basin, Morocco0.25°9 kmRF, SVRPr, ET, SM, Qs, NDVI, LST, DEM, Slope, Soil map, LULCNSE, RMSE, R2, PBIAS
[37]Barotse catchment, Zambia0.25°5 kmXGBoost and RFPr, ET, SM, LST, NDVI, EVINSE, R2, MAE, RMSE
[38]Central Yunnan, China1°0.1°RFPr, LST, NDVI, EVI, ET, SMCC, NSE, RMSE, MAE
[39]Cambrian Limestone Aquifer (CLA), Australia1.0°0.05°SVMPr, ET, Qs, GWLRMSE, NSE, MAE
[40]Transboundary Indus Basin (Pakistan, India, China, Afghanistan)0.25°1 kmEF, RFgwelevation, ET, SM, Pr, NDVI, population density, CWS, SWSNSE, R2, RMSE, KGE, CC
[41]Upper Indus Plain Aquifer (Pakistan)0.5°0.1°XGBoostSMS, ET, T, Qs, rainfall, DEM, slope, aspect, GWLPearson correlation, NSE, RMSE, PBIAS
[42]Songhua River Basin, NE China0.25°1 kmGWR, RFPr, NDVI, ETa, SM, TNSE, RMSE, CC
[43]Rift Valley Basin, Ethiopia1°0.25°ANNPr, ET, T, SM, CWS, SWSRMSE, CC, NSE
[44]Texas–Gulf Basin, USA3°0.5° LSTM, RF SM, SWE, canopy water, Qs, latent heat flux, GW, Pr, TR, MAE, ubRMSE,
[45]Rhine Basin, Central Europe (9 countries; focus on German part)0.25° 0.1° RFPr, T, SMS, ET, Qs, SWE, DEM, ParFlow–CLM (PFC) fully coupled hydrological model outputsPearson correlation, RMSE, NSE
[46]Northern Morocco1°1 kmRFPr, NDVI, LST, ET, DEM, NDSI NSE, RMSE, MAE, R2, CC
[47]Yangtze River Basin, China1°0.1°RFPRE, ET, SM, SWE, PCW, Qs, ST NSE, RMSE, CC, LCCC
[48]Karkheh & Karoon basins, Iran0.5°0.1°RFSM, T, air pressure, longwave radiation, Qs, ET, streamflow, Pr, snow coverCC, NSE, KGE, RMSE, MAE
[49]Mississippi Delta, USA0.5°1 km RFPr, T, NDVI, SM, Qs, aquifer thicknessR2, MAE, RMSE
[50]Tashk–Bakhtegan–Maharlo basin, Iran1°1 km LightGBM, RFSMS, SWE, CWS, Qs, ET, LST, NDVI, DEM, PrNSE, CC, RMSE
[51]Six major basins of Iran, Iran0.5°0.25° XGBoost SM, SWE, CWS, Qs, T, ET, Pr, DEM, Teleconnections data, Canadian Earth System Model (CanESM5) data RMSE, CC, MAE
[52]Seine River Basin, France0.25°0.11° RF and LSTM Pr, T, EKGE, RMSE, Pearson correlation
[53]North China Plain, China3°0.25°RFPr, ET, SM, T, CWS, Qs, SWE, SAR dataCC, NSE, RMSE
[54]Hang-Jia-Hu coastal plain, Zhejiang, China0.5°4 km RF, XGBoost, LightGBMTerraClimate vars, DEM, SM, CW, Qs, snow, GWCC, R2, NRMSE
[55]Bug River Basin, at the Poland–Ukraine–Belarus border0.25°0.1° RFSM, SWS, CWS, Pr, ET, Qs, GW levelRMSE, CC
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Al Nadabi, M.S.; El-Diasty, M.; Etri, T.; Nikoo, M.R. GRACE Downscaling and Machine Learning Models for Groundwater Prediction: A Systematic Review. Hydrology 2026, 13, 135. https://doi.org/10.3390/hydrology13050135

AMA Style

Al Nadabi MS, El-Diasty M, Etri T, Nikoo MR. GRACE Downscaling and Machine Learning Models for Groundwater Prediction: A Systematic Review. Hydrology. 2026; 13(5):135. https://doi.org/10.3390/hydrology13050135

Chicago/Turabian Style

Al Nadabi, Mohammed S., Mohammed El-Diasty, Talal Etri, and Mohammad Reza Nikoo. 2026. "GRACE Downscaling and Machine Learning Models for Groundwater Prediction: A Systematic Review" Hydrology 13, no. 5: 135. https://doi.org/10.3390/hydrology13050135

APA Style

Al Nadabi, M. S., El-Diasty, M., Etri, T., & Nikoo, M. R. (2026). GRACE Downscaling and Machine Learning Models for Groundwater Prediction: A Systematic Review. Hydrology, 13(5), 135. https://doi.org/10.3390/hydrology13050135

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop