Since 2002, the GRACE has supplied a highly valued dataset, which allows the study of TWS over large river basins globally. Shamsudduha et al. [105] used GRACE datasets for a long time period and found that for the recent time period (2003 to 2007), groundwater-storage changes correlate well with in situ borehole records in the Bengal Basin in Bangladesh. For the Indus Basin, the variation in groundwater storage at numerous spatial scales has been studied by Iqbal et al. [106] with GRACE data. They found that the groundwater storage changes through GRACE-based estimation is reasonably sufficient for the provision of monthly groundwater-storage changes.

Variations in TWS can be acquired through the GRACE data. After the changes in groundwater, the variation of soil moisture is normally the highest factor in TWS changes and intermediate zone storage [107]. Strassberg et al. [108] found a robust link between the sum of soil moisture and groundwater storage with TWS by means of GRACE data. Before GRACE satellite, estimates of TWS changes were used to derive from the in situ data of soil moisture and snow. These approaches cover analyzing the variations in groundwater, soil water, surface water, and snow through utilizing a long time series [107].

Since the GRACE satellite was launched in 2002, a great deal of researchers has conducted the studies of groundwater using GRACE satellite data. According to the research articles published in various journals (source: https://scholar.google.com/, dated 18 January 2022), an annual frequency graph of articles has been developed to envisage the work carried out using GRACE data (refer to Figure 3).

GRACE provides useful information about global groundwater depletion. The shallow GWS declines faster than the deep GWS [109]. Globally, GRACE could be the only hope for the assessment of groundwater reduction [104]. GRACE data are useful for the estimation of regional groundwater monitoring on monthly to seasonal scale. Yeh et al. [110] confirmed the ability of GRACE to simulate regional groundwater-storage changes from monthly to seasonal scale for an area of about 200,000 km² area in Illinois. It also provides reasonable results for the heterogeneous GWS variations as observed by Huang et al. [109]. Döll et al. [111] conducted groundwater storage monitoring for the Mississippi Basin. However, the study failed to obtain reliable results to monitor the total water storage due to water withdrawals at the whole scale of the basin. GRACE data were applied in global hydrological investigations, including water storage change evaluation, groundwater and evapotranspiration retrieving, drought analysis, and glacier response to global change [112]. In another study, complementary relationship (CR) approach was used to assess the evapotranspiration and a water balance method to estimate the runoff and base flow in the Conterminous United States (CONUS), which demonstrates how to use it with limited parameterization and datasets [113].

There are many GRACE-based studies of groundwater that showed substantial aquifer depletion for the large regions, for example, the North China, the northwest India, the Middle East, the Murray Darling Basin of Australia, and California’s Central Valley aquifers of the USA [17,101,114,115,116,117]. The shrinking phenomena of groundwater-fed lakes have also been monitored through GRACE and GLDAS [16]. Sun et al. [118] developed a method for estimation of aquifer storage parameters (specific yield and storability) utilizing the GRACE data for the connected aquifers of Edwards-Trinity Plateau and Pecos Valley in central Texas. Frappart and Ramillien [60] also found that these data were used to evaluate the parameters; the specific yield, for example, is used to relate groundwater level to storage or to describe the indices of groundwater depletion and stress. In another GRACE-GLDAS study, it was found that high consistency exists between GRACE observations and changes in water wells [119].

For groundwater management, the usage of GRACE data at a local scale has been inadequate due to in-built uncertainties in GRACE data and problems in disintegrating several TWS components.

Artificial Neural Network (ANN) models can be developed to foresee the variations of groundwater levels directly by means of a gridded GRACE product and other openly existing hydro-meteorological datasets. ANN is used as a statistical downscaling method, which is extensively used in streamflow projecting and management of water resources [120,121,122,123,124,125,126]. A trained ANN model is comparatively strong to data noise and is quite suitable to support real-time decision making [127,128]. Ensemble ANN models can be utilized to forecast monthly and seasonal groundwater-level changes for various wells situated in certain areas.

Because of the GRACE data, the ANN models forecast water-level variations instead of absolute water levels [128]. By combining GRACE, TRMM, and GLDAS, ANN predicts and evaluates groundwater drought [129]. Gemitzi et al. [130] used ANN to perform statistical downscaling of GRACE data using local meteorological data. A typical application process of ANN is shown in Figure 4.

Srivastava and Bhattacharya integrated numerous thematic plots on the GIS platform based on Saaty’s Analytical Hierarchy Process (AHP) with the Groundwater Potential Index (GWPI) values (0.175 to 0.940) to produce a good match with the resistivity and pumping test data [92]. Similarly, AHP was used by Mallick et al. [41] to find the most important contributing features to delineate the groundwater potential zones [131].

The Global Land Data Assimilation System (GLDAS) is a worldwide, high-resolution, offline terrestrial modeling system that integrates satellite and ground-based observations to produce optimum fields of land surface conditions and fluxes in near-real time.

It produces a sequences of worldwide land surface conditions (for example, soil moisture and surface temperature) and fluxes (for example, evaporation and sensible heat flux) products that were simulated by four Land Surface Models (LSM), that is, VIC, Mosaic, CLAM, and Noah.

Several current and proposed weather and climate prediction, water resources applications, and water-cycle investigations rely on GLDAS output. GLDAS runs globally and on user defined domains at resolutions ranging from 2.5° to 1 km. The GRACE and GLDAS simulations provide precise and reliable datasets that are used to describe the variations of groundwater storage regionally [132]. The assessed groundwater variations by GRACE-GLDAS are usually in line with in situ observations [22,97,98,109,110,132,133].

WGHM calculates timeseries of surface and subsurface runoff, groundwater recharge, and river flows along with water storage variations in canopy, snow, soil, groundwater, rivers, lakes, and wetlands. Therefore, it can quantify the whole water resources along with the groundwater resources. Institute of Physical Geography (IPG) developed WGHM to provide freshwater fluxes with storage data on the land surface globally [134,135]. The spatial resolution of the model is 0.5° × 0.5°, and computations are carried out on daily basis, whereas the output is analyzed on a monthly time step. The WGHM considers the human-induced water consumption along with the daily groundwater simulations [134]. Figure 5 explains the possible approaches to incorporated GLDAS and WGHM data to obtain GWS from the GRACE data. In this figure, it can be seen that TWS is obtained through GRACE satellite, whereas both GLDAS and WGHM give surface water, soil moisture, snow water equivalent, and groundwater storage. Final groundwater storage profiles are obtained through the interaction of GRACE with GLDAS and GRACE with WGHM, which can be compared with in situ groundwater measurements based upon the statistical selection criteria as given in the figure.