the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 04 Aug 2025
On the efficacy of downscaled GRACE total water storage products
Abstract. The Gravity Recovery and Climate Experiment (GRACE) satellite mission provides estimates of total water storage anomalies (TWSA) in terms of “mascons” (mass concentration blocks), representing integrated changes in surface water, soil moisture, and groundwater. However, the coarse spatial resolution of GRACE mascons (~3°) limits its utility for regional-scale hydrological analysis. Although several downscaling methods have been proposed to improve resolution, none have been comprehensively validated against in-situ observations. In this study, we are validating both native and downscaled GRACE TWSA products using well-based in-situ groundwater observations across India. Furthermore, we develop an improved downscaling method by modifying the approach of Vishwakarma et al. (2021), i.e., incorporating mass conservation constraints at the native GRACE resolution instead of catchment scale (Vishwakarma et al., 2021) to better capture spatial variability at a finer 0.5° grid scale. We assessed the efficacy of our downscaled product and two existing downscaled GRACE products (Vishwakarma et al., 2021; Gao and Soja, 2024) using performance metrics such as gain in correlation (r) and gain in root-mean-square error (RMSE) relative to native GRACE product. The entire India is covered by 50 mascons. Out of these 50 mascons groundwater observations are available only at 22 mascons. Our modified approach shows improved performance across 12 mascons in India, with median gains in r ranging from 0.58 % to 8.84 % and gain in RMSE improvements from 0.24 % to 23.59 %. Whereas, the Vishwakarma et al. (2021) and Gao and Soja (2024) methods perform well in only 3 and 10 mascons, respectively. These results demonstrate that our enhanced downscaling approach provides more accurate and spatially resolved estimates of groundwater storage changes, offering a valuable tool for regional water resource assessments in India. Further, the downscaled GRACE product developed in this study (Mascon wise Mass Conservation product) is publicly accessible at https://figshare.com/articles/dataset/GRACE_downscaled_TWSA_product_using_Mascon_wise_Mass_Conservation/29196617.
- Preprint
(4084 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
- RC1: 'Comment on egusphere-2025-2888', Anonymous Referee #1, 22 Sep 2025
-
RC2: 'Comment on egusphere-2025-2888', Sylvain Ferrant, 30 Sep 2025
General comments:Strengths:
The manuscript by Suryawanshi et al. represents a strong effort to develop and validate a downscaling method for GRACE total water storage (TWS) anomalies over India. The authors propose a modified approach, called "Mascon-wise Mass Conserved" (MMC), building upon the methodology of Vishwakarma et al. (2021). The primary innovation lies in applying a mass conservation constraint at the native GRACE mascon scale (~3°x3°), rather than at the coarser catchment scale used in previous studies. A notable strength of the study is the large-scale validation framework, which integrates both statistical modeling and data assimilation within a hydrological context. The authors also compare MMC with previously published downscaling approaches, namely CMC (Vishwakarma et al., 2021) and DL (Gou & Soja, 2024). For this evaluation, they used the temporal "gain" metric introduced in a previous GRACE downscaling study in India (Pascal et al., 2022), which represents a positive step toward standardizing validation practices. The public release of the resulting dataset is also a valuable contribution to the community.Weaknesses:
Despite these strengths, my principal concern relies on the hydrological validity of the downscaled product released. The manuscript lacks an appropriate discussion of the different hydrogeological contexts throughout India: huge irrigation, hydrogeological diversity, large dams and rivers, and small surface reservoirs simultaneously filling up with the monsoon and emptying with the dry season.
Missing downscaling methods and validation state of the art:
The introduction does not adequately position the study relative to the state of the art, notably by failing to cite Pascal et al. (2022) as a framework for spatial and temporal validation of downscaling approaches in this specific Indian highly irrigated context. Missing downscaling studies in India should be discussed (Jyolsna et al., 2021 and Karunakalage et al., 2021). Similarly, the manuscript does not incorporate the full spatial and temporal validation approach proposed in Pascal et al., 2022. Should be justified or discussed somewhere.Missing Table 2 Specific Yield of aquifers:
Methodologically, the transformation of groundwater levels into groundwater storage (GWS) relies on specific yield, yet Table 2, which should detail the assigned specific yield values for different aquifer types, is missing, limiting transparency and reproducibility. The choice of eight clusters for k-means clustering is arbitrary and unsubstantiated, raising concerns about the spatial reliability of the reference dataset (Ref-GWSC). The assigned specific yield values appear disconnected from geologically coherent units, and the exclusive use of temporal gain for validation neglects the assessment of spatial variability within mascons, which is a fundamental objective of downscaling.Surface water fluctuation neglected:
The validation relies on well-known Central Ground Water Board groundwater levels across 22 mascons, showing that native GRACE explains only 56% of in situ GWS variability. This implies that the remaining variability arises from surface and soil water storage, a well-documented phenomenon in both the Ganges-Brahmaputra basin (Salameh et al., 2017) and southern India, particularly Telangana. In Telangana, the cumulative capacity of large reservoirs on main rivers is estimated at 113 mm of equivalent water height, and small upstream reservoirs contribute an additional 30 mm, representing approximately 24% of the TWS signal over the period 2002–2021 (Pascal et al., 2021). This potential reservoir capacity of 143 mm represents about 24% of the annual GRACE TWS fluctuation in the area during 2002–2021 (600 mm). However, the authors interpretation that reservoirs are “rarely simultaneously full” is misleading: under the monsoonal regime, filling generally occurs simultaneously during the wet season and drawdown during the dry season as well. The decision to neglect this contribution can only be justified on practical grounds, namely the difficulty of obtaining reliable regional data on the filling dynamics of both large and small reservoirs, but not on theoretical arguments about negligible impact. Suryawanshi et al. consider surface water negligible based on tests over only two reservoirs, and they appear to generalize these results to the entire country, which is an overextension.Deconvolution problem for validation with the ref-GWSC:
Consequently, the downscaling approach risks misattributing surface water fluctuations, due to reservoir filling and releases, to groundwater storage. By considering only surface soil moisture (~5 cm) and neglecting the majority of reservoirs, the study likely overestimates GWS, and the signal attributed to GWS becomes a hybrid of groundwater and surface water. Such misinterpretation can distort seasonal dynamics, for instance by incorrectly suggesting rapid aquifer recharge when TWS increases during monsoon reservoir filling. Overall, the study underestimates the importance of surface water in TWS deconvolution, and the resulting GWS product may therefore be systematically biased, which leads to an impossibility to validate this published downscaled TWS dataset.I also agree with Reviewer 1 that the manuscript would benefit from a more thorough scientific discussion. Additionally, relevant references on GRACE downscaling that are missing include Jyolsna et al. (2021) and Karunakalage et al. (2021), which should be considered.
References cited in this review:
Salameh, E.; Frappart, F.; Papa, F.; Güntner, A.; Venugopal, V.; Getirana, A.; Prigent, C.; Aires, F.; Labat, D.; Laignel, B. Fifteen Years (1993–2007) of Surface Freshwater Storage Variability in the Ganges-Brahmaputra River Basin Using Multi-Satellite Observations. Water 2017, 9, 245. https://doi.org/10.3390/w9040245
Pascal, C., Ferrant, S., Rodriguez-Fernandez, N., Kerr, Y., Selles, A., Merlin, O. Indicator of Flood-Irrigated Crops From SMOS and SMAP Soil Moisture Products in Southern India. IEEE Geoscience and Remote Sensing Letters 2023, 20, 1–5. https://doi.org/10.1109/LGRS.2023.3267825
Pascal, C., Ferrant, S., Selles, A., Maréchal, J.-C., Gascoin, S., Merlin, O. High-Resolution Mapping of Rainwater Harvesting System Capacity from Satellite Derived Products in South India. IEEE IGARSS 2021, 7011–7014. https://doi.org/10.1109/IGARSS47720.2021.9553131
Missing downscaling studies in India: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. https://doi.org/10.1080/02626667.2021.1896719
Karunakalage, A., Sarkar, T., Kannaujiya, S., Chauhan, P., Pranjal, P., Taloor, A. K., Kumar, S. The appraisal of groundwater storage dwindling effect, by applying high resolution downscaling GRACE data in and around Mehsana district, Gujarat, India. Groundwater Sustain. Dev. 2021, 13, 100559. https://doi.org/10.1016/j.gsd.2021.100559
Citation: https://doi.org/10.5194/egusphere-2025-2888-RC2
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 1,286 | 48 | 19 | 1,353 | 31 | 37 |
- HTML: 1,286
- PDF: 48
- XML: 19
- Total: 1,353
- BibTeX: 31
- EndNote: 37
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
The manuscript mainly examines the downscaling of GRACE mascon-based total water storage anomalies over India, introducing a mascon-scale mass conservation approach and validating the product with in-situ groundwater well observations. The topic is of broad interest to researchers in hydrology and remote sensing, and the dataset produced could be potentially useful for regional water resource studies. However, the study builds closely on previous work (Vishwakarma et al., 2021), and the novelty is therefore somewhat limited. In addition, the manuscript suffers from some shortcomings in the literature review, methodological justification, interpretation of the results, and presentation of figures and references. Therefore, the manuscript requires substantial revisions before it can be considered for publication.
Major Comments
Minor Comments