Dear Authors,

I have read your manuscript with great interest. The study presents a data-driven “residual” approach to estimate groundwater storage, where time-variable terrestrial water storage (TWS) changes derived from GRACE and GRACE-FO are interpreted as the vertically integrated signal of total land water storage. Estimates of other land water components are then obtained from independent satellite observations or models and subtracted from TWS. To enhance consistency among the different datasets, the manuscript introduces spatial filtering techniques.

While the resulting groundwater estimates are potentially valuable, I believe several important technical issues need to be addressed to strengthen the robustness and interpretability of the results. These points likely require additional analysis, validation, and deeper discussion.

1. Filtering strategy

The filtering approach adopted from Sharifi et al. (2025) is not entirely convincing from a physical perspective. Defining an “optimal” filter based on maximizing correlations between individual storage components and TWS lacks a clear physical justification. If the objective is to ensure spectral consistency, a more transparent approach would be to represent the storage components in the spherical harmonic domain and apply filters consistent with GRACE(-FO) processing. This would allow consistency to be enforced directly in the spectral domain rather than through empirical tuning.

2. Consistency of storage magnitudes

Even if the filtering approach is accepted, there is no guarantee that independently derived storage components (from models or other satellite products) will match the magnitude of GRACE(-FO) TWS. Such inconsistencies are well documented and can be substantial. They are more appropriately addressed through data assimilation frameworks that explicitly account for uncertainties and model–data discrepancies. I encourage the authors to consider and discuss this aspect, for example as highlighted in Getirana et al. (2025, GRL, Inconsistencies in GRACE‐Based Groundwater Storage Estimation—A Call for a Proper Use of Land Surface Models - Getirana - 2025 - https://doi.org/10.1029/2025GL119197).

In addition, several offline data-merging approaches have been proposed in the literature that attempt to address such inconsistencies more systematically. These include ensemble-based approaches, statistical decomposition methods, and Bayesian merging frameworks (e.g., https://doi.org/10.5194/hess-18-2955-2014; doi:10.1016/j.rse.2013.09.025; https://doi.org/10.1016/j.scitotenv.2020.143579). A discussion of how the present method relates to or differs from these established techniques would significantly strengthen the manuscript.

3. Magnitude of groundwater estimates

The magnitude of the derived groundwater storage changes appears questionable, possibly due to over-filtering or the rescaling procedure. For instance, the spatial patterns and amplitudes shown in Figure 7 over Europe seem unrealistic. A comparison against independent estimates, such as outputs from global hydrological models (e.g., WGHM) or GRACE-based groundwater products (e.g., constrained estimates available via Figshare: https://doi.org/10.6084/m9.figshare.31073527), would be helpful to contextualize and validate the results.

4. Rescaling procedure

The manuscript would also benefit from a clearer description of the rescaling methodology. While filtering is described, it is not sufficiently clear how the final amplitudes are restored or interpreted. Since end users are typically interested in physically meaningful magnitudes, a transparent explanation of how near-realistic amplitudes are ensured is essential.

Overall, I believe the study addresses an important topic, and the proposed dataset could be valuable to the community. Addressing the points above, particularly with respect to methodological justification, consistency, and validation, would considerably improve the quality and reliability of the manuscript.

I hope these comments are helpful, and wish you all the best.

Ehsan Forootan