The paper contributes a reconstruction of the filling strategy of the Grand Ethiopian Renaissance Dam (GERD) in Ethiopia based on a combination of hydrological modeling and satellite data processing. The approach also explores the role of precipitation data uncertainty by considering five alternative rainfall products as input to the conceptual hydrological model (HBV). The study is very timely and interesting for HESS readership. However, despite the modeling framework looks solid and well designed, its implications and value for the ongoing water management dispute in the Nile River Basin should be better elaborated (see comments below) before accepting the paper for publication.

1) Since the abstract and in most of the introduction, the authors emphasize the value of inferring the GERD filling strategy to support a better management of the Nile River despite the long-lasting international tensions between Ethiopia and downstream countries (see lines 19-22; 57-58 74-75). I do second this point but would argue that the paper falls shy of these contributions to water management. The proposed approach uses a conceptual hydrological model calibrated at the Eldiem station (close to the Ethiopia-Sudan border) before the GERD construction to estimate the volume of water stored during the filling period as the difference between the simulated discharge in natural conditions minus the observed one. The resulting storage trajectory is validated against a trajectory reconstructed from Landsat images according to the method proposed by Vu et al. (2022). As said before, this modeling approach sounds solid and well-designed, except for a couple of minor points reported below. 

However, the discussion in Section 4.4 about the value of these findings for the ongoing water management dispute in the Nile River Basin is relatively simplistic. Here, the authors only comment about the reconstructed filling strategy (Fig. 10) and streamflow entering in Sudan (Fig. 11), raising the following doubt: is the proposed approach really necessary for informing water management? On the one hand, the reconstructed trajectory in Fig. 10 could be obtained with the approach by Vu et al (2022) only using satellite images; on the other hand, the flow entering in Sudan is directly measured at Eldiem station, regardless of the models developed for the upstream part. To satisfy the (high) expectations generated in the abstract-introduction, I believe the authors should try to expand this part of the manuscript in order to better show the potential value of their model. For example, can you use your results to infer a rule that could be used to simulate the rest of the filling period? can you quantify the value of the information produced by your model for supporting the pro-active operations of Roseires and Sennar dams (as mentioned at line 67)? how should these two dams be operated to adapt/coordinate with the upstream filling policy?

Replying to this type of questions is in my opinion necessary to make the paper's findings valuable on the policy side. If authors believe this is going beyond the scope of their work, I would suggest revising the narrative of the abstract and intro in order to downplay these aspects and better characterize their contribution.

2) The reconstruction of the filling strategy is built on the hydrologic simulation of the HBV-light model. This conceptual model was calibrated during the 2006-2019 period and validated over the period 1991-1996. How reliable is this strategy given the evident global warming/climate change trends? Did the authors check the presence of trends in precipitation and temperature data? Since the calibration relies on 10,000 random parameter sets that returned 1756 simulations with NSE>0.75 (lines 242-244), I suspect the "best" parameterization adopted might not necessarily be so valid when applied to the 2020-2022 time period.

3) The authors are validating the reconstruction based on the HBV-light model using the approach by Vu et al. (2022). However, they notice only 53% of the Landsat images are cloudless, with several missing data during the wet season, which is also the most critical in terms of filling. Why did they not consider also using radar altimetry data to complement Lansat images?

Summary: This study uses a lumped hydrological model, along with (not “coupled with” as noted in the abstract) remote sensing data to examine the filling strategies of the Grand Ethiopian Renaissance Dam (GERD). The model used is the HBC-light model, which is used to simulate the inflow into the reservoir, evaporation etc. Overall, it is an interesting study that presents substantial results on inflow, outflow, and filling strategies for GERD. The paper is generally well written and there is a lot to like in the paper, thus I am generally supportive of the work and believe that it could eventually be published in HESS; however, substantial revisions are necessary before the paper can be accepted. I provide my detailed comments below.

** Abstract, Line 6, “coupled”: I don’t think the model is coupled with remote sensing data. RS data is used in conjunction with the model. Please revise this statement.

** Abstract and conclusion: As I noted above, there is a lot going on in this study; however, I am not convinced that the study, at least as it stands in the current form, presents sufficient novel scientific insights. It surely presents substantial information that could be used to manage reservoirs in the study region, but I ask: what is the scientific merit? I suggest that authors revise the introduction to address this issue, and perhaps some changes in the results and conclusion sections should be made as well.

** End of introduction (Lines 70-79): this is not very convincing. Again, what is the major scientific contribution of this work? Please clearly specify scientific questions and objectives. The authors attempt to justify the study (toward the end of the paper) noting that the approach/framework could be generally applicable to other (data sparse) regions; I am not sure how valid this claim is given substantial uncertainties in the ability to simulate the flows by the model and the inherent limitations in remote sensing data.

** The simulated inflow is somewhat questionable as it is not validated with any observed data. Given many sources of uncertainty, how do the authors ensure that the simulated inflow is reasonable?

** Figure 4 (related to the above comment), “Best simulation”: I assume “observation” here is the outflow and “Best simulation” is the inflow. Please clarify by changing the legends.

** Figure 4: Is the unit “MCM”? Discharge should have a unit with per unit time, not just volume! Same applies to the right axis of Figure 11.

** Figure 7 and other relevant section: Given limitations in Landsat imageries, could the authors use other remote sensing products such as those from Sentinel?

** There are high uncertainties/errors in many of the results presented, which are not discussed in the present manuscript. The uncertainties arising from precipitation data are discussed, but there are many other sources of uncertainties including the use of PET data from ERA5. Is the PET data reliable? How do any uncertainties in the PET data affect the results. I suggest that the authors present a dedicated (concise) section on various sources of uncertainties, the implication on the results, and potentially the conclusion drawn.

** L403: The question here suffers from critical grammatical issues. Please carefully rewrite it.

Minor comments:

** There are excessive abbreviations used in this paper. I suggest removing those that are not necessary.

** I found that figure number is not done in an increasing order in the text. Also, the arrangement/placement of figures in the text is not good; many figures are 2-3 pages away from where they are referenced in the text, which makes it hard to read.

** L85: UBN was already spelled out.

** L118: What does “satellite imagery” mean here? Please specify.

** L210: “is” à “are”?

** L248: “shows” = “show”