This manuscript is overall well written and easy to follow. The topic fits GMD very well. The authors nevertheless are challenged to better bring out its novelty or significance beyond applying an existing reservoir scheme over Spain. The objective of this study was two folds. First, it “proposes” a global and parameterized reservoir model. Secondly, it performs sensitivity analysis on the model parameters.  Specific comments on the two aims are the following.

• The first aim of the study

Their statement is factual as the model simply converts deterministic values presented in Hanasaki et al. (2006) to parameters. However, even that is not the first attempt as it has been attempted by Shin et al. (2019). For example, in Equation (5), at M = 0.5, this Equation becomes Hanasaki. Similar case for Equation (6). Therefore, in general, this is an existing scheme. The naming of a scheme developed by others as new is also wrong from my point of view.

• The second aim of the study

Concerning the sensitivity, the identified most sensitive parameters are ones anyone could generalize through educated guesses. C_treshold is the parameter for the lower bound of the Hanasaki schemes, the storage capacity to mean total annual inflow ratio. Equation (6) represents the final release where ‘c’ is the determinant factor. Below the threshold, the Equation is a one-parameter function, and the threshold has no influence. Above the threshold, it is a three-parameter and uses the threshold value directly. This is also the case for the M parameter (Equation 5). Further, the sensitivity analysis is performed on the flow; both M and C_treshold play a major role in release estimation. So, one can easily reach the same conclusion without performing the sensitivity analysis.

Since the novelty on the modeling or parameterization side is somehow limited, perhaps the authors can provide more analysis and discussions on the similarity and difference between the reservoir/river dynamics in Spain and other regions, hence producing improved scientific understanding. 

General comments

The authors examined parameterizations of reservoir operations in Spain. They employed a series of formulations developed by Hanasaki et al. (2006) that reproduce reservoir release and storage. They conducted intensive parameter sensitivity tests and identified several highly sensitive parameters. They also analyzed the performance of 215 reservoirs to discuss why they performed better or worse.

The formulation of Hanasaki et al. (2006) has been employed in many global hydrological models, but is rarely additionally and carefully validated. This study is unique because it presents the results of intensive validation in 215 reservoirs in Spain where reliable data are well accessible. Another interesting result, somewhat surprisingly, is that the simple formulation works well with observations in many cases.

The manuscript is quite well written and clear.I only noticed that a few technical clarifications are necessary in several parts as detailed below.

Specific comments

Line 82 "direct runoff Rd": What is this? Is this different from "tributary inflows"?

Line 135 "Unlike Hanasaki et al. (2006) scheme...": It is a bit unclear as to what exactly is different from Hanasaki et al. (2006). Does it mean that industrial and domestic water requirements are not considered in this study? For readers' convenience, it would be nice if the authors add a list of the major changes from Hanasaki et al. (2006).

Line 400-415 “When using Sobol indices, …”: I couldn’t follow the discussion for these two paragraphs. In short, I couldn’t understand what Figure 12 is showing. What does it mean by “shifts in C_2M”? What is the PDF being discussed here? A bit more detailed and readable explanation is needed.

Line 452 “Zhou et al. (2021) suggested an efficient way to overcome this issue…”. I couldn’t fully understand what issue is focused on here and how Zhou et al. (2021) solved it. It would be helpful for readers if you could elaboration further. Perhaps, I guess this part discusses the cells that should be included in the estimation of irrigation water demand. The spatial resolution of Hanasaki et al. (2006) was 1 degree by 1 degree, or 110 km by 110km at the equator. Therefore, it seemed that summing the grid cells direct downstream would be sufficient to estimate irrigation water demand for large reservoirs. However, the spatial resolution in this study is 5km by 5km. It seems necessary to include not only the direct downstream grid cells, but also the surrounding ones. Some of the latest global hydrological models with a high spatial resolution have started to transfer water from main stream to the surrounding grid cells (e.g. the “aqueducts” of Hanasaki et al. 2018, 2022). Maybe such a concept is needed to estimate downstream water demand reasonably.

References

Hanasaki, N., Yoshikawa, S., Pokhrel, Y., and Kanae, S.: A global hydrological simulation to specify the sources of water used by humans, Hydrol. Earth Syst. Sci., 22, 789-817, 10.5194/hess-22-789-2018, 2018.

Hanasaki, N., Matsuda, H., Fujiwara, M., Hirabayashi, Y., Seto, S., Kanae, S., and Oki, T.: Toward hyper-resolution global hydrological models including human activities: application to Kyushu island, Japan, Hydrol. Earth Syst. Sci., 26, 1953-1975, 10.5194/hess-26-1953-2022, 2022.