| 16 Dec 2025
How does the choice of the input hydrograph affect reservoir and dam design?
Abstract. Reservoir and dam design requires a detailed understanding of the entire input hydrograph rather than relying solely on peak discharge estimates. Input hydrographs are essential both for verification purposes to evaluate the peak attenuation capacity of the reservoir and for design purposes to define reservoir height, its volume and outlet structures. However, no universal guidelines exist for selecting the input hydrograph, leaving designers to navigate conflicting methodologies often without clear evidence of the advantages and drawbacks of different hydrographs. As a result, the choice of the input hydrograph is typically based on local regulations, introducing subjectivity and potential inconsistencies. This study investigates how sensitive reservoir and dam design is to the choice of the input hydrograph by quantifying the differences in reservoir key parameters, such as the maximum outflow discharge, maximum storable volume, and peak reduction effect. The analysis compares the results of the most commonly used input hydrographs and continuous time series routing. The study highlights the advantages and potential limitations of commonly used input hydrographs, particularly regarding their ability to represent hydrological conditions of the time series accurately. The findings of this study aim to offer a more conscious approach to hydrograph selection, potentially reducing subjectivity and improving the robustness of reservoir and dam design practices. Finally, the study seeks to address computational challenges associated with reservoir routing by identifying efficient yet reliable hydrograph options.
The article ‘How does the choice of the input hydrograph affect reservoir and dam design?'’ aims to assess the influence of different methods for defining the hydrological load in the context of reservoir design and verification. Specifically, it relies on a design-event approach (which in this case is unusually independent from the return period) and compares results obtained by routing different reservoir configurations with (i) different combinations of rainfall design events and IUH models and (ii) a ‘flow time series’ (line 264; I do not understand how it is obtained from the rainfall time series presented in the manuscript – another rainfall-runoff model?). The results, which are not always clearly presented, highlight that, although the findings are not fully generalizable (line 444), there are differences in the mean annual maximum outflow, maximum storable volume, and peak reduction effect obtained by routing a reservoir with a continuous time series and commonly used design ‘hyetographs’ (line 411).
Although I found the idea of this comparison interesting and useful for researchers and practitioners, I found: (i) an introduction that does not sufficiently explore the problem, (ii) a structure of the paper which is confusing and difficult to follow, (iii) some assumptions regarding the simulated reservoir configurations that are not clearly presented or discussed, (iv) an approach that appears more hyetograph-oriented than hydrograph-oriented, and (v) a conclusion which does not offer new insights, as I expected. In particular, I found two important inconsistencies on which the paper relies. These inconsistencies affect the interpretation of the results and the overall contribution of the study.
The first one is that, in my opinion, a more detailed hyetograph could certainly lead to a solution closer to reality. However, often the assumption of a rectangular hyetograph and a triangular-trapezoidal hydrograph (i.e., based on the kinematic wave model) is necessary to provide analytical solutions, when it is important to simplify reality. If this is not the aim, of course, more sophisticated methods can be used. Hence, I think that it is important to clarify that each assumption should be consistent with the purpose of the analysis. In this regard, my second concern is related to the choice of the BLUE hyetograph and the recognition of its better performance in the proposed work. This aspect is not entirely clear to me, but, as stated in lines 182–186, it is obtained directly from the flow time series, as opposed to the other hyetographs, which are calculated mainly considering the IDF curves. This raises concerns about the independence of the comparison between the BLUE hyetograph and the continuous simulation. Hence, does the good agreement of the BLUE hyetograph with the continuous simulation introduce an inherent bias in the comparison? Also, when you have a flow time series, why should one use the BLUE hyetograph instead of directly defining a design hydrograph from the flow time series? (see also my comment at Lines 180-186)
I encourage the authors to clarify these points and address the potential inconsistencies highlighted above.
In addition, I provide the following specific comments:
Line 14: ‘maximum storable volume’ could you please be more precise about the meaning of this key variable?
Lines 15–20: From the abstract, I do not grasp the main conclusions of the work; could you please include some of your findings here?
Lines 48–49: Does a design-event approach necessarily assume a reservoir empty at the time of a flood event? I think that different reservoir-filled conditions can indeed be simulated.
Line 82: Why did you compute the mean of annual maxima of the variables? Why do you not analyze your results also in terms of return period? I missed this point.
Introduction: I feel that references to previous studies addressing this type of analysis and reporting relevant findings are missing. The state of the art is not discussed. I would encourage the authors to include, in the introduction, key results from the existing literature on this topic.
Lines 94–116: Why did you not include this part directly in the Methodology section (as usually done in manuscript structure)?
Lines 114–116: Here you define the equations describing the reservoir-routing system. Specifically, you include both the possibility of considering a bottom outlet and a free and unregulated spillway. Hence, for the same reservoir, you could have two different equations based on the water level in the reservoir and the spillway crest height. Could you please clarify this aspect? Also, I see that you consider in your simulations only bottom outlets. Am I right? And why did you not simulate the full reservoir behavior?
Lines 120–130: If I correctly understood, your aim is to compare a continuous simulation (with the same rainfall-runoff models) with a design-event approach. Am I right? At line 134, you refer in general to Equation 1. Hence, how did you evaluate your reference flow time series? It is not clear to me. At lines 210–212 it is better explained, but it should be addressed earlier.
Lines 180–186: I do not understand your assumption in the BLUE approach. Indeed, the procedure of Alfieri et al. (2008), as you mentioned, estimates the intensity considering the rainfall depth computed from the IDF curves and the time of concentration of the basin. I found it reasonable compared to your assumption, which depends on the flow time series. This point, as I mentioned before, is not clear to me.
Lines 236–240: Why did you consider only bottom outlets and not the unregulated free spillway? How could this affect the results?
Lines 258–260: Did you consider here the behavior of the spillway? In Table 4, only the parameters of the bottom outlet are shown.
Results and Conclusion: As I mentioned before, I found the discussion of the results and conclusions to be more hyetograph-choice oriented than hydrograph-choice oriented. This is because (i) I do not find a clear discussion of the specific rainfall–runoff models adopted, and (ii) I do not fully understand whether the comparison between the continuous simulation and the design-event simulation is carried out using the same IUH models. In the latter case, the manuscript should be framed explicitly as hyetograph-oriented.
Overall, while I find the topic of the manuscript relevant and the idea of comparing different approaches potentially valuable, the current version of the paper presents, in my opinion, several conceptual and methodological issues that significantly limit the robustness of the conclusions. It is also not well positioned within the scientific context (there is no clear introduction to the problem and no references to other manuscripts that investigate the same topic). For these reasons, I believe that the manuscript requires a major revision before it can be considered for publication.