Assessing the cumulative impact of on-farm reservoirs on modeled surface hydrology

Perin, Vinicius; Tulbure, Mirela G.; Fang, Shiqi; Arumugam, Sankarasubramanian; Reba, Michele L.; Yaeger, Mary A.

doi:10.5194/hess-29-6353-2025

Articles | Volume 29, issue 22

https://doi.org/10.5194/hess-29-6353-2025

© Author(s) 2025. This work is distributed under
the Creative Commons Attribution 4.0 License.

https://doi.org/10.5194/hess-29-6353-2025

© Author(s) 2025. This work is distributed under
the Creative Commons Attribution 4.0 License.

Articles | Volume 29, issue 22

Research article

|

18 Nov 2025

Research article |

| 18 Nov 2025

Assessing the cumulative impact of on-farm reservoirs on modeled surface hydrology

Vinicius Perin, Mirela G. Tulbure, Shiqi Fang, Sankarasubramanian Arumugam, Michele L. Reba, and Mary A. Yaeger

Download

Final revised paper (published on 18 Nov 2025)
Supplement to the final revised paper
Preprint (discussion started on 17 Jun 2024)
Supplement to the preprint

Interactive discussion

Status: closed

RC1:
'Comment on hess-2024-148', Anonymous Referee #1, 24 Jul 2024
General comments
The article deals on an important issue, the impact of on-farm reservoirs on the hydrology.

The article propose an interesting application, with an innovation that consist in imposing surface and hence volume of the reservoirs in the model as derived from satellite data.

The application is in Arkansas, USA, with about 300 OFRs on a basin of 530km2, known for the importance of irrigation.

One issue with such reservoirs is the lack of data on their management.

From what I understood, the study tries to retrieve some parts of the management of the OFRs by imposing the extension of surface water of the OFRs which is very interesting

However, some elements of the methods are not clear, and I couldn’t understand how the model really works, and how the OFRs are managed.

My main comments is that the way OFRs are models and their management should be clarified
Details questions

Introduction : Peak flow is mentioned in the introduction and analysed in the study, but never defined… In the study, it seems to be maximal annual monthly flow, which is quite far to what can expect as peak flow, ie, flood… Can you define ?
Section 2.1: It is stated that there are with about 330 OFRs. The size of the basin, given later in the text is 7107 km2. Thus, the density of the OFRs is lower than 0.05 OFRs/km². This corresponds to a small density, especially considering large annual precipitation (1300mm/year) when compared to previous studies review by Habets et al., 2018 refered in the article. It is stated that 95 % of the OFR are smallest than 50ha, and Fig 4, it is shown than only about 10 aggragated OFRs are smaller than 10ha. According to the hypothesis used in the study of an average depth of 10m, 10ha corresponds to a capacity of 1 million cubic meter, which is quite large.

So is there only 330 OFRs because there are actually rather large OFRs. Or Is it possible that smaller OFRs are missing ?
Line 274 « For each of the aggregated OFR, the water volume was calculated using SWAT+ default rule, which is a simple multiplication of the OFR surface area by a factor of 10 » : Do you mean that the maximum volume capacity of the OFRs is the maximum surface area multiplied by 10m ? Or do you mean that the water level within the OFR is constant and fixed to 10 ? This is unclear...
Line 282 : I find it weird to have details on how the river channel are divided in 4 classes, while, no details on the way the OFRs impacts the water balance are given…

You need to provide the water balance of the OFRs:
what are the inputs and ouptuts of the OFRs ?

Is there water abstraction from the OFRs ? How much ? How is it computed ? What are the temporal variations ?

Is there evaporation from the OFRs ?

Does the water level vary ? Is the water level affect the outflow ?, due to the spillway

The surface of the OFRs is changing. But is it the only way to have a change on the volume of the OFRs ?

What’s happening if the surface of the OFRs is decreasing ? Does it lead to an outflow of water downstream ? Does the associated volume is expected to be abstracted for irrigation ? How this volume is estimated ?

What happens when the surface area increase ? Does the inflow feels the reservoir ? Again how the corresponding volume is estimated ? What if the inflow is not large enough to fill the OFRs ?

I’d like to have a detail explanation of how it works.
Section 2.4 Scenario Analysis
line 297 ; "The daily surface area time series of each OFR was used to simulate three scenarios (i.e., lower, mean, and upper) representing the OFRs’ capacity in terms of surface area." Sorry, but, again, this part is not clear. Figure 2 is not very helpfull to understand what is done…
It is stated that the « daily OFRs’ surface area change between 2017 and 2020 » is derived,, and then that (line 300) « The mean scenario represents the regular condition of the OFRs, and it is the mean of the daily surface area time series derived from the Kalman filter , The lower and upper scenarios represent the lowest and highest capacities of the OFRs, and they are based on the surface area 95% confidence interval limits, calculated using the daily time series. »
So there is 4 value for each day of the year, how do you compute a 95 % confidence ?
Line 304 : » For each scenario, the OFRs were simulated at full capacity (i.e., maximum storage at the lower, mean and upper scenarios), and this capacity was kept constant during the simulation period »
⇒ this seems to be in contradiction with the previous sentence… Do you mean that the maximum capacity is set constant ? That the volume is set constant… ? Please, make it clearer...
Moreover, this part is the innovative part of the article. And only the distribution of the area of the aggregated OFRs is given Figure 4, and no details are given on the annual cycle of the OFRs … I strongly suggest that the daily evolution of the surface area of the OFRs be presented, for the 3 scenarios.

And again, please explain how the evolution of the surface impact the evolution of the volume….
Does this surface change gives indication on the volume of water used for irrigation ? If yes, please, provide the estimated values…
Line 360 : The impact of the OFRs on monthly flow varied throughout the year…. Ok, but here, the reason of this impact is not clear : there is no information on the management of the OFRs, so, no idea of when they are filled, when the water is used/ abstracted, the condition for the water to spill out… So, again, please provide an description of the water balance and describe the hypotheses… the dynamic of the storage is clearly missing….
All this part is difficult to follow since key informations are missing…
Section 3.3 : impact on peak flow… You have to define what you consider as peak flow...
Section 3.4 Again, difficult to understand as the hypotheses on the functioning of the OFRs are not clear. You should present the evolution of the OFRs volume and surface in this section, as well as the evolution of the abstracted water….
Section 4 : Discussion
line 512 : « our validation and calibration was done using the flow measurements, and the OFRs surface area scenarios were based on an algorithm that was validated with an independent higher spatial resolution dataset (Perin et al., 2022). »
==> This sentence is not clear : OFRs surface area is an input data, not an output to be validated….
line 515 : « Furthermore, differently from previous research, our results showed that the OFRs may have a positive (< 9%) impact on flow (Fig. 5, classes 3 and 4) »

⇒ This is true only for some months…But you don’t explain why. This can occur only if OFRs release some water. But the management of the OFRs is not presented at all…
line 527 : « This could be explained by the level of details in our analyses. »

⇒Well to achieve this, it is necessary to know how much of the water in the OFRs is used, directly in the OFRs, but also, if the OFRs releases water in the river, to have a clear idea of the pumping directly in the river.
Moreover, it is not clear that this increase of riverflow with OFRs improves the model results, thus, correct a bias of the model without OFRs.
line 542 « For instance, during January and May the mean monthly percent change ranged between -35.8 ± 6% and -32.0 ± 7%, and during June and December it varied between -8.8 ± 5% and -5.4 ± 6% for the three surface area scenarios »

⇒ Do these values refer to the evolution of the surface of the OFRs ? Such info is needed earlier to understand the method and the results
line 559 : « Our findings highlight that the impacts of the OFRs on flow and peak flow have a significant intra- and inter-annual variability (Figs. 5, 6, and 7) »

⇒ Fig 5 only present monthly flow. So, again, it depends on what you call peak flow...
line 593 « Our results indicate that OFRs do not have an equally distributed impact on mean and peak flow across the watershed. Hence, assessing the OFRs location as well as their numbers across the watershed is important when aiming to manage the construction of new OFRs. »

⇒ I don’t understand how you can reach such statement… 1st, because no indication is given on the OFRs management, 2nd, the density of the OFRs varies in space….
line 620 : « there is no bathymetrie » Does it means that the water level in the OFR is constant !
Reference
Yaeger et al : Trends in the construction of on-farm irrigation reservoirs in response to aquifer decline in eastern is not well referenced
Citation: https://doi.org/10.5194/hess-2024-148-RC1
- AC2: 'Reply on RC1', Vinicius Perin, 31 Oct 2024
  
  See response on file.
  
  Citation: https://doi.org/10.5194/hess-2024-148-AC2
RC2:
'Comment on hess-2024-148', Anonymous Referee #2, 30 Sep 2024
Dear Authors,
The study innovatively assessed the spatial and temporal variability of the cumulative impact of OFRs at the watershed and subwatersheds levels, quantifying the annual impacts of the OFRs on flow and peak flow at the channel scale. Although the manuscript presents relevant novelty, there are some issues, which must be addressed or clarified by the Authors, prior to the study's publication, in my point of view. Please, see below my comments and suggestions.
Major comments:
The influence of OFRs on surface hydrology and their mathematical representation could not be adequately evaluated, because there is no streamflow monitoring in the areas with the presence of the OFRs (see Fig. 1).

Very simple considerations were assumed for the water volume and water releases of OFRs (see Lines 273-278). This is a critical issue of the study, which should be addressed by fieldwork or at least by a sensitivity analysis.

Please, explain in detail how hydrologically the OFRs could positively impact streamflow, as you found.

I did not fully understand what kind of results you would like to explore in section 3.5 Overall impact of OFRs. Please, improve this section.

In the beginning of the manuscript, you mentioned that one of your focuses is the analysis of the interannual flow variability. However, I could not find such an analysis. An example of simulated time series (section 3.4) is not enough. For example, what are the impacts of OFRs on low-flow and high-low years?

Minor comments:
Lines 385-386: “In general, the OFRs contributed to decreased monthly flow. However, the OFRs' impact on flow had a significant intra- and inter-annual variability…” Figure 5 is only on intra-annual variability or, also called, seasonal variability.
Citation: https://doi.org/10.5194/hess-2024-148-RC2
- AC1: 'Reply on RC2', Vinicius Perin, 31 Oct 2024
  
  See response on file.
  
  Citation: https://doi.org/10.5194/hess-2024-148-AC1

Peer review completion

AR – Author's response | RR – Referee report | ED – Editor decision | EF – Editorial file upload

ED: Publish subject to revisions (further review by editor and referees) (16 Nov 2024) by Pieter van der Zaag

This paper addresses an interesting and important topic. If it succeeds in better quantifying the impact of small reservoirs on the hydrology (rather than the modelled hydrology, as in the title) than it is also a salient paper. But I am yet to be convinced what new knowledge this paper adds, despite of the use of the word “novel” in the abstract.

My main worry concerns my doubts whether the method that the authors adopt is suited to achieve their objective, that is to assess the cumulative impact of on-farm reservoirs that “store water from precipitation and runoff during the rainy season to irrigate … crops during the dry season” (lines 18-19).

These reservoirs (OFRs) are meant to displace water in time in order to provide water for irrigated lands during the dry season. But the model the authors employed cannot or does not model water abstractions from these reservoirs to the farm lands. The only thing the model apparently does is follow “the model default rule, which sets the OFRs to release water when the spillway volume is reached – 80% of the OFRs capacity” (lines 276-277). So the model cannot model the actual use of the OFRs, the impact of which the authors want to assess. It is also not clear how irrigation water fluxes are or are not included in the SWAT+ model.

An alternative way of assessing the impact of OFRs on the hydrology is to measure the change in one of the most important fluxes of the water balance, namely evaporation from the irrigated crops (transpiration or evapotranspiration), whereby it could be assumed that with each OFR there are associated irrigated fields, i.e. more OFRs imply more irrigated fields and more evapotranspiration. The added evaporation will obviously impact the surface hydrology. But the paper does not even discuss this, let alone model it.

The point is that the reservoirs themselves have a limited impact on the hydrology, compared to their capacity to divert water in time and place in order for it to (largely) evaporate. So statements such as “the presence of OFRs in the watershed decreased annual flow” (lines 27-28) are suggestive and strictly speaking incorrect. A more correct statement would be “the presence of OFRs in the watershed is associated with a decreased annual flow”. The OFRs themselves are unlikely to significantly decrease these annual flows, as the study region is not very arid, and the additional evaporation losses directly from the open-water reservoirs is probably limited. The impact largely comes from increased irrigation, and this is of course nothing new. A PhD student of mine ten years ago used actual evaporation estimates derived from remote sensing to force a hydrological model, which could adequately model an intensely (irrigated) cropped catchment in Africa (Kiptala et al., 2014). Nowadays there are better products, even and in particular from Planet Inc. (!), for estimating actual evapotranspiration.

The above concerns need to be addressed in the revised paper. The revised paper should obviously also adequately address the issues raised by the two reviewers.

Reference
Kiptala, J.K., M.L. Mul, Y. Mohamed and P. van der Zaag, 2014. Modelling stream flow and quantifying blue water using modified STREAM model for a heterogeneous, highly utilized and data-scarce river basin in Africa. Hydrol. Earth Syst. Sci. 18, 2287–2303 [doi:10.5194/hess-10-18-2287-2014]

Hide

AR by Vinicius Perin on behalf of the Authors (16 Jan 2025) Author's response Author's tracked changes Manuscript

ED: Referee Nomination & Report Request started (04 Feb 2025) by Pieter van der Zaag

RR by Anonymous Referee #3 (24 Mar 2025)

ED: Reconsider after major revisions (further review by editor and referees) (04 Apr 2025) by Pieter van der Zaag

AR by Vinicius Perin on behalf of the Authors (02 Jul 2025) Author's response Author's tracked changes Manuscript

ED: Publish subject to revisions (further review by editor and referees) (20 Jul 2025) by Pieter van der Zaag

(The line numbers quoted below refer to the track changes version of the revised manuscript.)

I find that the authors did not adequately respond to a major concern of the 3rd reviewer, namely

“The major uncertainty around OFRs is how they are managed and how much water for irrigation is actually abstracted from them. […] I find this aspect a little bit odd in the paper. The authors […] never consider irrigation in the model configuration.” And again: “… the paper does not really advance this field significantly because the major source of uncertainty, i.e., irrigation abstraction and other management-related aspects of the reservoirs, are completely neglected”

The premise of the paper is that OFRs impact the hydrology (e.g. lines 51-52). But the main impact of OFRs on the hydrology is not the OFR itself but rather how the OFR is being managed, namely for using the stored water for irrigation. Without the use of the stored water in OFRs for irrigation, OFRs will have hardly a noticeable impact on the hydrology. But as this irrigation use of the stored water is ignored in the model used, the innovation of this paper is limited.

For example, there might be a correlation between the change in the number or size of OFRs in a certain area and a change in the irrigated area in that same area, and with it the change in water consumption by that change in irrigated area. There are currently many RS products that can reliably estimate such changes in water consumption patterns. Including this could possibly significantly improve our understanding of the impact of OFRs on the hydrology. I write this just to invite the authors to more adequately reflect on the 3rd reviewer’s concern.

I have a few more detailed comments on the revised manuscript, as follows, which I would like the authors also to address:

1. In your rebuttal you promised the following, but I could not find this in the revised manuscript:

“We reviewed and cited the two studies in our manuscript. The new lines in the manuscript are copied below: “For example, other studies have found that OFRs reduce annual and monthly runoffs in southeastern China (Yan et al. 2023) and Australia’s Murray-Darling Basin (Robertson et al. 2023).”

2. In lines 58-59 you write “… the spatial and temporal variability of OFRs …” What do you mean by this? Kindly clarify.

3. The paper frequently refers to inflows and outflows of OFRs (e.g. line 74, 568-569, 639, 692, 716, 719). But farmers also abstract water from OFRs through e.g. pumps and pipes, in order to irrigate their crops. This abstraction is also not covered in the OFR water balance given in the new equation 3 (line 294). Kindly explain why you omit this, in view important, feature of OFRs.

4. In line 280-281 the paper now reads: “The total volume of water in the OFR fluctuates in response to changes in surface area …” I think this is an incorrect statement. The volume of water in an OFR changes due to changes in inflows, outflows or abstractions. Such changes of water volumes are associated with changes in surface area of the OFR, but are not caused by it, as is suggested in this text.

5. Lines 311-314: I do not understand what the volume of water and the surface area at the emergency spillway or at the principal spillway mean. For readers that are unfamiliar with the specific design of an OFR, like myself, this needs an explanation.

6. Line 352: “SWOT+” Is this correct?

7. Lines 557-560: Why does this sentence start with “Although”? In my understanding the sentence could more appropriately start with the word “As”.

8. Line 641: “… may not lead to differences (e.g., > 10%) …” I would add: “significant”, as follows: “… may not lead to significant differences (e.g., > 10%) …”

9. At lines 184 and 735 you refer to Mollina-Navarro et al., but that should read Molina-Navarro et al.

Hide

AR by Vinicius Perin on behalf of the Authors (04 Sep 2025) Author's response Author's tracked changes

EF by Katja Gänger (08 Sep 2025) Manuscript

ED: Publish as is (22 Sep 2025) by Pieter van der Zaag

AR by Vinicius Perin on behalf of the Authors (02 Oct 2025) Manuscript

Short summary

This study assesses how small on-farm reservoirs (OFRs) can impact surface hydrology using a new framework combining remote sensing and hydrological modeling. Findings show that reservoirs can reduce annual flow by 14–24 % and peak flow by 43–60 %, with effects varying by location and reservoir's capacity. This is relevant as the number of OFRs is expected to increase globally as an adaptation to climate change under severe drought conditions.

Assessing the cumulative impact of on-farm reservoirs on modeled surface hydrology

Download

Interactive discussion

Peer review completion

Suggestions for revision or reasons for rejection