| 15 Oct 2025
The influence of small farm reservoir network characteristics on their cumulative hydrological impacts
Abstract. In many regions of the world, the use of infrastructure to store runoff and stream water, such as small farm reservoirs, is the only way to enable irrigation and thereby secure and increase food production. The presence of multiple reservoirs in one catchment has cumulative impacts that are not necessarily the sum of the individual impacts. However, we still have little knowledge of the spatial factors that drive these cumulative impacts. In this work, the effects of the distribution of small reservoirs in a catchment on their hydrological impacts are investigated with a modeling approach. Our numerical experiment consists of randomly generating multiple small reservoir networks in the same catchment with realistic reservoir numbers, capacities, and spatial distributions and then comparing their hydrological impacts over a 20-year period. We focused on two variables, namely, the outlet discharge and the mean proportion of the network in low flow, which we computed annually and seasonally. We used the distributed agrohydrological model MHYDAS-small-reservoir, which represents small reservoirs and their links with the hydrological network and the irrigated plots. In our context and with current reservoir management rules, we found that the impacts of reservoirs are more important in summer, with discharges reduced by more than 20 % and up to 60 % compared with the reference situation without reservoirs. Moreover, low flow proportions are always higher than those in the reference situation. For these two indicators, the main explanatory factors are the number and distribution of reservoirs, with a limited effect of the storage capacity. The effects of the study factors on the seasonal and annual indicators were thoroughly interpreted with respect to the hydrological functioning of the catchment and the timing and amount of irrigation. This work contributes to a better understanding of the drivers of the cumulative hydrological impacts of small reservoirs. Although many questions remain, our results can help scientists and water managers choose the best representation of small reservoirs in their models to address their needs.
This manuscript describes the cumulative impact of small reservoirs on the hydrology of a small catchment area in south-western France. These reservoirs are generated randomly and are connected to the river network. Their characteristics include their water storage capacity, their spatial distribution across the catchment area and their number. The metrics used to assess the impact of these different components are annual flows, summer flows, and the proportion of the hydrographic network at low flow each year. The study is based on 20-year numerical simulations. It shows that the impact of reservoirs is cumulative along the river network and assesses the effects of the distribution, number and capacity of reservoirs on flows, and studies the associated processes. This work is very interesting and also sensitive because it attempts to provide answers to societal questions, in particular the storage of water for irrigation in dedicated reservoirs.
I find this article very well structured and well written. The conclusions are based on objective evidence and clear figures, and the limitations of the study are well presented in the discussion. I recommend publication of this article after minor revisions.
Comments:
- L109: evaporation from the reservoir is considered to be 60% of the reference evaporation, which I assume corresponds to potential evapotranspiration. How valid is this approximation and what is the associated uncertainty? It is important to give an order of magnitude for annual evaporation, as well as annual withdrawals, as these contribute to direct mass loss.
- L117: withdrawals are possible if the volume of water is greater than 1/4 of the reservoir's capacity: what average water depth does this correspond to, given that the shape is an inverted pyramid? Is this compatible with the characteristics of the withdrawal pumps?
- L114: on the map of France in Figure 1, there are white pixels in the Rhone valley that should not be there. What do they correspond to?
- L127: average annual rainfall is 675 mm: how was this calculated? Using SAFRAN data or measurements from rain gauges located in the catchment area?
- L128: The total volume capacity of the reservoirs in the basin is estimated at 205,000 m³. Is this estimate based on the pyramidal shape of the reservoirs or does it come from data describing the various structures?
- L140: What do you mean by “better match expectations”?
- L147: An example of one or two reservoir distributions would be helpful. For example, showing one distribution of the 7, 14 or 21 downstream reservoirs with the two capacities (two colours) as in Figure 1 and the main irrigable crop would help to clarify the ideas.
- L180: what is the proportion of reservoirs not connected to the hydrographic network compared to those that are connected? They also contribute to water storage for irrigation.
- L225: a warm-up period of 5 years is considered: does this mean that the simulated data used start in September 2000? Or is the warm-up from 1990 to 1995? Please clarify.
- L233: references to Vidal et al. (2010) and Le Moigne et al. (2020) can be added for SAFRAN. Can you clarify if the reference evapotranspiration is computed from SAFRAN data and at what frequency? Same question for min and max temperatures, do they come from the SAFRAN reanalysis?
- L236: You state that meteorological data variability is taken into account: I would temper this statement, as the climate in such a small area, covered by four contiguous points, probably does not vary greatly, at least you have not demonstrated that it does. I suggest removing this idea of variability. However, it is interesting to note that Gélon covers only four cells of the SAFRAN grid. Further on, you only take one cell (8558) into account in your comparison. You could at least show that the annual precipitation for these four grid points is very similar, which would allow you to take only one into account.
- L252: Figure 2, Medians are extending before 2001 and after 2019: do they use data covering 2000-2020? If not lines must be croped to adjust to 2001-2019.
- L300: the sentence 'highlights the role of weather' is not very adapted and accurate: please rephrase
- L321: (i) and (v) are the same, (v) is the autumn proportion of the framework in low flow
- L327: Fig 5 (a) and (d) respectively
- L329: 'the boxes are more separated' is not very adapted, perhaps change it into 'each year, the departure to the median is systematically more pronounced in (a) as compared to (d)'
- L403: equation should be y=-(3/4)x ; in Figure 7 equations should contain brakets y=-(a/b)x
Edits:
- L222: parameterization
- L166: reservoir
- L176: 1.06 km⁻²
- L250: reference simulation