Modelling the impact of on-farm reservoirs on dry season water availability in an agricultural catchment area of the Brazilian savannah

Abstract

The São Marcos catchment (SAM) encompasses one of the largest center pivot areas in the Southern Hemisphere. Irrigation at SAM is highly dependent on small on-farm reservoirs (SR), specifically at the headwater sub-catchments, where large center pivot irrigation areas are evident. However, the capacity and hydrological performance of those SR are largely unknown as their hydrological characteristics have not been widely assessed. The objective of this work was to establish a method to assess the impact of the SR network over dry season water availability at SAM. For this purpose, a model was developed to test the hypothesis that SR increase water availability at the local, sub-catchment and catchment scales. To this end, SR were mapped with sub-metric satellite images, and their volumes estimated by linear regression with other on-farm reservoirs with known volumes. Streamflow based on four gauged historical data was regionalized in a monthly time step for 4161 sub-catchments. Subsequently, SR were modelled individually, at the local scale, and as a group both at the sub-catchment scale (medium scale) and at the catchment scale (broad scale) in scenarios using Q95 and mean inflows. Results showed that remotely sensed surface area at full supply level (FSL) is the factor that caused the largest variation in the outputs, which has direct correlation with the upstream drainage area discharging to each SR. Simulations suggested that the combination of SR storage capacities have a high proportional impact in increasing water availability at headwater sub-catchments. However, the capacity of a single SR to augment local water availability is more pronounced when its upstream drainage area is 3 km² or greater. Furthermore, assuming there are no inflows of water during the dry season, a single SR must have a minimum FSL of 0.08 ha to result in any improvement in local water availability.