Representing a dense network of ponds and reservoirs in a semi-distributed dryland catchment model

•The impact of numerous reservoirs on horizontal hydrological connectivity is modeled.•SWAT satisfactorily represents daily runoff from meso-scale catchments in drylands.•New approaches for parameterization of reservoirs in drylands are presented.•Small reservoirs had a minor influence on stream flow and storage volumes of large reservoirs.•Robust hydrological modeling is at handy for a large underdeveloped dryland. The mismatch between natural water availability and demand in dryland regions is overcome by reservoirs of different sizes with the purpose of storing water. The increase in population in dryland regions and the consequent growth in water demand expanded the construction of small reservoirs, generating in these regions a dense network of reservoirs, which increases the complexity of modeling these hydrological systems. For dryland watersheds modeling with daily time-step, the horizontal connectivity of the reservoir network needs careful representation in order to achieve acceptable model performance, including cumulative effects of reservoirs. However, the horizontal connectivity of reservoir networks is often less investigated in large-scale catchment models. This work presents an innovative way of implementing the dense-reservoir network into the widely used eco-hydrological model Soil and Water Assessment Tool (SWAT), with detailed representation of large and small reservoirs, and an extensive analysis about the cumulative impact of small reservoirs on the horizontal hydrological connectivity for large-scale dryland catchments. A two-fold cross-validation was used against streamflow at a catchment outlet and against in-catchment reservoir water levels. The model daily performance was acceptable despite the input data uncertainty, with good reliability for peak flow in wet years, for nonflow periods and for the rising limb of the hydrograph. The efforts in the parameterization of reservoirs and aggregation of ponds allowed a better analysis of the hydrological processes and their impacts in the catchment. The results showed that small reservoirs decreased the streamflow, but had a low impact on catchment retention and water losses, with 2% of water retention in wet years. However, the water retention reached 9% in dry years, which may worsen periods of water scarcity in the large reservoirs. The spatial representation of small reservoirs for a high-density network in the SWAT model and the results of the cumulative impact of small reservoirs