Quantifying the downstream impacts of dams with hydrologic signatures and a large-scale hydrologic model




Liu, Yiran

Journal Title

Journal ISSN

Volume Title



Rivers are regulated worldwide and are segmented by dams, primarily built in the last century. The impact of dams on flow immediately downstream has been studied for decades. However, the downstream propagation of the hydrologic impacts in a river network still needs to be better quantified, particularly in humid regions. This study aims to test a method of investigating the impacts of large reservoirs by using different hydraulic signatures, determining how the signatures propagate downstream of the dams, and exploring the potential to utilize a hydrologic model to determine dam impacts. By solely analyzing streamflow observations in the Delaware River Basin, hydrologic signatures can detect the dams’ impacts immediately downstream of the New York City reservoirs and are able to show the length scale at which hydrologic signatures return to unimpacted values within the Delaware River basin. Results show that the streamflow downstream was influenced by the large reservoirs but with a propagation distance of approximately 35 kilometers, after which the signatures are likely to be consistent with gauges upstream of the reservoirs. Unlike the western U.S., the streamflow below large dams shows the potential to recover via streamflow contributions from the tributaries in a humid watershed, resulting in a less vulnerable river regime; however, the flashier streamflow and the change in timing locally downstream of the dams may result in ecological degradation. This method developed for the Delaware basin to compare hydrologic signatures in time and space is transferable to other river basins. Another potential approach is to use a hydrologic model to simulate unregulated streamflow which can serve as a proxy for naturalized hydrologic signatures. The Variable Infiltration Capacity (VIC) model was selected to explore the possibility of expanding the observation-based method. Due to the poor streamflow performance in an initial simulation, especially for the baseflow, the second chapter of this thesis focuses on better understanding the VIC baseflow generation and its sensitivity to model parameters. To do this, seven calibration parameters with three different values each are selected for a sensitivity analysis. The results suggest that the infiltration parameter (infil), the exponent in the equation for hydraulic conductivity (exp), and the parameter that specifies the maximum baseflow (dsmax) play important roles in determining the baseflow simulation and also influence the baseflow sensitivity of other parameters. This emphasizes the importance of precipitation partitioning and the vertical drainage process in baseflow simulation, in addition to the actual baseflow curve for the bottom soil layer. Based on the results of the sensitivity analysis, VIC is expected to be able to correctly simulate baseflow and total streamflow after calibration but this modeling exercise revealed significant challenges and limitations. Typical calibration methods that rely on monthly streamflow need to be improved, while calibrating to daily streamflow is important because of the need to capture the hydrograph characteristics, like the correct recession curve. Spatial heterogeneity needs to be considered, and the calibration parameters may not be transferable between gauges, even at a relatively small distance. If VIC can represent the naturalized streamflow regime, there is a potential to do a similar tracing downstream as in Chapter 2. Both methods could then be applied to other basins.



streamflow, Delaware River Basin, dam impact, dam impact propagation, NYC reserviors, hydrologic siganitures