The impact of run-of-river dams on sediment longitudinal connectivity and downstream channel equilibrium

Abstract

Considerable research over the past several decades shows that dams, especially large, flow regulating structures, fragment watersheds and serve to disconnect the normative downstream flux of sediment and nutrients. Less attention has addressed smaller, channel-spanning Run-of-River (RoR) dams that are more commonly distributed throughout watersheds. Taking advantage of a suite of RoR dams in New England (USA), we quantify bedload flux into, through, and beyond the reservoir, and we calculate the residence time of gravel clasts.We used traditional channel surveys to evaluate (dis)equilibrium channel form and develop two equilibrium metrics based on bankfull shear stress and the bankfull Shields parameter. Additionally, we compartmentalize the bankfull channel Shields parameter as a linear combination of bedload and suspended load components to better quantify channel evolution in response to changes in sediment supply. To accomplish these goals, we embedded Radio Frequency Identification (RFID) PIT tags in 791 gravel clasts ranging in size from 15 mm to 81 mm, which were subsequently deployed within and upstream of the impounded reservoirs. Among the 503 tracers that were transported from their deployment location, the median cumulative distance traveled was 30 m and the maximum cumulative displacement during the study period was 758 m. Of the total tagged rocks placed at all five sites, 276 rocks were displaced over the dam, 204 of which spent time in the reservoir between high discharge events; the rest were transmitted through the reservoir and over the dam in a single high discharge event. Among those tracers that spent time in the reservoir prior to transmission over the dam, the average reservoir residence times at the different sites ranged from 19 to 203 days. The median grain sizes of tracers that were transported over the dam were identical to those that moved during the study period and similar to the median grain size of the channel bed. The distribution of virtual velocities of those tracers that moved was approximately log-normal and very broadly distributed over more than six orders of magnitude. An analysis of variance revealed that the distribution of velocities was partitioned into two statistically similar groups; with slower velocities in the two smaller watersheds (13 km2–21 km2) with higher average slopes compared to the larger watersheds (89 km2–438 km2) with lower average slopes. We conclude that RoR dams transmit and trap the upstream sediment supply within the same range of physical conditions that produce mobility and trapping in the river's natural reach-scale morphological units. Because RoR dams are likely not trapping more sediment than is typically sequestered in natural river reaches, these dams do not disconnect the normative downstream flux of sediment nor result in channel morphological disequilibrium downstream of the dam. Reaches below RoR dams have similar geomorphic properties to comparable equilibrium reaches unaffected by dams. However, the minimal effect that small, channel spanning RoR dams have on the morphological equilibrium state of a channel does not suggest that RoR dams have no ecological footprint.