Decadal-scale effects of a dam removal on channel geomorphology, sediment and large wood on the Elwha River, Washington, USA

Abstract

The removal of Glines Canyon Dam on the Elwha River in western Washington, USA, from 2011 to 2014 introduced a 20-Mt pulse of stored sediment and logs into the downstream channel. We used terrestrial laser scanning, high-resolution orthoimages, and surveys of large wood (LW) and sediment grain-size distribution to quantify changes to the channel and LW in four different geomorphic settings spanning a 16-year period before, during, and after dam removal. The results provide insights into the role of site-specific geomorphology on the interplay among sediment size and supply, wood, and channel form in the aftermath of a dam removal. Sediment-size distribution, braiding index, and number of log jams rapidly reached new steady states. Other factors, such as channel sinuosity and log jam area, were still evolving six years after the dam was removed. The rate and type of river response was partially dictated by the geomorphic setting and the accumulation of LW. Complex reaches trapped more sediment and LW, initiating immediate changes in the channel count, position and lateral migration that continued to evolve through positive feedbacks. Single-channel sites experienced less initial erosion and deposition, but channel migration continued for years once it was underway. The post-dam sediment composition progressively shifted to a mixed size distribution midway between the armored cobbles when the dam was in place and the influx of primarily sand and fine gravel during dam removal. Reworking of sediment was most rapid in the first year after dam removal, especially at the site with the greatest channel complexity. The relation between log jams and channel divisions fundamentally changed. There were 11 log jams in the middle reach of the Elwha River downstream of Glines Canyon Dam, and all log jams associated with channel divisions occurred at the heads of stable, vegetated islands. During dam removal, the number of log jams rapidly increased to 86 and stabilized near that level in the post-dam period. While log jams on stable islands persisted, more were added at divisions around transient sediment bars, scattered across gravel bars, or outside of the active channel. Following a brief spike during dam removal, there was a net long-term increase of ∼10 % in the number of channel divisions associated with log jams. The sediment deposits, LW, vegetation, channel morphology and river discharge continued to cause adjustments within the fluvial system a decade after the start of dam removal. This state of greater variability could be the new equilibrium for years to come.