Homestead Dam Removal: a Natural Scale Experiment in Sandy and Coarse-grained Channels

Removal of a dam provides an opportunity to test current understanding of fluvial processes by creating a known disturbance at a known place and time. Here we present initial findings from a dam removal that provides a natural scale experiment of fluvial processes in both sand-dominated channels as well as gravel- and cobble-dominated channels. The approximately 200-year-old, run-of-river Homestead Dam on the Ashuelot River in southeastern New Hampshire was breached on August 23, 2010. Upstream drainage area at the dam is 820 km2, average channel width is 50 m, and the 2-year discharge is 95 m3/s. This former, 4-meter-high dam had created a ~ 5.5-km long backwater with a water surface slope of about 0.0001 that inundated both coarse-grained bed material (D50 ~ 10 cm) in the first 0.5-km upstream of the dam as well as finer-grained sandy bed forms (comprising sediment with D50 ~ 2 mm) from 0.5-km to 8-km upstream. Thick alluvium, from 2-m to greater than 4-m, mantles glacial till throughout the impoundment. Prior to removal of the dam, we obtained aerial lidar data in a 2- by 14-km swath centered at the dam; measured 42 cross sections over 14-km up- and downstream of the dam to document channel depth, width, slope, and bed grain size; sampled 210Pbex and 7Be fallout radionuclide activity of bed material at cross sections for potential use as a tracer of sediment movement, and measured alluvium thickness using ground-penetrating radar for 6-km upstream of the dam. The initial breaching of the dam exposed a historic riffle from the former structure to 0.5-km immediately upstream of the dam. Over this 0.5-km reach of the river, the water surface slope increased from 0.0001 to 0.004 due to the lowering of the water level at the breached dam. During the week following the dam breaching, flows never exceeded 2.5 m3/s, and measured sediment transport and channel form adjustment were insignificant at cross sections in this proximal reach. In contrast, sediment transport in short reaches (~30-m to 200-m long) surrounding dune crests was observed during the first week between 4-km and 7-km upstream of the former dam. In these locations, the lowering of the downstream water surface increased the slope across dune crests and initiated transport of the mobile, sandy bed. We anticipate additional, more significant changes in channel form following high flow events. Future measurements of channel geometry will be utilized to test hypotheses on covariate changes in channel slope, grain size, width, depth and sediment transport. The results of this study will improve predictions of channel and habitat change following dam removal, which has become an increasingly common watershed restoration tool.