Channel response to sediment release: insights from a paired analysis of dam removal

Dam removals with unmanaged sediment releases are good opportunities to learn about channel response to abruptly increased bed material supply. Understanding these events is important because they affect aquatic habitats and human uses of floodplains. A longstanding paradigm in geomorphology holds that response rates to landscape disturbance exponentially decay through time. However, a previous study of the Merrimack Village Dam (MVD) removal on the Souhegan River in New Hampshire, USA, showed that an exponential function poorly described the early geomorphic response. Erosion of impounded sediments there was two-phased. We had an opportunity to quantitatively test the two-phase response model proposed for MVD by extending the record there and comparing it with data from the Simkins Dam removal on the Patapsco River in Maryland, USA. The watershed sizes are the same order of magnitude (10² km²), and at both sites low-head dams were removed (~3–4 m) and ~65 000 m³ of sand-sized sediments were discharged to low-gradient reaches. Analyzing four years of repeat morphometry and sediment surveys at the Simkins site, as well as continuous discharge and turbidity data, we observed the two-phase erosion response described for MVD. In the early phase, approximately 50% of the impounded sediment at Simkins was eroded rapidly during modest flows. After incision to base level and widening, a second phase began when further erosion depended on floods large enough to go over bank and access impounded sediments more distant from the newly-formed channel. Fitting functional forms to the data for both sites, we found that two-phase exponential models with changing decay constants fit the erosion data better than single-phase models. Valley width influences the two-phase erosion responses upstream, but downstream responses appear more closely related to local gradient, sediment re-supply from the upstream impoundments, and base flows.