The Secor dam was a low-head dam that was 17m wide and 2.5 m tall on the Ottawa River in Toledo, Ohio. The dam was removed for liability reasons and to facilitate improvements in water quality and fisheries habitat on November 19, 2007. This study documents the fluvial response of dam removal using: (1) repeated high resolution channel surveying with a total station, (2) differential GPS measurements of bedform migration, (3) bedload sampling using a Helley-Smith bedload trap, (4) multiple cores and (5) grain size analyses of the channel substrate. The research also examined the applicability of the conceptual channel evolution model of Doyle et al. (2003) and the predictive success of Dam Removal Express Assessment model (DREAM-1) of Cui et al. (2006).
The results highlight the impact of reservoir sediment characteristics in fluvial responses to dam removal. The Secor dam only trapped sand-sized bedload, thus rather than the initial flush of suspended load sediment, the response was rapid incision and mobilization of the material from the upstream delta of the former impoundment. The breaching of the dam resulted in rapid upstream migration of a diffuse nick-zone approximately 10-m in length and downstream migration of a sediment wave that translated at rates up to 0.5 m/hr. Within five months, an estimated 514-m3 of sediment had been removed from the former delta and was redistributed into pools immediately upstream and downstream of the former dam.
Channel incision was the dominate process over the first two weeks of the study, with net incision as much as approximately 1-m in some locations. One month after removal, the channel began to slowly widen and widening has been the dominate process during the subsequent 4 months.
The Secor dam removal differed fundamentally from other dam removals in the literature for the following reasons: (1) the impounded sediment was relatively homogenous, (2) a well defined channel already existed behind the impoundment, (3) the substrate was cohesionless and (4) incision was dominated by nick-zone migration not nick-point migration. As a result, channel evolution occurred quickly, mobilizing sediment from the former delta almost immediately, as opposed to other studies which have reported erosion from the upstream delta taking years to decades. Thus, the channel evolution model proposed by Doyle et al. (2003) failed to predict the initial phases of removal of the Secor dam, which was dominated by uniform degradation of the channel bed behind the dam instead of evolving via nick-point migration.
The DREAM model adequately predicted the net volume of sediment removed from the former impoundment, only differing from the estimated value by -1.7 % The model, however, may not be used during the later phases of dam removal because the model failed to predict channel widening which may result in an underestimation of the volume of sediment removed from the former impoundment.