Four dams on the Kalamazoo River between the cities of Plainwell and Allegan, Mich., are in varying states of disrepair. The Michigan Department of Environmental Quality (MDEQ) and U.S. Environmental Protection Agency (USEPA) are considering removing these dams to restore the river channels to pre-dam conditions.

This study was initiated to identify sediment characteristics, monitor sediment transport, and predict sediment resuspension and deposition under varying hydraulic conditions. The mathematical model SEDMOD was used to simulate streamflow and sediment transport using three modeling scenarios: (1) sediment transport simulations for 730 days (Jan. 2001 to Dec. 2002), with existing dam structures, (2) sediment transport simulations based on flows from the 1947 flood at the Kalamazoo River with existing dam structures, and (3) sediment transport simulations based on flows from the 1947 flood at the Kalamazoo River with dams removed. Sediment transport simulations based on the 1947 flood hydrograph provide an estimate of sediment transport rates under maximum flow conditions. These scenarios can be used as an assessment of the sediment load that may erode from the study reach at this flow magnitude during a dam failure.

The model was calibrated using suspended sediment as a calibration parameter and root mean squared error (RMSE) as an objective function. Analyses of the calibrated model show a slight bias in the model results at flows higher than 75 m³/s; this means that the model-simulated suspended-sediment transport rates are higher than the observed rates; however, the overall calibrated model results show close agreement between simulated and measured values of suspended sediment.

Simulation results show that the Kalamazoo River sediment transport mechanism is in a dynamic equilibrium state. Model results during the 730-day simulations indicate significant sediment erosion from the study reach at flow rates higher than 55 m³/s. Similarly, significant sediment deposition occurs during low to average flows (monthly mean flows between 25.49 m³/s and 50.97 m³/s) after a high-flow event. If the flow continues to stay in the low to average range the system shifts towards equilibrium, resulting in a balancing effect between sediment deposition and erosion rates.

The 1947 flood-flow simulations show approximately 30,000 m³ more instream sediments erosion for the first 21 days of the dams removed scenario than for the existing-dams scenario, with the same initial conditions for both scenarios. Application of a locally weighted regression smoothing (LOWESS) function to simulation results of the dams removed scenario indicates a steep downtrend with high sediment transport rates during the first 21 days. In comparison, the LOWESS curve for the existing-dams scenario shows a smooth transition of sediment transport rates in response to the change in streamflow. The high erosion rates during the dams-removed scenario are due to the absence of the dams; in contrast, the presence of dams in the existing-dams scenario helps reduce sediment erosion to some extent.

The overall results of 60-day simulations for the 1947 flood show no significant difference in total volume of eroded sediment between the two scenarios, because the dams in the study reach have low heads and no control gates. It is important to note that the existing-dams and dams-removed scenarios simulations are run for only 60 days; therefore, the simulations take into account the changes in sediment erosion and deposition rates only during that time period. Over an extended period, more erosion of instream sediments would be expected to occur if the dams are not properly removed than under the existing conditions. On the basis of model simulations, removal of dams would further lower the head in all the channels. This lowering of head could produce higher flow velocities in the study reach, which ultimately would result in accelerated erosion rates.