Erosion dynamics of a stepwise small dam removal, Brewster Creek Dam near St. Charles, Illinois

Abstract

A stepwise dam-removal project, consisting of cutting five 12- to 18-in. notches in the dam, was completed from June 2003 through February 2004. The study area included Willow Lake on Brewster Creek near St. Charles, Illinois. The dam was 8-ft high and had a 30-ft wide crest. In 2002, the lake had a surface area of 3.96 acres and had filled with sediment to an average water depth of less than 1 ft, sediment volume equaling 14.47 acre-ft, and average sediment thickness of 3.7 ft. The deposited sediments consisted of 67 to 99 percent silts and clays. The stream channel downstream of the dam consisted primarily of sand, gravel, and cobble. The total sediment eroded during and after dam removal was approximately 13 percent of the lake sediment. The downstream normalized sediment yield (sediment yield divided by the mean-daily flow) during the notchings and the first 7 months after the notchings (June 15, 2003, through September 30, 2004) was 14 tons/mi2-yr-ft3/s. From October 1, 2004, through September 30, 2006, the downstream normalized sediment yield (6.5 tons/mi2-yr-ft3/s) was within 17 percent of the upstream normalized sediment yield (5.4 tons/mi2-yr-ft3/s). During and after dam removal, a knickpoint established and developed upstream of the dam. A flow model was applied to determine the shear stress acting upon the knickpoint for five surveyed conditions. The maximum knickpoint shear stress for each condition was always a result of flow between 50 ft3/s and the 1.5-yr streamflow (139 ft3/s). The backwater effects caused by the remnant abutments and embankments limited the shear stresses acting on the knickpoints for the two higher streamflows modeled (10- and 100-yr streamflows, 477 ft3/s and 920 ft3/s, respectively). Empirical equations were determined to relate the knickpoint shear stress with eroded sediment for sixteen storms from November 2003 through February 2005. A Cohesive Knickpoint Parallel Retreat method was developed utilizing the erosion-rate equation determined in this research from the cohesive-erosion laboratory testing results. The method was applied to the Brewster Creek study and predicted the eroded volume, knickpoint height, and migration length within 17, 20, and 6 percent, respectively, of measured values.