The 1959 Hebgen Lake earthquake caused a massive landslide (Madison Slide) that dammed the Madison River and formed Earthquake Lake. The U.S. Army Corps of Engineers excavated a spillway through the Madison Slide to permit outflow from Earthquake Lake. In June 1970, high streamflows on the Madison River severely eroded the spillway channel and damaged the roadway embankment along U.S. Highway 287 downstream from the Madison Slide. Investigations undertaken following the 1970 flood events concluded that substantial erosion through and downstream from the spillway could be expected for streamflows greater than 3,500 cubic feet per second (ft³/s). Accordingly, the owners of Hebgen Dam, upstream from Earthquake Lake, have tried to manage releases from Hebgen Lake to prevent streamflows from exceeding 3,500 ft³/s measured at the U.S. Geological Survey (USGS) gaging station 0638800 Madison River at Kirby Ranch, near Cameron, Montana.

Management of flow releases from Hebgen Lake to avoid exceeding the threshold streamflow at USGS gaging station 06038800 is difficult, and has been questioned for two reasons. First, no road damage was reported downstream from the Earthquake Lake outlet in 1993, 1996, and 1997 when streamflows exceeded the 3,500-ft³/s threshold. Second, the 3,500-ft³/s threshold generally precludes releases of higher flows that could be beneficial to the blue-ribbon trout fishery downstream in the Madison River.

In response to concerns about minimizing streamflow downstream from Earthquake Lake and the possible armoring of the spillway, the USGS, in cooperation with the Madison River Fisheries Technical Advisory Committee (MADTAC; Bureau of Land Management; Montana Department of Environmental Quality; Montana Fish, Wildlife and Parks; PPL-Montana; U.S. Department of Agriculture Forest Service - Gallatin National Forest; and U.S. Fish and Wildlife Service), conducted a study to determine movement of the Madison River channel downstream from Earthquake Lake and to investigate the potential for bed material movement along the same reach. The purpose of this report is to present information about the lateral and vertical movement of the Madison River from 1970 to 2006 for a 1-mile reach downstream from Earthquake Lake and for Raynolds Pass Bridge, and to provide an analysis of the potential for bed-material movement so that MADTAC can evaluate the applicability of the previously determined threshold streamflow for initiation of damaging erosion.

As part of this study channel cross sections originally surveyed by the USGS in 1971 were resurveyed in 2006. Incremental channel-movement distances were determined by comparing the stream centerlines from 14 aerial photographs taken between 1970 and 2006. Depths of channel incision and aggregation were determined by comparing the 2006 and 1971 cross-section and water-surface data. Particle sizes of bed and bank materials were measured in 2006 and 2008 using the pebble-count method and sieve analyses. A one-dimensional hydraulic-flow model (HEC-RAS) was used to calculate mean boundary-shear stresses for various streamflows; these calculated boundary-shear stresses were compared to calculated critical-shear stresses for the bed materials to determine the potential for bed-material movement.

A comparison of lateral channel movement distances with annual peak streamflows shows that streamflows higher than the 3,500-ft³/s threshold were followed by lateral channel movement except from 1991 to 1992 and possibly from 1996 to 1997. However, it was not possible to discern whether the channel moved gradually or suddenly, or in response to one peak flow, to several peak flows, or to sustained flows. The channel moved between 2002 and 2005 even when streamflows were less than the threshold streamflow of 3,500 ft³/s.

Comparisons of cross sections and aerial photographs show that the channel has moved laterally and incised and aggraded to varying degrees. The channel has developed meander bends and has incised as much as 5–12 feet (ft) through the upstream part of the Madison Slide (cross sections 1400–800). Near cross section 800, the stream has eroded into the steep right bank between the stream and the road where fill was mechanically placed after 1970. Channel movement also was noted downstream from the Madison Slide.

Near Raynolds Pass Bridge, about 3 miles (mi) downstream from Earthquake Lake, elevations across the channel have changed by -1.4 ft to +1.9 ft, but these changes were local in nature and could represent a few rocks or depressions in the bed. Overall, it does not appear that the materials eroded from the Madison Slide are causing aggradation in the subreach near the Raynolds Pass Bridge.

Comparisons of critical shear stresses to mean boundary-shear stresses indicate that the D50 particle sizes (median size) along the right side of the bed between cross sections 400 and 500 and along the right side of the bed between cross sections 1300 and 1400 could move at the threshold streamflow. In contrast, most of the D84 particle sizes at those two locations probably will not move at the threshold streamflow. This lack of movement for the larger particles at the threshold streamflow could lead to further armoring of the bed as the D50 and smaller-sized particles are removed from the bed and transported downstream.

The Shields parameter values from 0.04 to 0.08 that were used to calculate critical shear stresses could be conservative for a high-gradient stream such as the Madison. A higher, less conservative, Shields parameter would result in higher critical shear stresses, meaning that higher streamflows would be required to move material than those reported herein. In addition, because materials in the channel thalweg are exposed to higher boundary-shear stresses than the materials along the sides of the channel, larger, more erosion-resistant materials likely exist in the deeper parts of the channel where high-flow depths and velocities prevented sediment sampling. Movement of these materials might require higher critical shear stresses than estimated in this report. Characterization of sediment sizes in the center of the stream and observation of bed-material movement for a range of streamflows could provide information to help refine the Shields parameter and critical-shear stress estimates for bed materials in the Madison River downstream from Earthquake Lake. Furthermore, resurveying cross sections and water-surface elevations more frequently (either annually or after high streamflows) could better define the relation between streamflow and lateral and vertical channel movement.