A streamside landslide delivered 60 000 m3 of mixed-size sediment to the Navarro River, a sinuous gravel-bed channel (drainage area = 535 km2), at the end of the annual high-runoff period in spring 1995. The deposit formed a 9-m-high dam that partially breached within several hours, but recessional flows entrained little material until the following high-runoff season. The landslide afforded the opportunity to measure the evolution of a sediment wave from its inception to near-obliteration and, particularly, to test relative tendencies for translation and dispersion of a sediment wave in a natural gravel-bed channel. This study represents a simple case: The wave originated from a single input, the preexisting channel was relatively uniform, and resistant banks prevented adjustments in width. We surveyed channel topography over a 1.5–4.5 km reach centered on the landslide dam each year from 1995 to 1999, and we sampled bed material downstream of the dam in 1995 and 1997. Landslide material was coarser than ambient bed material, but all sizes were mobilized by subsequent peak flows. Abrasion of weathered and fractured graywacke sandstone landslide material was roughly an order of magnitude greater than the ambient river gravel.
The sediment wave dispersed and mostly disappeared within a few years with no measurable translation. Sediment filled the reservoir created by the eroding landslide dam until throughput of bed load was restored in 1998. The stationary wave crest eroded until in 1999 it was <1 m higher than the preslide elevation. As the wave profile flattened, its detectable leading edge extended downstream from 620 m in 1995 to ∼1600 m in 1997. Downstream advance of the wave was associated with coarsening of bed material.
The sediment wave created a longitudinal disturbance in sediment transport. By using the dam as a reference datum of zero bed-load transport, we computed longitudinal variations in annual bed-load and suspended-sediment transport rates in 100 m increments downstream of the dam. These longitudinal variations were controlled by scour and fill of the bed and by abrasion of bed-load particles. Bed-load transport rates in the first and second years after the landslide increased in the landslide vicinity and then decreased downstream as sediment deposited behind the advancing leading edge of the wave. The location of peak bed-load transport rate advanced from the first year (400 m) to the second (800 m).
We used a physically based, one- dimensional model (Cui et al., 2002b) to hindcast annual changes in transversely averaged bed elevation over the study reach. Agreement between measured and predicted bed elevations was very good. This result supports our conclusion that, once emplaced, sediment waves in gravel-bed rivers tend to disperse, with little or no translation.