Using high-resolution suspended-sediment measurements to infer changes in the topographic distribution and grain size of bed sediment in the Colorado River downstream from Glen Canyon Dam

Eddy sandbars and other sandy deposits in and along the Colorado River in Grand Canyon National Park (GCNP) were an integral part of the pre-dam riverscape, and are still important for habitat, protection of archeological sites, and recreation. Recent work has shown that eddy bars are dynamic landforms and represent the bulk of the ecosystem's sand reserves. These deposits began eroding following the 1963 closure of Glen Canyon Dam that reduced the supply of sand at the upstream boundary of GCNP by about 94% and are still eroding today. Sand transport in the post-dam river is limited by episodic resupply from tributaries, and is equally regulated by the discharge of water and short-term changes in the grain size of sand available for transport (Rubin and Topping, WRR, 2001). During tributary floods, sand on the bed of the Colorado River fines; this causes the suspended sand to fine and the suspended-sand concentration to increase even when the discharge of water remains constant. Subsequently, the bed is winnowed of finer sand, the suspended sand coarsens, and the suspended-sand concentration decreases independently of discharge. This prohibits the computation of sand-transport rates in the Colorado River using stable relations between water discharge and sand transport (i.e., sediment rating curves) and requires a more continuous method for measuring sand transport. To monitor suspended sediment at higher (i.e., 15-minute) resolutions, we began testing a laser-acoustic system at four locations along the Colorado River in Grand Canyon in August 2002. Because they are much easier to acquire, the high-resolution suspended-sediment datasets collected using the laser-acoustic systems greatly outnumber (by >5 orders of magnitude) direct grain-size measurements of the upstream bed sediment. Furthermore, suspension processes effectively provide an average "sample" of the bed sediment on the perimeter of the upstream channel and the underwater portions of the banks and eddy bars. Thus, it is advantageous to analyze suspended-sediment concentration and grain-size data to infer changes in the topographic distribution and grain size of the upstream bed sediment. Rubin and Topping (2001) developed and tested a theory-based technique that can be used for this purpose. Their parameter "β " is a nondimensional measure of the average bed-surface grain-size that interacts with the suspended sediment in the flow. Analyses of the laser-acoustic datasets indicate that, when the Colorado River is relatively enriched with respect to finer sand, the discharge of water, and the concentration and grain size of the suspended sand are all positively correlated. During these periods, β is negatively correlated with discharge, indicating that the sand on the bed is finer at higher elevations along the banks. Although water discharge and the concentration of suspended sand remain positively correlated when the river is relatively depleted in finer sand, grain size of suspended sand then becomes negatively correlated with both the discharge of water and suspended-sand concentration. The greater decrease in β as a function of discharge demonstrates the decrease in the grain size of the bed sand as a function of elevation is much greater under sand-depleted conditions than under sand-enriched conditions. Thus, these analyses indicate that, during periods of erosion, sand is winnowed preferentially from lower elevations along the river.