ENTRAINMENT OF BED SEDIMENT BY DEBRIS FLOWS: LARGE-SCALE EXPERIMENTS

Exceptionally voluminous, mobile, and destructive debris flows can result from the entrainment of sediment during flow, but quantifying controls on entrainment can be difficult in the field. As an alternative, we conducted nine large-scale experiments in the USGS debris-flow flume in Oregon. In each experiment, we placed a 10-15 cm thick layer of loose, loamy sand and gravel over a roughened concrete bed in the upslope half of the 95-m long, 2-m wide, 31º flume. We then used overhead sprinkling combined with real-time monitoring of 16 to 23 buried moisture sensors and tensiometers to systematically vary the volumetric moisture content of the bed sediment between 15% and 28%; at higher moisture contents the bed began to fail spontaneously as small positive pore pressures developed at its base. To initiate a debris flow, we released 6 m³ of saturated loamy sand and gravel from the flume headgate. As this flow interacted with the erodible bed, we measured sediment scour using nests of buried electronic tripwires, and recorded basal total normal stress, shear stress, and pore-water pressure. We also measured flow thickness using overhead lasers, flow-front speeds, runout distances, and volumes of entrained sediment.

Moisture content of the bed sediment dramatically influenced both the entrainment process and flow dynamics. When debris flows traveled over wet bed sediment, nearly all was entrained and the debris flow volume grew 2- to 3-fold after only ~50 m of travel. Flow front speeds (~11m/s) were often 10-20% faster than those of similar debris flows without entrainment, and runout distances at the base of the flume roughly doubled. In stark contrast, when debris flows traveled over drier sediments, net bed sediment scour was minimal (5-30%), and flow front speeds and runout were greatly reduced. Overall, the net volume of sediment scoured increased monotonically with increasing sediment moisture content.

Data from the electronic tripwires showed that scour occurred progressively rather than en masse. Given wet bed sediment, the scour front progressed into the sediment very rapidly, with complete removal of the 10-15 cm layer in less than 1 s. This process was facilitated by development of positive pore pressures measured in the bed sediment as it was rapidly loaded by the overriding debris flow. In contrast, pore pressure development in drier bed sediment was minimal.