Geomorphic response of the North Fork Stillaguamish River to the State Route 530 landslide near Oso, Washington

Abstract

On March 22, 2014, the State Route 530 Landslide near Oso, Washington mobilized 8 million cubic meters of unconsolidated Pleistocene material, creating a valley‑spanning deposit that fully impounded the North Fork Stillaguamish River. The river overtopped the 8-meter high debris impoundment within 25 hours and began steadily incising a new channel through the center of the deposit. Repeat topographic surveys, sediment transport measurements, bedload transport models, and observations of downstream channel change were used to document the establishment of that new channel through the landslide and assess the potential for downstream aggradation or channel change that might increase downstream flood hazards. Efficient erosion of the landslide deposit, associated with the steep knickzone formed by the downstream edge of the deposit, resulted in the re-establishment of a 20–40 meters wide, deeply inset channel through the entire deposit by May 2014, 2 months after the landslide. The mean water-surface elevation of the channel through the landslide decreased 7 meters during that 2-month period, and was about 1 meter above the pre-landslide profile in July 2014. The 2014–15 flood season, which included flows near the 0.5 annual exceedance probability discharge (2-year flood), widened the channel tens of meters, and further lowered the water-surface profile 0.5 meter. The planform position evolved slowly as a result of 5–20-meter high banks predominantly composed of clay-rich, cohesive lacustrine material. Erosion of the landslide deposit delivered a total of 820 thousand metric tons of sediment to the North Fork Stillaguamish River over the 18 months following the landslide. The sediment delivery from the deposit was predominantly fine grained: 77 percent (by mass) of the eroded material was silt or clay (less than 0.063 millimeter [mm]), 19 percent sand (0.063–2 mm), and 4 percent pebbles and cobbles (greater than 2 mm). Over the 18 months following the landslide, the bedload at a site 5 kilometers downstream of the landslide was estimated to be 310±65 thousand metric tons, and the suspended load at that same site was estimated to be 990±110 thousand metric tons. These loads represent the combined input from the landslide and ambient upstream sources; over the study interval, landslide sediment made up about 20–40 percent of the bedload, and 65–85 percent of the suspended-sediment load at this site. At a site 70 kilometers downstream of the landslide, near the mouth of the main‑stem Stillaguamish River, suspended sediment loads were estimated to be about 1,440 thousand metric tons, of which about 600 thousand metric tons, or 30 percent, likely was derived from the landslide. The mass of landslide sediment in suspension at the mouth of the river, and the timing of arrival of that sediment, indicates that about 70 percent of the landslide sediment eroded during the study period was quickly transported through the entire basin, exiting into Puget Sound within weeks of initial entrainment. Empirical bedload transport equations, in conjunction with surficial grain-size data and output from a one‑dimensional hydraulic model, were used to estimate spatial trends in bedload transport capacity, highlighting areas where reach-scale conditions would be most likely to promote deposition of coarse landslide sediment. Transport capacities decreased sharply over a reach about 5 kilometers downstream of the landslide and remained relatively low over the next 10 kilometers downstream. However, the magnitude of calculated transport capacities are large relative to the coarse sediment input from the landslide, suggesting that substantial deposition of landslide sediment was not likely to occur. These assessments were corroborated by observations of channel change, which indicated that the downstream channel response to the landslide was modest and short-lived. The most pronounced downstream effects included a wedge of aggradation just downstream of the landslide, about 1 meter high and extending a kilometer downstream, and a 0.3-meter pulse of aggradation observed 5 kilometers downstream of the landslide. In both locations, peak aggradation and channel response occurred within about a month of the landslide, and both sites had largely recovered to pre-landslide conditions by July 2014. No substantial channel change clearly linked to the landslide was observed after July 2014 except for a modest fining of surficial gravel size distributions and continued recovery and incision of the reach just downstream of the landslide. The muted downstream response of the North Fork Stillaguamish River to the State Route 530 Landslide primarily can be attributed to the cohesive, silt- and clay-rich material that bounded most of the new channel. Although the river efficiently incised a new channel through the deposit, subsequent rates of lateral erosion were slowed by the tall, cohesive banks, limiting the total volume of sediment delivery. Once entrained, however, most landslide material was rapidly transported downstream in suspension with little geomorphic effect. Landslide material coarse enough to travel as bedload was predominantly sand and fine gravel, and sediment transport models and observations of downstream change indicated that the rate of coarse sediment delivery from the landslide did not exceed the rivers ability to transport that material. The generally muted downstream response to sediment delivery from the State Route 530 Landslide, as well as the mechanics of that delivery and response, were generally consistent with observations made following the intentional removal of constructed dams. The rate and efficiency of erosion from the landslide decreased over the period of analysis, as the new channel approached a quasi-equilibrium form. In the absence of additional hillslope activity, rates of erosion from the landslide are likely to be small compared to those over the first 18 months after the landslide. The modest channel response to the highest rates of sediment delivery, and rapid recovery thereafter, indicate that the river should be able to convey the continued supply of landslide-derived sediment effectively with little effect on the downstream morphology and flood risks.