Monitoring fluvial erosion of the Oso landslide, Washington, using repeat Structure-from-Motion photogrammetry
Friday, 18 December 2015
Poster Hall (Moscone South)
On March 22, 2014,
the Oso landslide, located in northwestern Washington State, catastrophically mobilized about 8 million m3
of mixed glacial sediment, creating a valley-wide blockage that impounded the North Fork Stillaguamish River to a height of 8 m. The river overtopped the landslide blockage within several days and incised a new channel through predominately fine-grained, cohesive glaciolacustrine sediment in the center of the deposit. Our research focuses on the evolution of this new channel. Using a consumer-grade digital camera mounted on a fixed wing-aircraft, we used structure-from-motion (SfM) photogrammetry to produce 25 cm digital elevation models (DEMs) of the channel at one-month intervals between November 2014 and July 2015. A large RTK GPS validation dataset and inter-survey comparisons documents sub-decimeter vertical and horizontal accuracies. In combination with aerial lidar surveys acquired in March and April 2014, this dataset provides a uniquely resolved look at the erosion of a landslide dam. The newly-formed channel incised rapidly, lowering to within a meter of its pre-slide elevation by May 2014 despite modest flows. During high flows of the 2014-2015 winter flood season, erosion was dominated by channel widening of tens of meters with an overall stable planform. Incision fully returned the channel to pre-slide elevations by December 2014. A total of 510,000 +/- 50,000 m3
of material was eroded between March 2014 and July 2015, split evenly between the initial period of incision and the later period of widening. Sediment yield and channel morphology showed asymptotic trends towards stability. Measurements of deposit bulk density and grain size allowed conversion of volumetric sediment yields to mass yields by size classes. Over the 16 months after the slide, the river eroded about 0.82 +/- 0.1 Mt of sediment, of which 0.78 Mt was finer than 2mm. This yield agrees within 15% of an independent estimate based on concurrent sediment gaging in the reach, and represents about 400% of the background sediment yield over that same period.