EP33C-1082
Reconstructing Tsunami Deposits in the Eastern Aleutians Using Forward and Inverse Sediment Transport Models
Wednesday, 16 December 2015
Poster Hall (Moscone South)
Guy R Gelfenbaum, US Geological Survey, Pacific Coastal and Marine Science Center, Santa Cruz, CA, United States, SeanPaul La Selle, Pacific Coastal and Marine Science Center Santa Cruz, Santa Cruz, CA, United States, Bruce E Jaffe, USGS Pacific Coastal and Marine Science Center Santa Cruz, Santa Cruz, CA, United States and Robert Carleton Witter, USGS Alaska Science Center, Anchorage, AK, United States
Abstract:
Tsunami deposits on coastal plains are commonly observed to gradually thin inland and contain upward fining sand units. These characteristics help validate the assumptions (steady and uniform onshore flow and sediment settling from suspension) that are employed by inverse sediment transport models, which predict flow speed from thickness and grain size data. On Sedanka Island in the eastern Aleutian Islands, a sequence of 6 tsunami deposits from the last 1700 years have been described that extend across an 800 m strandplain and reach elevations up to 15 m. The youngest deposit is attributed to the 1957 Andreanof Island earthquake (Mw 8.6) and is 1-13 cm thick. The older deposits are thicker (6-50 cm) and all of the layers contain upward fining sand units. Although the total volume of sediment varies among deposits, they all thicken landward, toward the back of the valley. We developed a Delft3D forward tsunami sediment transport model to better understand the conditions that resulted in this spatial pattern of deposit thickening. Results from a profile model suggest that sediment eroded from the beach and berm is transported to the back of the valley during uprush. Significant deposition does not occur until the initial wave reflects off the steep topography at the back of the valley and the local flow velocity drops, with the highest rate of deposition occurring during slow return flow as sediment settles out of the water column. A 3D model will be used to determine if the funnel-shape of the valley produces convergences that can explain the observed deposit thickening, and to see if a majority of the deposition still occurs during return flow. Finally, we will use flow depths from the forward model to constrain a TSUSEDMOD inverse model and see if it can reproduce the observed deposit grading and modeled flow velocities. If so, the inverse model may be applied to deposits in other locations where similar hydrodynamic conditions are suspected occur.