EP13A-3506:
Megaflood erosion of the Tsangpo Gorge constrained by hydraulic modeling, geochronology, and geochemical fingerprinting

Monday, 15 December 2014
Michael D Turzewski1, Katharine W Huntington1, Randall J LeVeque1, James K Feathers1, Isaac J Larsen2 and David R Montgomery1, (1)University of Washington, Seattle, WA, United States, (2)California Institute of Technology, Pasadena, CA, United States
Abstract:
The role of extreme floods in long-term erosion and landscape evolution is difficult to study because these events occur infrequently on human timescales and cannot often be observed directly. We address this difficulty using the rich sedimentary record of extreme floods preserved within the Siang River valley of the Eastern Himalaya, which hosts slackwater deposits from historical landslide-dam outburst floods and prehistoric glacial lake outburst megafloods sourced upstream of the Tsangpo gorge in Tibet. Previous workers used detrital zircon U-Pb ages of modern river sediments and flood sands to suggest that megafloods more effectively focus erosion in the Tsangpo gorge than modern peak flows and the year 2000 flood; however, this finding is based on only four, undated megaflood deposits, and a thorough investigation of the history, hydraulics, and erosive impact of different magnitude flood events is lacking.

To expand this record and better understand the relationship between floods, erosion, and deposition, we combined field observations and dating of flood deposits with sediment provenance constraints and numerical flood modeling. We surveyed the trim line from the 2000 flood and identified two historical and 18 ancient slackwater flood deposits. Radiocarbon dates of megaflood deposits range from 1200-1650 14C yr B.P and correlate with upstream glacial lake terraces in Tibet dated by previous workers. Single-grain infrared stimulated luminescence (IRSL) dating of feldspar from megaflood deposits produced ages of 10 to 27 ka that correlate to older glacial lake terraces in Tibet. Using reconstructed glacial and landslide-dammed lakes and modern topography, we simulated the 2000 flood and two megaflood events with GeoClaw, an open source program designed to model geophysical flows. Modeled inundation for the 2000 flood is comparable to the observed flood stage, trim line, and distribution of slackwater deposits. Inundation maps from megaflood simulations provide context for interpreting the distribution of dated megaflood deposits, and modeled water velocity and depth are used to map stream power and bed shear stress to predict zones of erosion and deposition to compare with deposit provenance constraints from new detrital zircon U-Pb ages.