Erosion of the Yarlung-Tsangpo Gorge sustained by episodic megaflooding

As the Yarlung-Tsangpo river descends through the easternmost Himalaya to join the Brahmaputra river in India, it carves a ~2 km-deep knickzone: the Yarlung-Tsangpo Gorge. Within the gorge, fluvial incision has kept pace with an active crustal scale antiform, exhuming mid-crustal material over Pliocene time at rates up to 7-10 km/Myr. This close spatial correspondence of erosion and exhumation has led previous researchers to hypothesize a self-perpetuating relationship between focused erosion and rock uplift in this ~1200 km^2 region. While high discharge in the river today produces bed stresses that may have the erosional capability to match rock uplift rates, this hypothesis also requires erosion to be localized in the gorge region over timescales spanning multiple glacial periods during which the upstream drainage was episodically dammed by glaciers and discharge was variable. We interpret downstream flood deposits corresponding to this episodic damming as direct evidence of megaflood action as the principal agent of gorge erosion, spanning multiple glacial periods and potentially matching the million-year exhumation rates independently measured in the gorge. Previous researchers documented evidence of glacial damming throughout the fluvial network upstream of the gorge, where moraine and landslide dams are abundant, often beneath dissected lake deposits. We report flood deposits downstream of the gorge, remaining along valley walls at multiple elevations up to 150 m off the modern channel. Using LA-ICPMS U-Pb dating of detrital zircons from these flood sediments, we demonstrate that these high-magnitude flood events disproportionately excavated material from the Yarlung-Tsangpo Gorge, and propose an alternative erosional mechanism to match rock uplift in the gorge over a 10^6 year timescale. We analyzed five flood deposits spanning a range of elevations above the modern channel and four deposits from a documented lower-magnitude flood in 2000. We constrain flood sediment provenance with petrographic analysis, using U-Pb dating to discern the proportion of sediment derived from Himalayan bedrock in the immediate gorge region from that transported through the gorge. Himalayan sourced U-Pb ages are >300 Ma, peaking at ~500 Ma, whereas zircons sourced upstream of the gorge are dominated by the <100 Ma ages from Trans-Himalayan plutons and Tethyan sediments. While the dry-season zircon population downstream of the gorge reflects a ~30-50% bedload enrichment of gorge-derived Himalayan zircons already, flood sediments from high-magnitude events are nearly twice as enriched. Interestingly however, sediments from the lower-magnitude 2000 flood are not demonstrably more enriched, suggesting a threshold flood magnitude necessary for gorge excavation. Including this new data, we propose that it is high-magnitude events, likely during glacial periods, that contribute most of the total erosion of the gorge. Erosion is achieved by a combined process of direct bed incision during flooding and post-flood landsliding of undercut slopes. Our results compliment a growing collection of detrital data from within the Yarlung-Brahamaputra network and contribute to the broader discussion of the influence of high-magnitude, low-frequency events on long-term landscape evolution and how that evolution may be preserved in the sedimentary record.