Reconciling Geomorphic Observations with Simulations of a Modern Landslide-dam Outburst Flood Using GeoClaw Software, Eastern Himalaya

Abstract:

High-magnitude (>10⁵ m³/s) outburst floods have the potential to dramatically alter landscapes and greatly impact human lives and infrastructure. Numerical modeling can help us understand the hydraulics of these infrequent and difficult to observe floods, but their scale makes simulation challenging and computationally expensive, particularly where rugged mountain topography produces complex flow hydraulics. Here we simulate the second largest historical outburst flood on record using GeoClaw open source software for modeling geophysical flows, and ground-truth the results of these simulations using observations and geomorphic evidence of the event. This landslide-dam outburst flood was sourced in Tibet on the Yigong River in June 2000, scouring vegetation, triggering landslides and depositing flood sands in hydraulically sheltered areas downstream. We mapped these features in the field and remotely using Google Earth and Landsat-7 imagery, and simulated the flood with a reconstructed 2 Gm³ impounded lake using instantaneous dam failure. Simulated inundation and the arrival-time of the flood wave downstream are relatively insensitive to the Manning bed-roughness parameter implemented in GeoClaw, but are very sensitive to grid-resolution and the chosen Adaptive Mesh Refinement scheme. High-resolution simulations produce estimates of discharge and the arrival-time for the initial flood wave that compare favorably to observations of the event recorded at locations up to 450 km downstream, and inundation maps that match the mapped distribution of high water marks. GeoClaw simulations (1) show inundation and decreasing bed shear stresses during the waning stage of the flood in the areas that contain observed slackwater deposits and (2) produce sustained deep flows in regions where landslides were observed directly after the event, showing a clear link between flood hydraulics and geomorphic change due to erosion/deposition. Results suggest that GeoClaw can accurately simulate high-magnitude outburst flood events through mountainous topography, showing the potential of this modeling approach to improve hazard predictions and our understanding of the geomorphic impact of outburst floods.