Using D-Claw to Inform Lahar Hazard Assessment at Mount Baker Volcano, Washington, USA

Abstract:

Lahars are one of the most destructive volcanic processes worldwide. Over the past two decades, several models have been employed (e.g., LaharZ, Titan 2-D) to forecast lahar inundation areas for hazard mitigation and response purposes. Here we present the results of a new study, using the D-Claw software developed at the USGS, to model lahars resulting from hypothetically derived landslide sources on Mount Baker, Washington. The results are used to improve forecasts of lahar travel times, inundation areas, and flow depths in affected communities. D-Claw uses finite-volume methods and adaptive mesh refinement to solve five depth-averaged hyperbolic partial differential equations. In addition to conservation of mass and momentum, D-Claw solves a novel set of equations for the coevolution of the solid-volume fraction and basal pore-fluid pressure, which regulates debris-flow mobility. Thus, D-Claw simulates flow initiation from increasing pore-fluid pressure to the point of slope failure, rather than instantaneous dam-break conditions with unrealistic force imbalances. The downslope inundation area, travel speed, and runout distance depends on the evolving pore-fluid pressure, which is influenced by the material composition of the lahar.

To test D-Claw simulations against geologic constraints at Mount Baker, we explore the effects of varying the initial hydraulic permeability (10^-12 to 10^-10 m²) and source volume on lahar flow depth and runout. We use two initial source volumes, a hypothetical failure volume (108 x 10⁶ m³; the approximate volume of near vent altered rock) and the bulked volume (260 x 10⁶ m³) of the 6.7 ka Middle Fork lahar, the largest known lahar at Mount Baker. All of our scenarios match the geologic data regarding flow depths of >80 m in a narrow canyon of the Middle Fork Nooksack River about 15 km downstream of source, but only the lowest permeability reproduces the inferred maximum runout distance of 70 km. Nevertheless, each simulation produces a lahar that would inundate communities approximately 30 km downstream of the source area within 20–30 minutes and flow depths would exceed 10 m. Importantly, D-Claw simulations provide useful visual aids for communication of potential lahar inundation areas, flow depths, and arrival times to public officials and communities downstream from Mount Baker.