Authors: Colin Phillips , Jose Castejon, Noelle Patterson, Anzy Lee, and Belize Lane – Utah State University; Rebecca Diehl – University of Vermont; Claire Masteller – Washington University in St. Louis
Title: Leveraging River Channel Geometry to Enable Probabilistic Flood Inundation Forecasts
Presentation Type: Lightning Talk and Poster
Abstract: Flooding represents an ever-growing and increasingly tragic environmental hazard. The physics of weather and rainfall-runoff modeling have advanced to where extreme discharges can be forecast. In the United States, the National Water Model (NWM) represents an operational hydrometeorological and flood forecast model for over two million river reaches. Currently, routed discharges are mapped to depth and inundation through Manning’s equation on 10-m topography. However, river flooding is a topographic problem which requires that advances in river and hydraulic processes are translated into operational knowledge to support the decades of hydrometeorological modeling that forms the engine of the NWM. Flood inundation forecasting at the reach scale across a continent remains fundamentally limited by the need to locally calibrate the relation between discharge and depth. Here we leverage the concept of the bankfull flood and channel self-organization to provide a parsimonious calibration procedure. To address topographic calibration, we’ve developed tools that leverage high-resolution topography to extract the elevation of flooding which occurs as a zone of high-curvature within reach-averaged channel and valley geometry. The bankfull discharge is a natural calibration target, however, downstream hydraulic geometry compilations indicate that bankfull discharge spans over six orders of magnitude, while velocity, which is limited by channel stability, appears to vary by a factor of three. We compiled a large dataset of flood velocities to develop a channel stability inspired velocity calibration. Lastly, to account for sub-reach variation in topography, we demonstrate how variability in channel depth can be encapsulated within a log-normal distribution to enable probabilistic flood inundation maps highlighting the potential maximum extent of flood inundation. Paired with our topographic extraction and velocity correction, we present a methodology for locally calibrating rating curves to enable largescale flood inundation forecasting. Accurate flood forecasts are essential to sever the link between natural flood hazards and natural disasters.