FLUVIAL EROSION AND INUNDATION HAZARD ASSESSMENTS FOR VERMONT USING HIGH-RESOLUTION TOPOGRAPHY, A LOW-COMPLEXITY HYDRAULIC MODEL, AND THE PROCESS DOMAIN CONCEPT

Extreme discharge events in Vermont are projected to occur more frequently with climate change, elevating the need for accurate identification of vulnerable portions of the river network. Yet, flood inundation mapping is limited in coverage, and characterization of the potential for fluvial erosion is often absent altogether from regional flood hazard assessments. Predicting fluvial hazards over large spatial domains with fully coupled hydraulic and sediment transport models is a challenge due to the high overhead involved in their reach-by-reach construction. In this project, we aim to predict both inundation and erosion hazards at regional scales through the combination of a low-complexity hydraulic model paired with a machine-learning predictive model of sediment process domains. Sediment process domains describe the tendency of a river reach to source, transport, and deposit sediment by grain texture as well as the degree of connectivity between the channel and adjacent floodplains. We extend process domain predictions regionally with data-driven methods that leverage geomorphological field data collected from diverse rivers across Vermont. We use a probabilistic height-above-nearest-drainage inundation model to predict flood inundation hazards and to extract features that describe the interplay between discharge and hydraulic radius along a given reach. Both models rely on high-resolution topography to identify reach-scale variables relevant to lateral flood dissipation and stream power generation under varying hydrological forcings in order to predict fluvial erosion hazards at regional scales. This workflow is generalizable and will enable low-cost mapping of process-domain-specific fluvial hazards in support of community and watershed resilience efforts.