Authors: Anzy Lee – Utah State University
Title: Revising stream segmentation in the hydrofabric to improve OWP flood inundation mapping
Presentation Type:
Abstract: Improving the skill of hydrologic and flood inundation modeling (FIM) as part of NextGen requires accurate representation of the underlying hydrofabric, the spatial data architecture that describes key channel and landscape features. Currently, the NextGen hydrofabric defines reaches hydrologically, relying on catchment and flowline boundaries dictated by major surface water inputs refactored to optimize path length. This approach does not account for longitudinal changes in important attributes for flooding such as channel slope, valley width and hydraulic geometry. We aim to improve the predictive skills of the Office of Water Prediction’s terrain-based height above nearest drainage (HAND) FIM within the existing terrain-based framework by refining the computational scales used to represent stream reaches and hydraulic attributes in the hydrofabric to reflect downstream changes in hydraulic geometry.
We demonstrate that HAND-FIM performance can be improved by adding select breakpoints to the hydrofabric. We introduce an automatable workflow for hydraulics-informed network segmentation based on significant downstream changes in floodplain width, slope and other hydraulic attributes based on various terrain and remote sensing products, segmentation parameters (e.g. significant difference threshold, minimum reach length), and change detection algorithms. The impact of alternative segmentation schemes on HAND-FIM performance is sensitive to the benchmark flood size and HAND resolution. The proposed data-driven framework provides a pathway for refining reach segmentation to better capture hydraulic variability and enhance large-scale FIM forecasting.