Estimation of River Channel Shape Using Big Data-Driven Approaches

Abstract

Enhanced precision in channel geometric parameters has the potential to significantly improve hydrologic modeling and the generation of flood inundation mapping (FIM). Emerging datasets containing large-scale measurements of channel morphological and hydraulic characteristics offer pathways for developing novel models estimating channel cross-sectional geometry parameters, including shape. This project develops first-order channel geometry parameter models based on large-scale datasets for the CONtiguous United States (CONUS). Four methods of reconstructing simplified channel cross-sections, using different channel shapes, from observational at-a-station data are compared to observed channel cross-sections. Statistical analysis of the channel geometry parameters cross-sectional area, wetted perimeter, and hydraulic radius are used to identify the most accurate reconstruction approach. Two reconstruction methods use the common trapezoidal channel shape and the other two implement curved channel shape parameters. The Moramarco’s w reconstruction, which derives its channel shape parameter based on the ratio of mean to maximum depth observations at a site, is identified as the best channel reconstruction. The cross-sections generated from this optimal reconstruction are used for the development of models of the channel geometry parameters cross-sectional area and hydraulic radius, which can be applied for the NHDPlus V2.1 stream network to provide bathymetry estimations across the entire CONUS.