U.S. Environmental Protection Agency


Date of this Version



Published in Geomorphology 108 (2009) 292–311. Doi:10.1016/j.geomorph.2009.02.005


Bankfull channel width is a fundamental measure of stream size and a key parameter of interest for many applications in hydrology, fluvial geomorphology, and stream ecology. We developed downstream hydraulic geometry relationships for bankfull channel width w as a function of drainage area A, w=α Aβ, (DHGwA) for nine aggregate ecoregions comprising the conterminous United States using 1588 sites from the U.S. Environmental Protection Agency's National Wadeable Streams Assessment (WSA), including 1152 sites from a randomized probability survey sample. Sampled stream reaches ranged from 1 to 75 m in bankfull width and 1 to 10,000 km2 in drainage area. The DHGwA exponent β, which expresses the rate at which bankfull stream width scales with drainage area, fell into three distinct clusters ranging from 0.22 to 0.38. Width increases more rapidly with basin area in the humid Eastern Highlands (encompassing the Northern and Southern Appalachians and the Ozark Mountains) and the Upper Midwest (Great Lakes region) than for the West (both mountainous and xeric areas), the southeastern Coastal Plain, and the Northern Plains (the Dakotas and Montana). Stream width increases least rapidly with basin area in the Temperate Plains (cornbelt) and Southern Plains (Great Prairies) in the heartland. The coefficient of determination (r2) was least in the noncoastal plains (0.36–0.41) and greatest in the Appalachians and Upper Midwest (0.68–0.77). DHGwA equations differed between streams with dominantly fine bed material (silt/sand) and those with dominantly coarse bed material (gravel/cobble/boulder) in six of the nine analysis regions. Where DHGwA equations varied by sediment size, fine-bedded streams were consistently narrower than coarse-bedded streams. Within the Western Mountains ecoregion, where there were sufficient sites to develop DHGwA relationships at a finer spatial scale, α and β ranged from 1.23 to 3.79 and 0.23 to 0.40, respectively, with r2>0.50 for 10 of 13 subregions (range: 0.36 to 0.92). Enhanced DHG equations incorporating additional data for three landscape variables that can be derived from GIS—mean annual precipitation, elevation, and mean reach slope—significantly improved equation fit and predictive value in several regions, most notably the Western Mountains and the Temperate Plains. Channel width was also related to human disturbance. We examined the influence of human disturbance on channel width using several indices of local and basinwide disturbance. Contrary to our expectations, the data suggest that the dominant response of channel width to human disturbance in the United States is a reduction in bankfull width in streams with greater disturbance, particularly in the Western Mountains (where population density, road density, agricultural land use, and local riparian disturbance were all negatively related to channel width) and in the Appalachians and New England (where urban and agricultural land cover and riparian disturbance were all negatively associated with channel width).