CONSTRUCTING A 260-KM LONG, 25-CM RESOLUTION MAP OF COLORADO RIVER BED SEDIMENT IN GLEN, MARBLE, AND GRAND CANYONS

High-resolution (25-cm grid) maps of surface riverbed sediment have been constructed for an almost continuous 260-km stretch of the Colorado River in Glen, Marble and Grand Canyons. The maps have been compiled using tens of billions of soundings from a Reson 7125 multibeam echosounder, utilizing the acoustic backscatter and bathymetric data in conjunction with thousands of photographic observations of the bed collected with camera systems developed in-house by the USGS Grand Canyon and Monitoring Research Center. The resulting map is the most spatially extensive and highest resolution substrate map yet compiled for any river in the world.The data were collected over a six-year period (2012-2017, inclusive) in modal water depths of between 2 and 15 m (maximum depth around 30 m). The bed of this large canyon river is various among sand, gravel, cobbles, boulders and submerged vegetation. The study area is divided into two segments based on geomorphic characteristics and the location of major tributaries: 1) Glen Canyon, upstream from the first major tributary below Glen Canyon Dam, where the bed is dominated by gravel- and cobble-supported submerged aquatic vegetation, and 2) Marble and Grand Canyons where frequent tributaries that contribute fine- and coarse-sediment result in a mixed sand-gravel-cobble-bedrock riverbed with little to no vegetation. The sediment classifications are developed using novel methods harnessing the grain-size information in high-frequency acoustic backscatter, calibrated to an extensive underwater video dataset. We filter out the topographic contributions to acoustic backscatter using Fourier series that decomposes backscatter into the (unwanted) high-pass component associated with bedform topography (ripples, dunes, and sand waves) and vegetation, and the (desired) low-frequency component associated with the composition of sediment patches superimposed on the topography. This process strengthens relationships between backscatter and sediment composition. We then classify each filtered backscatter measurement into substrate type using a Gaussian mixture model. A 4-part substrate classification has been developed for the partially vegetated Glen Canyon reach, and a 5-part classification for unvegetated riverbeds in Marble and Grand Canyons.