Joint Federal Interagency Sedimentation Conferenet and Federal Interagency Hydrologic Modeling Conference 2015

Full Program »

Hydroacoustic Signatures of Colorado Riverbed Sediments in Marble and Grand Canyons Using Multibeam Sonar.

Technical Paper
View File

Conventional sampling of submerged riverbed sediments (e.g. grabs, cores, and dredges) is costly, labor-intensive, and limited in spatial coverage. Video and photographic sampling is more cost-effective and does not require time-consuming laboratory analyses, which allows sampling at greater frequency and coverage. However, the use of high-frequency acoustic backscatter from swath-mapping sonar systems to classify sediment by grain size has perhaps the potential to provide complete coverage of large areas of the bed (km2), rapidly (minutes to hours) and at high resolution (cm2 to m2), with high positional accuracy (cm).

In the absence of theory concerning heterogeneous riverbed surfaces scatter high-frequency sound, we have adopted a data-driven statistical approach to classify bed sediments, by comparing the spectral properties of high-frequency acoustic echoes received by multibeam sonar from the surface of a heterogeneous riverbed, with bed-sediment observations using geo-referenced underwater video.

Statistics of backscatter magnitudes alone are found to be poor discriminators between sediment types. However, the found that use of combinations of spectral parameters calculated from geo-referenced point clouds of acoustic echoes (namely, the spectral variance, and the intercept and slope from a power-law spectral form) has the potential to delineate patches of similar riverbed sediment types (sand, gravel, cobbles and boulders). An advantage of a spectral approach is that it make no assumptions about the data, such as the distributional form of the fluctuating component of backscatter over small spatial scales.

A decision-tree approach to sediment classification was able to classify spatially heterogeneous patches of homogeneous sands, gravels (and sand-gravel mixtures), and cobbles/boulders with 95, 88, and 91% accuracy, respectively. Application to sites outside the calibration, and surveys made at calibration sites at different times, were found to be physically plausible based on observations from underwater video transects.

Based on analysis of various decision tress built with different training data sets, we tentatively suggest that spectral width, or the number of component frequencies required to describe the spatial variation in backscatter, is a sensitive indicator of different sediment types. However, we argue that no one spectral parameter can definitively separate the relative contributions of roughness and acoustic impedance (hardness) in the acoustic signal from a particular substrate type.


Daniel Buscombe    
Flagstaff, AZ
U.S. Geological Survey

Paul Grams    
Flagstaff, AZ
U.S. Geological Survey

Matthew Kaplinski    
Flagstaff, AZ
Northern Arizona University

Robert Tusso    
Flagstaff, AZ
U.S. Geological Survey

David Rubin    
Santa Cruz, CA
University of California Santa Cruz


Powered by OpenConf®
Copyright ©2002-2013 Zakon Group LLC