Detection of the Bedload Movement with an Acoustic Sensor in the Yangtze River, China

Abstract

The acoustic method, which enables continuous monitoring with great temporal resolution, is an alternative technique for detecting bedload movement. In order to record the sound signals produced by the impacts between gravel particles and detect the bedload motion, in this study, a hydrophone is placed close to the riverbed at the upper Yangtze River. Three categories of raw audio signals—moving gravel particles, ship engines, and flow turbulence—are collected and investigated. Signal preprocessing is performed using spectral subtraction to reduce the noise of the background sound, and the sound signal characteristic parameters are then calculated. In this paper, we propose a novel method for detecting and extracting bedload motion parameters, including peak frequency, pitch frequency, and energy eigenvector. When a segment of a speech signal meets the indicators for all three feature parameters simultaneously, the segment signal is classified as a bedload motion sound signal. Further work will be conducted to investigate bedload transport using the extracted audio signal.