Resumen
Bed-load discharge of a river can be monitored by indirectly measuring the acoustic pulses generated when the bed load collides with a steel pipe installed on the riverbed (i.e., pipe hydrophone measurement). However, existing methods used for filtering pulses from acoustic signals reflect a combination of bed-load collision frequency bands, thereby limiting characterization capabilities. This study proposes an improved filtering method that separates and efficiently examines frequency bands that are highly correlated with bed-load collision characteristics. Herein, an experimental hydraulic model and bed-load collision sound-measurement system were constructed, and bed-load collision experiments were repeatedly performed for collecting acoustic data using a pipe hydrophone. Fast Fourier Transform analysis was performed on data to select the specific frequency bands and pressures reflecting the bed-load particle size. Furthermore, a bandpass method to examine bed-load collision sounds is also presented herein. Results indicate that in comparison with existing filtering methods, the proposed bandpass method yields higher detection rates under bed-load conditions of low flow rate and small particle size, thereby demonstrating its enhanced effectiveness.