Joint Seismic and Acoustic Signals in a Boulder-Bed Channel along a Bedrock Canyon

Bedload transport is the dominant fluvial mechanism shaping canyons and constitutes a significant portion of sediments transported from mountains to oceans. Monitoring of bedload flux in bedrock canyons during large magnitude floods remains an outstanding problem in fluvial geomorphology, due to extreme hydraulic conditions and the risk to equipment and human life. Surrogate monitoring methods include the interpretation of seismic and acoustic signals generated by colliding transported grains. Establishing a reliable relationship between seismic and acoustic signals and bedload flux in such floods has hitherto not been attempted. Here we present seismic and acoustic data from two pairs of adjacent channel reaches in the Liwu river, Taiwan, that differ by the concentration of boulders, but otherwise share similar hydraulic conditions and drainage area (~ 60 km²). In each of the paired locations (Shakadang and Baiyang) we have setup a field experiment where seismic sensors and data loggers with a frequency acquisition resolution of 200 Hz, were deployed aside each of the boulder-bed - boulder-free channels. In Baiyang we have also installed a hydrophone, i.e., a microphone submerged in the water column, recording acoustic signals in frequencies up to 16,000 Hz. Our monitoring system recorded a flood during August 2019 where water stage rose from 1 to 3-4 meters within few hours. In the Baiyang stations, bedload transport onset, peak and cessation were resolved through manual listening to audio files recorded by the hydrophone and counting the number of inferred impacts that occurred during one-minute time intervals. The bedload transport event lasted more than 78 hours, peaked in the recession, thus producing a counterclockwise hysteresis between the power of the acoustic signal and the water depth, not observed in the seismic signals. Signals generated by bedload impacts excited the hydrophone at frequencies of 600 to 3000 Hz, and the seismic sensor at the boulder-free stations at 18 to 30 Hz. In contrast, the highest seismic power at the boulder-bed channels peaked at a frequency band of 50 to 80 Hz, which is commonly not associated with bedload nor with water turbulence. The peak in the high frequency bands suggests that boulder-bed channels may differ in how bedload and turbulence are expressed in terms of the seismic content. We hypothesize that the high frequency content may be a combination of (i) small bedload grains colliding onto boulders, and (ii) the coupled effect of enhanced turbulence with the proximity of the seismic station to the channel.