River sediment observatory in the mountainous catchment: Long-term and high spatiotemporal monitoring with distributed acoustic sensing (DAS), large-N geophone array, hydrophone, smart rock, AI-based grain-size scanner, and photogrammetric survey

The big challenge addressed in this study involves understanding Earth surface processes, such as landslides, sediment transport, and bedrock incision, which shape landscapes and link climate, tectonics, and erosion. These processes require long-term monitoring of experimental catchments to capture the full range of timescales involved in their evolution. Hydrophones deployed in stream can record fluvial soundscapes over frequencies from Hz to tens of kHz, which are possibly corresponding to sediment transport. Large-N array geophone deployed along riverbank can also provide additional constraints on bedload flux and turbulent flow using seismic physical models. However, high-frequency of fluvial-related seismic signals are rapidly attenuated before arrival riverine geophones. Discrepancies in frequency contents and sensing space resulted that is difficult to fully connect passive seismo-acoustic signals to fluvial processes. In this study, distributed Acoustic Sensing (DAS) is proposed as possible solution to advance in understanding of sediment transport. DAS not only records strain-rate at meter scales, similar to large-N geophone array, but also monitors frequencies from mHz to kHz, similar to hydrophone. This study aims to establish a river sediment observatory in a mountainous catchment in Taiwan. This observatory will serve as a research and educational hub for long-term monitoring, providing valuable data for sediment transport studies and environmental conservation. We have conducted a preliminary experiment in 2024. The artificial channel segment reach approximately 30 m long, 1 m depth, and up to 3.8 m width, with an average slope of 4°. Based on a series analysis of photogrammetric survey, we measure the flow configurations during experiment with flow discharge ranging from 1.1 m³/s to 1.5 m³/s and flow surface velocity of 1.9 – 3.0 m/s. The riverbed is covered with sediment particles that have D50 before the experiment ranged between 10 mm and 12 mm, all of which were smaller than the D50 values measured after the experiment (15 mm to 22 mm) as derived from pebble scanner using a deep learning model. Regarding the movement of sediment particles, data from smart rocks showed that 8 impacts occurred along the channel; with an average saltation velocity of ~1.5 m/s. The time-series monitoring data from the field channel experiment showed notable time-frequency differences in the microseismic signals at stations located in the concave bank scour and sediment deposition areas. The PSD at the concave bank scour station exhibited stronger PSD energy and a broader frequency range, with a dominant frequency range of approximately 20 – 80 Hz. This was consistent with the time-frequency results from the impact tests. The study hypothesizes that this dominant frequency characteristic is caused by the sediment material saltation effects. Fiber optic records showed that when the upstream flow reached the test area, it caused a strain rate of approximately 10^-5/s. We further conduct PSD estimation of in-stream and along-riverbank DAS data to explore spectral characteristics corresponding to riverbed scour, lateral erosion, sediment transport and flow dynamics. The goals in this study include developing new monitoring instruments, validating seismic models, and exploring the impact of extreme events on sediment dynamics.