Breaching parameter assessment of landslide dams using vibration signals: insights from flume experiments

Technical Summary: Breaching Parameter Assessment of Landslide Dams Using Vibration Signals

Problem Statement
Landslide dams, formed by geological masses blocking river channels, pose significant threats due to their high failure rates (approximately 30% within one week of formation) and the massive flood volumes generated upon breaching. While revealing the breaching mechanism and monitoring parameters are critical for disaster prevention, the process is inherently complex. Traditional monitoring methods struggle to accurately capture breaching parameters directly during the event. Although environmental seismology has successfully utilized seismic signals to study landslides, rockfalls, and debris flows, previous research on landslide dam breaching has largely focused on qualitative signal identification or limited correlations with specific parameters. There is a lack of systematic experimental data linking dam material gradation, inflow discharge, and dam geometry to both breaching dynamics and the resulting local vibration signals.

Methodology
To address these gaps, the authors conducted sixteen controlled flume experiments using a custom-built rectangular flume (970 cm × 50 cm × 50 cm) with a 1° inclination. The study employed a "local monitoring" configuration, placing three vertical accelerometers directly beneath the dam and at the toe/downstream areas to record vibration signals, thereby minimizing propagation path effects and external noise.

The experimental design varied three key parameters across three groups:

1. Material Gradation (Tests T1–T7): Median particle size ($D_{50}$) ranged from fine sand (0.075 mm) to coarse gravel (40 mm).
2. Inflow Discharge (Tests T8–T12): Inflow rates varied from 0.5 to 2.5 L/s.
3. Dam Geometry (Tests T13–T16): Dam height ($H$) and downstream slope angles were varied to alter the height-to-width ratio ($H/B$).

The breaching process was initiated via a pre-excavated discharge chute to ensure controlled overtopping. Data acquisition included high-speed cameras for visual monitoring, water level sensors for flow calculation, and accelerometers sampling at 1 kHz. The vibration signals were processed using Fast Fourier Transform (FFT), Short-Time Fourier Transform (STFT), Power Spectral Density (PSD), and Arias Intensity ($I_A$) analysis.

Key Results
The study identified a three-stage breaching evolution: formation, expansion, and attenuation/stabilization.

• Breaching Dynamics: Finer dam materials, higher inflow discharges, and larger $H/B$ ratios resulted in higher peak outflow discharges ($Q_{max}$) and shorter total breaching durations. Conversely, coarser materials led to higher residual dam heights due to selective erosion where fine particles are transported while coarse gravel remains.
• Vibration Signal Characteristics:
  • Signals were primarily triggered during the expansion stage (Stage 2).
  • Coarser materials, higher inflows, and larger dam heights produced vibration signals with richer frequency components and higher energy.
  • The dominant frequency range was 50–450 Hz. As particle size increased, the frequency centroid shifted toward higher frequencies, attributed to intense coarse-particle collisions.
• Quantitative Relationships:
  • PSD and Discharge: A strong linear relationship was observed between the breaching outflow discharge ($Q$) and the logarithm of the PSD envelope ($\lg(PSD)$), with a correlation coefficient of $R^2 = 0.79$. This suggests the PSD envelope can effectively track temporal variations in outflow.
  • Arias Intensity and Parameters: The maximum Arias intensity ($I_{A,max}$) showed strong linear correlations with the logarithm of the median particle size ($R^2 = 0.88$), inflow discharge ($R^2 = 0.93$), and the $H/B$ ratio ($R^2 = 0.99$).

Significance and Claims
The paper claims to provide laboratory-scale empirical evidence linking breaching dynamics to recorded local vibration responses. The primary contributions are:

1. Empirical Basis for Monitoring: The study establishes that vibration signals contain quantifiable information regarding breaching parameters, offering a potential basis for signal-based monitoring of landslide dam breaching.
2. Quantitative Models: The derived linear relationships between vibration energy metrics (PSD envelope, Arias intensity) and physical parameters ($Q$, $D_{50}$, $H/B$) allow for the estimation of breaching discharge and energy release from vibration data.
3. Physical Interpretation: The results support the interpretation that vibration energy is driven by a combination of flow-bed interaction, bedload transport, and slope collapse, with specific signal features (e.g., high-frequency peaks) corresponding to coarse particle impacts.

Limitations and Scope
The authors explicitly state that these findings are limited to laboratory-scale, local monitoring conditions. The study acknowledges that the absolute amplitudes and fitted relationships cannot be directly extrapolated to field-scale landslide dams due to the absence of propagation path effects (attenuation, scattering, geometric spreading) present in natural environments. Consequently, the paper positions these results as a foundational step for future research, emphasizing that field-scale applicability requires further evaluation and that the current relationships serve as controlled-condition correlations rather than validated field inversion methods.