Seismic instruments placed outside of spatially extensive hazard zones can be used to rapidly sense a range of mass movements. However, it remains challenging to automatically detect specific events of interest. Benford's law, which states that first non-zero-digit of given datasets follow a specific probability distribution, can provide a computationally cheap approach to identifying anomalies in large datasets and potentially be used for event detection. Here, we select raw seismic signals to derive the first-digit distribution. The seismic signals generated by debris flows, landslides, lahars, and glacier-lake-outburst floods follow Benford's law, while those generated by ambient noise, rockfalls, and bedload transports do not. Focusing on debris flows, our Benford's-law-based detector is comparable to an existing random forest method for the Illgraben, Switzerland, but requires only single station data and three non-dimensional parameters. We suggest this computationally cheap, novel technique offers an alternative for event recognition and potentially for real-time warnings.

An important component of quantifying bedload transport flux is the identification of the onset of bedload motion. Bedload transport can be monitored with high temporal resolution using passive acoustic methods, e.g., hydrophones. Yet, an efficient method for identifying the onset of bedload transport from long-term continuous acoustic data is still lacking. Benford’s Law defines a probability distribution of the first-digit of datasets and has been used to identify anomalies. We apply Benford’s Law to the three years of acoustic recordings from a stationary hydrophone in the Taroko National Park, Taiwan. Our workflow allows for monitoring bedload motion in near-real-time, and it is convenient for others to reference. Two bedload transport events were identified during the examined period, lasting 17 and 45 hours, accounting for approximately 0.35% of the time per year.

Machine learning can improve the accuracy of identifying mass movements in seismic signals and extend early warning times. However, we lack a profound understanding of the effective seismic features and the limitations of different machine learning models, especially for debris-flow warning. Here, we investigate the importance of seismic features for the binary debris flow classification tasks using two ensemble models: Random Forest (RF) and eXtreme Gradient Boosting (XGB) models. We find that an established approach to training machine learning models for debris flow classification task based on more than seventy seismic signal features may be affected by redundant input information. These seismic features are derived from physical and statistical knowledge of impact sources and are grouped into waveform, spectrum, spectrogram, and network sets. Our results show that only six selected seismic features can perform similarly for the binary debris flow classification task compared to published benchmark results trained with seventy features. Considering models that aim to capture patterns in sequential data rather than focusing on information only in one given window as ensemble models, using the Long Short-Term Memory (LSTM) algorithm does not improve the performance of binary debris flow classification tasks over RF and XGB. As a debris flow alarm task, the LSTM model predicts debris flow initiation more consistently and generates fewer false warnings. Our proposed framework simplifies seismic signal-driven early warning for debris flows and provides an appropriate workflow for identifying other mass movements.