How fractal dimension changes during mass movement events in seismic signals?

Enhancing real-time detection of mass movement events is critical for improving early warning systems and reducing risks to individuals and communities. Seismic monitoring offers an effective tool for hazard detection and timely alerts. However, a significant challenge remains in successfully isolating seismic signals associated with mass movements from continuous recordings, often obscured by persistent background noise. Therefore, it is essential to develop robust and reliable algorithms for automatic detection. This study proposes utilizing fractal geometry to quantify signal patterns across various scales, distinguishing seismic signals from background noise based on fractal dimension (FD). The study analyzed seismic data from various mass movement events, including debris flows and rockfalls in the Illgraben catchment of Switzerland and a landslide event from the Askja caldera in Iceland. Two methods were employed to estimate the FD: (i) the variogram estimator and (ii) detrended fluctuation analysis. The results show that noise typically exhibits a higher FD than the seismic signals produced by mass movements. Additionally, this study established distinct FD ranges for each type of mass movement, facilitating their classification. The outcomes also show that landslide seismic landslide signals exhibit high variability, particularly with low (signal-to-noise ratio) SNR and increased distance from the source. The findings highlight the potential for this method to improve seismic event detection in real-time monitoring systems.