Seismic Signal Analysis for Landslide: Detection and Classification

This research presents a simple technique to detect and classify landslide seismic signals from other seismic noise signals. Research aims to develop non-structural mitigation measures for landslide hazards, including monitoring landside-induced seismicity. The data was collected in real time from an 18-seismometer array situated in North Western Himalayas. The methodology to detect events in the time history record is based on STA/LTA, where managed event was analyzed through Peterson noise model, and finally, an event was considered a potential landslide event if it crossed the maximum noise boundary. The Time–Frequency Analysis (TFA) technique was used to classify the detected events. Based on literature review, Short-time Fourier Transform (STFT), Fourier Synchrosqueezed Transform (FSST), Wavelet Synchrosqueezed Transform (WSST), and Smoothed Pseudo-Wigner-Ville Distribution (WVD) are the best suitable for seismic signal analysis. As a result, in our study region, seismic events can be classified with high-frequency resolution and time localization with WSST; also, we found some frequency bands for dominating frequency with the event source. The results have been verified with actual events records, and all these events happened in recent years.