Near-surface Fracture Discovery from Landslides by Sonogram Screening

Fracture signals from near-surface processes like creeping landslides and pending rockfalls carry important information on formation stability and promise to enhance early warning approaches significantly. In principal, these fracture signals behave like any local earthquake signal. However, two aspects differ so much that for long time even the existence of these signals was unknown: (I) low energy of local magnitudes Ml = –2 and below, unknown source region and thus sub-optimal network layout with source-receiver distances of several hundred to thousand meter demand signal detection with SNR near one, (II) short signal duration, and a priori unknown signatures result in frequent confusion of fracture signals with noise bursts. In any field installation, noise forensics are crucial to differentiate between signals-of-interest and local noise. Classical detector approaches like STA/LTA are not suited to perform these tasks with sufficient rate of success due to very high false positive rates. Instead, we developed a scheme of visual inspection by seismologists, based on an optimum spectral display of detectable energy spots by combined sonograms of seismic mini-arrays. Applications from landslide monitoring in Austria and France and rockslide precursory detection in Austria demonstrate the appropriateness of our approach.