Machine Learning Improves Debris Flow Warning

Automatic identification of debris flow signals in continuous seismic records remains a challenge. To tackle this problem, we use machine learning, which can be applied to real-time data. We show that learning model based on the random forest algorithm recognizes different stages formation and propagation at Illgraben torrent, Switzerland, with an accuracy exceeding 90 %. In contrast typical detection requiring instrumentation installed our approach provides significant gain warning times tens minutes hours. For data from 2020, detector raises alarms for all 13 independently confirmed events, giving no false alarms. suggest machine-learning is critical step toward next generation debris-flow warning, increases using simpler compared existing operational systems.