Long Short-Term Memory Networks for Earthquake Detection in Venezuelan Regions

Reliable earthquake detection and location algorithms are necessary to properly catalog and analyze the continuously growing seismic records. This paper reports the results of applying Long Short-Term Memory (LSTM) networks to single-station three-channel waveforms for P-wave earthquake detection in western and north central regions of Venezuela. Precisely, we apply our technique to study the seismicity along the dextral strike-slip Bocono and La Victoria - San Sebastian faults, with complex tectonics driven by the interactions between the South American and Caribbean plates.