Estimating the Physical State of a Laboratory Slow Slipping Fault from\n Seismic Signals

Over the last two decades, strain and GPS measurements have shown that slow\nslip on earthquake faults is a widespread phenomenon. Slow slip is also\ninferred from correlated small amplitude seismic signals known as nonvolcanic\ntremor and low frequency earthquakes (LFEs). Slow slip has been reproduced in\nlaboratory and simulation studies, however the fundamental physics of these\nphenomena and their relationship to dynamic earthquake rupture remains poorly\nunderstood. Here we show that, in a laboratory setting, continuous seismic\nwaves are imprinted with fundamental signatures of the fault's physical state.\nUsing machine learning on continuous seismic waves, we can infer several bulk\ncharacteristics of the fault (friction, shear displacement, gouge thickness),\nat any time during the slow slip cycle. This analysis also allows us to infer\nmany properties of the future behavior of the fault, including the time\nremaining before the next slow slip event. Our work suggests that by applying\nmachine learning approaches to continuous seismic data, new insight into the\nphysics of slow slip could be obtained in Earth.\n