The Finite-Fault Rupture Detector (FinDer) algorithm provides real-time finite-source models for on-going fault ruptures. The FinDer algorithm determines a line-source model that is best suited to explain the current spatial ground-motion pattern observed at any given time in a dense seismic network. This is done via the cross-correlation of observed ground-motion amplitudes with theoretical templates modeled from empirical ground-motion prediction equations and a range of magnitude-dependent rupture lengths. The estimated line-source parameters (centroid, length, and strike) and uncertainties are updated every second to accommodate the temporal changes in the observed amplitudes caused by the evolving fault rupture. The FinDer v.2 algorithm has been implemented in C++ and integrated into the ShakeAlert and SeisComP3 seismic processing systems. We report the real-time and offline performances of FinDer in various regions including the Pacific Northwest, Switzerland, Chile, Central America, Central Italy, and Japan. We compare FinDer performance for different ground-motion metrics used for template generation, including peak-ground acceleration and velocity amplitudes, with and without site corrections. We typically obtain stable line-source solutions within a few seconds from event origin, provided the seismic network is sufficiently dense. Even when ground motions are available only at the nearest stations, FinDer can generate line-source estimates. From these estimates, seismic ground-motions can be predicted for surrounding areas often before the onset of strong shaking. These results demonstrate the potential of FinDer for earthquake early warning. For small earthquakes, generally 2.5 < M < 5.0, the FinDer-estimated rupture strike tends to coincide within 25o or less with one of the nodal planes of the fault plane solutions, suggesting an important role of rupture orientation in small earthquakes, which creates asymmetric ground-motion patterns.
Media has not been submitted for this Presentation