Moment tensors for rapid characterization of megathrust earthquakes: the example of the 2011 M 9 Tohoku-oki, Japan earthquake

The rapid detection and characterization of megathrust earthquakes is a difficult task given their large rupture zone and duration. These events produce very strong ground vibrations in the near field that can cause weak motion instruments to clip, and they are also capable of generating large-scale tsunamis. The 2011 M 9 Tohoku-oki earthquake that occurred offshore Japan is one member of a series of great earthquakes for which extended geophysical observations are available. Here, we test an automated scanning algorithm for great earthquakes using continuous very long-period (100-200 s) seismic records from K-NET strong-motion seismograms of the earthquake. By continuously performing the cross-correlation of data and Green's functions (GFs) in a moment tensor analysis, we show that the algorithm automatically detects, locates and determines source parameters including the moment magnitude and mechanism of the great Tohoku-oki earthquake within 8 min of its origin time. The method does not saturate. We also show that quasi-finite-source GFs, which take into account the effects of a finite-source, in a single-point source moment tensor algorithm better fit the data, especially in the near-field. We show that this technique allows the correct characterization of the earthquake using a limited number of stations. This can yield information usable for tsunami early warning.