Rapid Source Characterization of the 2011 Tohoku-oki Earthquake with Real-Time GPS and Strong Motion Data: Implications for Tsunami Warning

Saturation of seismometers in the near field and problems with the double integration of accelerometer data into displacements degrades rapid magnitude estimation and source characterization for large earthquakes. Although Japan possesses the most advanced earthquake and tsunami early warning system in the world, it took about 20 minutes using teleseismic waves from distant seismic networks to determine that the 11 March 2011 Tohoku-oki earthquake was an Mw 9.0 event - earlier estimates based on local seismic networks ranged from M7.2 to M8.0. We present a seismogeodetic network model that can estimate on-the-fly the full dynamic range and spectrum of seismic motions for large earthquakes using observations starting in the near field. Replaying local Japanese data in a simulated real-time mode from 785 GPS stations and 190 accelerometers during the Tohoku-oki earthquake, we demonstrate that an accurate centroid moment tensor solution to ascertain the type of earthquake, immediately followed by a finite fault slip model could have been obtained in about 3 minutes, providing more accurate and timely warnings of the severity of the impending tsunami and assisting first responders with evacuation and recovery efforts. The seismogeodetic network model is sufficiently accurate to detect P wave arrivals in the near field for large events, leading to improved earthquake early warning (EEW) methodologies that predict the arrival and intensity of S waves. Using a maximum P wave amplitude scaling relationship from displacements estimated for 175 stations during three historical large earthquakes including Tohoku-oki, we demonstrate that an accurate first estimate of magnitude may be able to be obtained within a few seconds of P wave arrival at a handful of stations closest to the epicentre. Furthermore we use the rapid slip inversion to model the ensuing tsunami, we will show comparisons between the predicted waveforms and those measured by DART buoys, near shore tide gauges and other geophysical instruments to demonstrate that the model is robust enough to produce reasonable predictions of tsunami intensity,