Online Material: Matlab scripts and example data for adaptive Kalman filter. The 2011 M w 9.0 Tohoku‐Oki earthquake was the most disastrous known event in Japan,triggering a powerful tsunami wave with a maximum amplitude of ∼40 m (Mori et al. , 2012). At least 15,883 people were killed and over 6145 injured, with a total economic loss of more than $235 billion U.S., making it the costliest natural disaster in world history (Emergency Disaster Countermeasures Headquarters, 2011; Hennessy‐Fiske, 2011). The earthquake occurred at 14:46:24 Japan Standard Time (05:46:24 UTC), with epicenter location at 38.297° N, 142.372° E and depth of 30 km (U.S. Geological Survey Earthquake Hazards Program, 2011). Seismic waveforms were recorded by continuous Global Navigation Satellite Systems (GNSS) monitoring network of GNSS Earth Observation Network (GEONET) (Sagiya et al. , 2001) and strong‐motion seismograph networks of the Kyoshin Network (K‐NET) and Kiban Kyoshin Network (KiK‐net) (Aoi et al. , 2004). These have been separately or jointly applied to rapid earthquake magnitude determination (Ohta et al. , 2012; Wright et al. , 2012; Melgar, Crowell, et al. , 2013), tsunami warning (Melgar and Bock, 2013), earthquake rupture process inversion (Ide et al. , 2011; Suzuki et al. , 2011; Yagi and Fukahata, 2011; Yokota et al. , 2011; Yue and Lay, 2011; Frankel, 2013), ionosphere perturbation detection (Heki, 2011; Tsugawa et al. , 2011; Komjathy et al. , 2012), study of Earth free oscillation (Mitsui and Heki, 2012), etc. For those studies, the acquisition of robust seismic waveforms is vital but defective, especially in a real‐time mode. High‐rate Global Positioning System (GPS) is capable of recording seismic waveforms with centimeter‐level accuracy (Genrich and Bock, 2006), regardless of relative kinematic positioning mode (Nikolaidis et al. , 2001; Larson et al. , 2003; Blewitt et al. , 2006, 2009; Bilich et al. , 2008; Crowell …
Read full abstract