M9.0 Sumatra-Andaman Island Earthquake on 26 December 2004. (A) Vertical component seismogram of velocity (m/sec.). (B) High frequency (1 Hz high-pass fi ltered) seismogram of velocity (m/sec.). Shorter wave trains are shown about 450 seconds which estimated as source duration. (C) Seismogram of the integrated displacement (m · sec.) in absolute values. Theoretical arrival time of P and later phases (pP, sP, PcP and PP) are shown. (D) Calculated Mwp from the integrated displacement seismogram. 

Context in source publication

Context 1

... 1977), for each vertical broadband seismic channel, without a correction for the radiation pattern. We add 0.2 to Mw obtained using the above procedure to compensate for to get Mwp. The results show that Mwp correlates well with Harvard’s Mw’s obtained from their Centroid Moment Tensor (CMT) solutions (Dziewonski et al. , 1981). Both U.S. Tsunami warning centers, at the WC/ATWC and at the PTWC, use Mwp for their first estimate of Mw. It provides the fastest possible estimate of a large, potentially tsunamigenic earthquake’s moment magnitude, enabling a fast Tsunami Warning. We calculated Mwp for both of these two great Sumatra earthquakes. In both cases, we analyzed 130 seconds of signal from the vertical component at MAJO beginning 10 seconds before the initial P-wave onset time. The velocity waveform (A), the 1 Hz high pass filtered velocity waveform (B), the absolute value of the integrated displacement waveform (C), and the Mwp value (D), are shown separately for the December 26, 2004 event in Fig. 3, and for the March 28, 2005 event in Fig. 4. From Eqs. (1) and (2) we calculated Mwp 8.0 for the December 26, 2004 event from the integrated displacement seismogram from MAJO. Using a magnitude correction of (Mwp _ initial-1.03)/0.83 (Whitmore et al. , 2002), PTWC determined an Mwp from MAJO of 8.2. This value is still extremely small compared to Harvard’s value for Mw of 9.0, and Stein and Okal’s value of Mw 9.3. A scatterplot of 3748 individually determined Mwp deficits versus epicentral distance shows no obvious trend (Whitmore et al. , 2002). Mwp deficit trend to epicenter distances is shown at MAJO for 18 events which determined the Mw greater than 7.0 by Harvard in the 2 years from 2003 to 2004 (Fig. 5). We then recalculated Mwp using α = 0 . 16 + 7 . 9 km./sec. derived from a fit to the P wave apparent velocity from the IASP91 earth model instead of α = 7 . 9 km/sec (P wave velocity) in Eq. (1); the original equation for Mwp. We determined Mwp = 8 . 5 from MAJO (Fig. 3D), and Mwp 8.6 from the average of 15 stations for the December 26, 2004 event, and Mwp 8.7 from MAJO (Fig. 4D), and Mwp 8.7 from the average of 15 stations for the March 28, 2005 event using the distance-dependent formulation for α . Our new value of 8.5 for Mwp is still much smaller than Harvard’s Mw of 9.0, and Stein and Okal’s Mw of 9.3. The December 26, 2004 event was one of largest ever recorded, the rupture expanded at a speed of about 2.5 kilometers per second toward the north northwest, extending about 1,300 kilometers along the Andaman trough, composed of multiple fractures (Ammon et al. , 2005). The value of Mwp 8.5 nearly equal to Mw 8.7 ( Mo = 0.16E + 23N · m) estimated from the peak moment ratios (4E + 20N · m/s) in a range of 120 seconds from P-wave arrival (Ammon et al. , 2005). The value is large enough to be useful in evaluat- ing the probability of tsunami generation. The duration of high frequency energy radiation more than 600 seconds (Ni et al. , 2005, Ishii et al. , 2005). The high frequency durations are about 500, and 100 seconds respectively for the December 2004 and March 2005 events on 1 second high- pass filtered broadband vertical velocity records from the IRIS MAJO station. These duration times show the differences between fracture lengths, and also provide an imme- diate estimate of the moment released from these events. The high frequency signal duration for the December 26, 2004 (Fig. 3B) event is about 500 seconds, as compared to about 100 seconds for the March 28, 2005 (Fig. 4B) event. This difference re fl ects the propagation of the rupture front along different fault lengths (Ni et al. , 2005). The Matsushiro Seismological Observatory (MSO) of JMA uses the data from USGS LISS and IRIS BUD server to determine hypocenter and moment magnitude. The hypocenter is calculated by HYPOSAT/NORSAR and a Grid-search method developed in house at the MSO. We estimate moment magnitude from the mean of 4 minutes of vertical component broadband squared velocity amplitudes beginning at the P-wave arrival time. Using this method on data from the IRIS GSN station at MAJO, we estimate Mw values of 8.8 and 8.7 for the two events off of the North West Coast of Sumatra, Indonesia on December 26, 2004, and March 28, 2005 respectively. Both U.S. Tsunami warning centers (the WC/ATWC, and the PTWC) use Mwp to estimate moment magnitude, based in the integrated displacement amplitude of the vertical broadband P-wave portion of the seismogram. Mwp provides the fastest estimate of an of a large earthquake ’ s moment magnitude enabling the Warning centers to issue a tsunami warning. The constant value for α , the mean P-wave velocity along the propagation path was used in the original equation because Mwp was initially derived as a method for local earthquakes, using data from seismic stations within a few hundred kilometers of the epicenter. In this paper we use α = 0 . 16 + 7 . 9 (the apparent P-wave velocity from a fi t to the IASP91 earth model) instead of the constant value of 7.9 for α in the original equation for Mwp. This change gives a value for Mwp from the IRIS MAJO station of 8.5 (the PTWC ’ s initial value for MAJO was 8.2), for the December 26, 2004 event, and a value of 8.7 (the PTWC ’ s original value for MAJO was 8.4) for the March 28,2005 event. Mwp is not suit- able for a precise estimate of the total moment magnitude of complex earthquakes, involving multiple fractures, such as the great December 26, 2004 Sumatra earthquake. Mwp does, however, provide the fastest possible moment magnitude estimate, and, using a distance-dependant P-wave velocity, is accurate enough to justify an initial tsunami warning. Indeed, had a Sumatra type event occurred in the Paci fi c, the corrected Mwp discussed here would have well ex- ceeded the tsunami regional watch/warning threshold used by PTWC and the WC/ATWC for the Paci fi c ...