TY - JOUR
T1 - Observation of core phase ScS from the Mw 9.0 Tohoku-Oki earthquake with high-rate GPS
AU - Guo, Aizhi
AU - Wang, Yong
AU - Li, Zhiwei
AU - Ni, Sidao
AU - Wu, Wenbo
AU - Liu, Genyou
AU - Zheng, Yong
AU - Simons, Mark
PY - 2013/7
Y1 - 2013/7
N2 - We collected high-rate GPS data (1-Hz) from 192 stations in China recording the Tohoku-Oki earthquake. Not only do we retrieve P, S, and surface waves, but we also retrieve the outercore-reflected ScS wave. This inference is supported by comparison with both seismographic records and the synthetic waveforms. However, it is still challenging to include ScS waveforms for earthquake source studies, because the complexity of lowermost mantle structure (such as the sharp boundaries and ultra-low velocity zones [ULVZs]) may substantially change amplitude and even waveforms of ScS (Helmberger et al., 2000; Ni and Helmberger, 2001; Ni et al., 2002). For observation of seismic phases with large apparent velocity such as ScS, the PPP processing approach is preferred over the double-difference approach. For example, the traveltime difference of ScS at distances of 10° and 20° are only about 25 s, the minor difference in travel time makes ScS (duration about 100 s) from the Tohoku-Oki earthquake difficult to observe with the double-difference approach even with a reference station as far as 1000 km away. To demonstrate the distortion of ScS due to the double-difference approach, we show ScS signals on the east-west component at stations HIA and MDJ and the difference of ScS between HIA and MDJ (see {circled E} supplementary Figs. S1 and S2, available in the electronic supplement that accompanies this paper). The ScS signals on HIA and MDJ are similar, but the differenced ScS signals are substantially different from ScS signals on individual stations (see {circled E} supplementary Fig. S2 in supplement). However, signals retrieved with the PPP approach could be contaminated by common mode error (CME), which may be present for multiple reasons (Dong et al., 2006). CME for time scale of days has been explored, but CME from high-rate GPS for a time scale of one hundred to one thousand seconds are little studied. To assess effects of CME on our study, we perform principal component analysis on the high-rate GPS time series from time of 2000-4000 s before the Tohoku-Oki earthquake. On the north-south and east-west components, CME from the first principal component for the time scale of a few thousand seconds has amplitude of less than 3 mm (see {circled E} Supplemental Fig. S3 in the supplement). But for the up-down direction, CME shows long-period drift and reaches almost 1 cm. For the time scale of 100 s, CME has amplitude of less than 2-3 mm, substantially smaller than the 7 mm displacement observed for ScS. Of course, ScS signal could be enhanced if the sources of CME are better understood and more advanced processing techniques are available for suppressing CME. By applying the methods usually used in seismology to high-rate GPS such as stacking and other array analysis techniques, more seismic phases could be identified and retrieved for strong earthquakes. Combined with conventional seismic instrumentation, we should routinely use these plentiful data sources for seismological studies (Bock et al., 2011; Crowell et al., 2012).
AB - We collected high-rate GPS data (1-Hz) from 192 stations in China recording the Tohoku-Oki earthquake. Not only do we retrieve P, S, and surface waves, but we also retrieve the outercore-reflected ScS wave. This inference is supported by comparison with both seismographic records and the synthetic waveforms. However, it is still challenging to include ScS waveforms for earthquake source studies, because the complexity of lowermost mantle structure (such as the sharp boundaries and ultra-low velocity zones [ULVZs]) may substantially change amplitude and even waveforms of ScS (Helmberger et al., 2000; Ni and Helmberger, 2001; Ni et al., 2002). For observation of seismic phases with large apparent velocity such as ScS, the PPP processing approach is preferred over the double-difference approach. For example, the traveltime difference of ScS at distances of 10° and 20° are only about 25 s, the minor difference in travel time makes ScS (duration about 100 s) from the Tohoku-Oki earthquake difficult to observe with the double-difference approach even with a reference station as far as 1000 km away. To demonstrate the distortion of ScS due to the double-difference approach, we show ScS signals on the east-west component at stations HIA and MDJ and the difference of ScS between HIA and MDJ (see {circled E} supplementary Figs. S1 and S2, available in the electronic supplement that accompanies this paper). The ScS signals on HIA and MDJ are similar, but the differenced ScS signals are substantially different from ScS signals on individual stations (see {circled E} supplementary Fig. S2 in supplement). However, signals retrieved with the PPP approach could be contaminated by common mode error (CME), which may be present for multiple reasons (Dong et al., 2006). CME for time scale of days has been explored, but CME from high-rate GPS for a time scale of one hundred to one thousand seconds are little studied. To assess effects of CME on our study, we perform principal component analysis on the high-rate GPS time series from time of 2000-4000 s before the Tohoku-Oki earthquake. On the north-south and east-west components, CME from the first principal component for the time scale of a few thousand seconds has amplitude of less than 3 mm (see {circled E} Supplemental Fig. S3 in the supplement). But for the up-down direction, CME shows long-period drift and reaches almost 1 cm. For the time scale of 100 s, CME has amplitude of less than 2-3 mm, substantially smaller than the 7 mm displacement observed for ScS. Of course, ScS signal could be enhanced if the sources of CME are better understood and more advanced processing techniques are available for suppressing CME. By applying the methods usually used in seismology to high-rate GPS such as stacking and other array analysis techniques, more seismic phases could be identified and retrieved for strong earthquakes. Combined with conventional seismic instrumentation, we should routinely use these plentiful data sources for seismological studies (Bock et al., 2011; Crowell et al., 2012).
UR - http://www.scopus.com/inward/record.url?scp=84880246616&partnerID=8YFLogxK
U2 - 10.1785/0220120143
DO - 10.1785/0220120143
M3 - Article
AN - SCOPUS:84880246616
SN - 0895-0695
VL - 84
SP - 594
EP - 599
JO - Seismological Research Letters
JF - Seismological Research Letters
IS - 4
ER -