نوع مقاله : مقاله تحقیقی (پژوهشی)
مؤسسة ژئوفیزیک دانشگاه تهران، ایران
عنوان مقاله [English]
It is widely believed that the moment tensor (MT) solution of an earthquake with a non- double couple moment tensor gives significant information about the source mechanism. An interesting topic in non-DC source studies is the relationship among non-DC events, multiple DC events, and complexity in fault segmentations.
This study presents a non-DC tectonic earthquake (forward modeling). This means that in the modeled moment tensor, there are no volumetric components. This model was made by adding the seismograms of the two 100% double couple event close lying, thus their seismograms overlap with each other. Then, an investigation is made of the seismic moment of the second subevent, frequency band used in the inversion and also the temporal and spacial separation between two subevents on the result of the inversion. This study focuses on the spacial complex source with Mw ~ 5. Based on empirical relations between moment magnitude and ruptured area (Somerville et al, 1999), the fault plane of an earthquake could not be larger than 11 km2, which could not be “seen” in our MT analysis (our MT inversion was run at frequency bands between 0.02-0.1 Hz). Thus, this inversion was not a pure DC MT inversion for two subevents at the same point but allowed their time separation to vary. The results show that the mechanism of the second subevent is dependent on the frequency band use and its accuracy is related to the higher frequencies used in the MT inversion (> 0.09). It must be noted that the DC percentage of the moment tensor is dependent on the separation between subevents. The investigation shows that, by increasing the separation between two subevents, the DC percentage reaches higher values, perhaps because of fewer overlaps of the seismograms.
We used the ISOLA package developed by Zahradnik et al, (2005) for the MT inversion, centroid location and time. This package is based on the Kikuchi and Kanamori, (1991) method for the teleseismic study of large earthquakes. ISOLA was developed for local and regional distances. For forward modeling of the complex source, we used the CPS package developed by R.B Herrmann in Saint Louis University.
Finally, we attempt to model the real data. The 2005/05/01 Mw 5 Central Iran earthquake is investigated based on the results of the synthetic test conducted in the first part. A waveform inversion is performed at near-regional stations at frequencies 0.02–0.07 Hz to search for the optimum 3D location and time of the centroid, using broadband station waveforms from the IIEES network. For calculating the Green function, two crustal velocity models were used, namely Ghods et al, (2010) and Walter, (2000). The Ghods et al, (2010) provides a higher correlation between observed and synthetic data. The optimum depth for centroid is 7 km. The stability of the MT inversion was investigated by the jackknifing method, and, finally, those stations which had low correlation between the observed and synthetic data were eliminated. The DC percentage of the moment tensor during the inversion was changed from 7 to 50%, which made this event a candidate for source complexity in order to study whether this earthquake could be interpreted as having two point sources lying close to each other. MT inversions were executed in frequency ranges from 0.02-0.07 to 0.02-0.13 Hz by adding 0.01 in each run. The second subevent had no significant impact on the waveform matches (about 2%). The mechanism of the first subevent was stable during the inversion and made the same correlation as the deviatoric MT inversion, but the mechanism of the second subevent changed considerably as a result of the changes in the frequency band of the inversion. The time separation between two subevents was about 6 to 8 seconds in different runs, which is unacceptable for earthquakes with magnitudes of approximately ~ 5, for which the fault area is approximately 11 km2. Thus, the two point sources model for this earthquake seems not to be a better model than the one point source. We also endeavored to model the Mw 5.7, 2006/03/25 Fin earthquake in southern Iran. The pure DC source model (for frequencies higher than 0.1 Hz) for this earthquake shows two subevents with a 10 km separation from each other.