An investigation of factors affecting the inversion of a complex earthquake source using synthetic seismograms

In many earthquake source studies, the seismic source is assumed to have a double couple (DC) source mechanism, matching shear motion on a planar fault. Observations of increasing quality and coverage, however, resolve departures from the DC model for many earthquakes and find some earthquakes, especially in volcanic and geothermal areas, that have strongly non-DC mechanisms. Burdick and Mellman (1976) used a timedomain inversion method to determine some of the complexity of the source time function. Several attempts have been made to explain the complexity of body waves from large earthquakes by using a multiple event model (Kikuchi and kanamori, 1982). Deviation of earthquakes from the double-couple (DC) mechanism is an important, but delicate tool to study their source processes. It is believed that the earthquake focal mechanisms not only provide information regarding the stress field, but also give information about the rupture phenomena e.g. crack opening. Hence, the deviation from the double-couple mechanism is a matter of investigation. In fact, the study of the non- DC components makes opportunity to provide information about the earth processes. This information may be used to facilitate the operation of the geothermal energy and help predict volcanic activity. On the other side, the complex time history of energy release is a common attribute of large earthquake failures, as is the presence of non-uniform surface displacement along the outcrop of surface-breaking faults. An interesting topic is the search for the connections between the non-DC events, multiple-double-couple events, segmentation of faults and their fractal properties. Critical papers emphasize difficulties in obtaining reliable non-DC components, e.g. due to noise, poor station coverage or incomplete structural models. Understanding non-DC earthquakes is important both for studying the process of faulting in detail and for identifying nonshear-faulting processes that apparently occur in some earthquakes. To assess the non-double-couple component, a new method is suggested, i.e., a hierarchic grid search of the centroid position and time, during which the double-couple percentage (DC%) convergence is studied (Sokos and Zahradnik, 2008). This article uses the iterative deconvolution of multiple point sources, based on Kikuchi and Kanamori (1991), often used to study complexity of earthquakes. The method was modified for regional distances by Zahradnik et al. (2005). The modification involves the full Green’s functions (Bouchon 1981, 2003). Possibly complex events are represented by multiple point source models, which may represent their isolated asperities (Zahradnik and Sokos, 2008). ISOLA calculates the moment tensors (MT) by the least-square fitting of the complete waveforms in the time domain (Zahradnik and Sokos, 2005). Due to the significance of an earthquake rupture process and investigating whether an event of high non-DC percentage can consist of several subevents with slight spatial and temporal intervals, this survey involves more accurate investigation using synthetic seismograms and their analysis. In this study, synthetic seismograms are produced using CPS software package and the seismograms of the main earthquake and its subevent were summed together and the resulting seismogram provided a basis for further studies. In fact, by modeling a source of high non-DC percentage (resulted from summing two 100 DC subevents), the ability of ISOLA to distinguish two subevents with different focal mechanisms is analyzed. Frequency band, crustal velocity model, moment ratio of earthquakes and the added noise value are determined as factors affecting successful retrieval of the two subevents.