Dynamically Consistent Source Time Functions to Invert Kinematic Rupture Histories

In the present study we aim to understand the importance of adopting source time functions (STF) compatible with earthquake dynamics to image the kinematic rupture history on a finite fault. We consider different slip velocity STFs: a boxcar, a modified cosine function and the regularized Yoffe function (Tinti et al., 2005). The latter is consistent with dynamic 'pulse-like' propagation of earthquake ruptures and makes feasible the dynamic interpretation of kinematic slip models. First, considering an array of stations distributed around the source, we compute synthetic waveforms in the frequency band 0-2 Hz commonly used in kinematic inversion. We find that the computed ground motions depend on the choice of the STF, especially at stations located near the source (within about 10 km from the fault). This suggests that the STF may have an important role when used to retrieve kinematic source models by means of waveform inversion. To this purpose we use a two-stages global search algorithm to invert strong motions to recover the rupture history of the earthquake on a finite fault (Piatanesi et al., 2006). This technique performs a statistical analysis of the model ensemble and computes a weighted mean model and its standard deviation, allowing extraction of the most stable features of the earthquake rupture that are consistent with the data and giving an estimate of the variability of each model parameter. We present several synthetic tests and an application to the 2000 western Tottori, Japan, earthquake (M_w 6.6). We find that the choice of the STF affects the inverted rupture model, especially the peak slip velocity and rise time, which are important when kinematic rupture models are used to infer dynamic parameters, such as the slip weakening distance and the dynamic stress drop. We use the inverted rupture histories as boundary conditions in pseudo-dynamic rupture modeling to compute the traction evolutions on the fault plane; we find that relevant dynamic parameters of the rupture process definitely depend on the STF used in the inverse procedure to retrieve the kinematic models.