Previous studies (Pinel-Puysségur et al., 2014; Pezzo et al., 2014) have suggested a straight fault
geometry, and our data did not strongly indicate curved geometries either. Therefore, we first
tested different depths and dip angles, by keeping the strike of all the faults straight. However,
significant residuals remained, especially in the descending data, even for the best straight fault
model (Figs. S1, S2). To resolve these unexplained signals, we carefully examined the data, and
found that the descending observation in Fig. 4a suggests a clear bend in RLF1 around the
central point. In addition, RLF2 exhibits a curved strike. The modeling conducted by following
the curved geometry for right lateral faults (RLF1 and RLF2), explained the observed
deformation quite successfully (Figs. 3, 4, 9).
Examination of the fault mechanism solutions (Fig. 2) suggests that the strike-slip contribution
was more dominant than the dip-slip component of the main shocks and most of the aftershock
crustal deformation. The maximum observed magnitudes were 6.4 and 6.4 (Mw) for the main
shocks and 5.7 (Mw) for the aftershocks (Table 1). Thus, for the source modeling, we attempted
to produce a model that exhibits nearly the same characteristics in terms of slip components, and
calculated the magnitudes along with minimum possible residuals and logical slip distributions.
As it was established that LLF1 contributed to the crustal deformations resulting from both the
main shocks and aftershocks, we produced an optimum fault model (Figs. 5 and 10) by following
a trial-and-error process, i.e., various possibilities related to the depth and dip of LLF1 were
analyzed, and the best result was selected. LLF1, with the same geometry, was imported to the
co-seismic model, and again, a trial-and-error process was performed by changing the depths and
dip angles of RLF1, RLF2, and LLF2. The best model selected (Figs. 3, 4, and 10) has a strikeslip component dominating over the dip-slip, with maximum slip distributed along conjugate
faults (RLF1 and LLF1), and the calculated magnitudes (Mw) are 6.3 for LLF1 and 6.5 for RLF1,
very close to the observed moment magnitudes.
Figure S1. Modeling results following straight fault geometry; areas enclosed in rectangular
boxes in the descending and ascending observation are shown in Figs. 6 and 7.
Figure S2. Slip distribution for the seismic sequence (faults with straight strikes). The top of the
faults is 300 m below the crust surface. The calculated magnitudes (Mw) for RLF1, RLF2, LLF1,
and LLF2 are 6.3, 6, 6.3 and 5.8, respectively.