A sequence of M7-class interplate earthquakes (M6.7-7.3) occurred from 2003 to 2011 before the occurrence of the M9 2011 Tohoku-oki earthquake. To investigate the relationship between these previous events and the giant earthquake, we determined the source processes of these events by joint inversion method. We performed the inversions by using both near source and teleseismic body-wave data. To perform a stable inversion, we applied smoothing constraints, and determined their optimal relative weights on the observed data using an optimized Akaike\\\\\\\'s Bayesian Information Criterion (ABIC). The results show that the rupture areas of these M7-class events were nearly covered by the rupture area of the 2011 Tohoku-oki earthquake determined by Yagi and Fukahata (2011). The epicenter or the initial rupture of the giant earthquake was located in the un-rupture region surrounded by the rupture areas of the 2003 Fukushima-ken-oki earthquake, the 2005 Miyagi-ken-oki earthquake and the 2011 Sanriku-oki earthquake. At the first stage of the source process, the rupture propagated southeast and broke the asperity where the maximum slip occurred, and the asperity was also located in the un-rupture region which was close to the source regions of the 2003 Fukushima-ken-oki earthquake and the 2011 Sanriku-oki earthquake.