## CMT solution determined by Direct Solution Method

#### We performed centroid moment tensor inversion using long period
(50-100s) body wave data following the grid search approach by
Hara (2004, 2005). The obtained solution probably corresponds to
the large moment release in the later part of the source process
found by the finite fault models (e.g., Yagi, http://www.geo.tsukuba.ac.jp/press_HP/yagi/EQ/20070816/;
Ji and Zeng, http://earthquake.usgs.gov/eqcenter/eqinthenews/2007/us2007gbcv/finite_fault.php;
Yamanaka, http://www.seis.nagoya-u.ac.jp/sanchu/Seismo_Note/2007/NGY3.html).

####
0rigin time (BIGQUAKE): 2007/08/15 23:40:57

Centroid time: 2007/08/15 23:42:13

Centroid (lat, lon): -13.76 -76.64

Centroid depth (km): 40.00

Half duration (s): 13.40

Mrr: 4.89 Mtt:-1.29 Mff:-3.59 Mrt: 0.72 Mrf:-6.49 Mtf: 2.93

(unit of moment tensor components: 1e27 dyne*cm)

Figure 1. The obtained CMT solution. The green star denotes the PDE epicenter.

Figure 2. Summary of the grid search. The upper and lower panels show
the variance reductions and focal mechanisms, respectively.
The optimal solution is represented by the solid circle in the
upper panel and the red focal mechanism in the lower panel, respectively.
After fixing the centroid location, we perform the grid search
only for centroid time. We used the centroid location of the Global CMT solution (http://www.globalcmt.org/)
as a reference point for grid search. We obtained a similar solution when
we used the PDE hypocenter as a reference.

#####
References

Hara, T., 2004. Earth Planets Space, 56, 307-310.

Hara, T., 2005. Earth Planets Space, 57, 179-183.