Performance of regional distance centroid moment tensor inversion applied to the 2004 mid-Niigata prefecture earthquake, Japan

We develop a centroid moment tensor (CMT) inversion method by using a densely distributed regional seismic network in order to calculate the centroid locations and MT solutions of regional earthquakes. Centroid location and time are obtained using a grid search algorithm with the new CMT inversion method. This method is applied to the earthquakes that occurred in the focal area of the 2004 mid-Niigata prefecture earthquake, central Japan. The centroid locations and focal mechanisms of the main shock and aftershocks with moment magnitudes greater than 3.4 are effectively constrained. We assess the reliability of the CMT inversion method by comparing its results with accurate hypocentres calculated with a local temporary seismic network. The accuracy of the centroid locations in the horizontal and vertical directions is approximately less than 4 km. The MT solutions are consistent with the P-wave first-motion polarities. By using the nodal planes of focal mechanisms and the centroid locations of many earthquakes, we successfully imaged the major faults distributed on a vertical cross-section across the main shock fault. Further, we evaluate the validity of the application to 'offshore earthquakes' by using the aftershocks that occur in the landward area, wherein the seismic network does not cover the focal area but is located in a one-sided region of the focal area. These results show that the focal mechanism and depth are very stable, although the accuracy of the horizontal distribution is reduced to half.