Seite - 460 - in Book of Full Papers - Symposium Hydro Engineering

1, F3-2 and F3-3) with approximately the same trike of N310°W. As is shown in Fig. 4, the seismic events after the impoundment are mainly distributed in the area near fault F2 and F3. The strike of 2014-2016 comes much more concentrated and closer to that of fault F3 than the 2002~2005 and 2006~2013. In fact, according to the field survey, the permeability of fault F3 is much bigger than that of fault F1 and F2. Furthermore, the permeability order within the 3 branch faults of F3 is F3-3 > F3-2 > F3-1. And the northwest trending faults such as F2 and F3 are parallel tension-shear fault with steep dip[4]. All the geological conditions mentioned above are beneficial to the infiltration of the increasing reservoir water. 3. DOUBLE-DIFFERENCE LOCATION ALGORITHM 3.1. METHODOLOGY The source parameters such as hypocenter location and focal depth are key points in the study of earthquake mechanism. The ordinary location methods for these source parameters is very complicate and require accurate earth models. Due to our limited knowledge of structures in the earth, it is scarcely possible to get precise location of single earthquake at a time. The main points of the double- difference location algorithm is reformulate the problem to solve for changes in the distance between pairs of seismic events. By forming differences in this manner, we could steer clear of the travel time errors due to the uncomprehending Earth structure and obtain high resolution results of hypocenter location as well as focal depth[5]. 3.2. RESULTS In this study, the double-difference location algorithm is used to conduct data processing of 2014 seismic cluster. The processed results are shown in Fig.5, which indicate the spatial distribution of 2014 seismic clusters are much more concentrated along the strike of branch fault F3-3 compared with that shown in Fig.4. Therefore, we consider the branch fault F3-3 is the causative fault of 2014 seismic cluster. Actually, almost all of the seismic events in 2014 are occurred after the typhoon season. We consider the most possibly reason is that there are some hydraulic connections between the raised water and the branch fault F3-3. The water will erode the fault plane, and reduce the effective stress. In case the water level dropped or changed rapidly, the related fault will loss stability and cause earthquake. This point of view is corroborate with the assumption mentioned in Section2.1. Consider the statistical results of seismic activities and the operating experience, we suggest that the changing rate of water level should be controlled in a relative low level (no more than 2m/day). 460