Based on matched filter technique (MFT) and earthquake relative location technique, a new method is proposed to detect fault structure. The weak matched filter technique (WMFT) is modified from MFT to find more similar earthquakes, which is now used to detected events different from known templates in location. After new events are detected, their locations relative to templates can be given by using time differences from cross correlation. Due to low requirement of the WMFT in wave similarity and its low computational cost, more events can be detected and might help depict the structure of faults clearly. The feasibility of this method is primarily proved by the study of Bohai Bay area using records of 40 days.

The correctness of the earthquake source spectra derived from array data with an iteratively stacking method is checked by analyzing the expressions of iterative stacking in each step. The expression of the finally derived source spectra term shows that it has nothing of the station term, but will be affected by the path term dependent on the source-receiver configuration, which is further confirmed by numerical simulations with iteratively stacking method. Considering stress drop might be wrongly estimated when stations or events are unevenly distributed, the paper provides a strategy to derive the correct stress drop in typical conditions of stationevent configurations. It will be helpful to correctly acquire seismic source information from seismic data.

Theoretical analysis quantitatively shows that high velocity anomaly near source, low velocity anomaly near receiver and the lateral velocity variation above the target inversion area have the influence of the same dimension of anomaly on the traditional inversion of 1-D wave velocity by triplicated wave arrival times. A quantitative computation scheme is proposed to remove the smearing effects with the help of regional or global tomography results when using 1-D inversion by triplicated wave arrival times. Tests imply that the velocity smearing could be eliminated to great extent and the real 1-D structure might be recovered.