The localization of seismic events is of great importance for hydro frac and reservoir monitoring. For deposits with weak 4-D signatures the passive seismic method may provide an alternative option for reservoir characterization. We introduce a new localization technique which does not require any picking of events in the individual seismograms of the recording network. The localization is performed by a modified diffraction stack of the squared amplitudes of the input seismograms resulting in the image section. The method is target oriented and is best suited for large networks of surface and/or downhole receivers. The source location is obtained from the maximum of the image section for the time window under consideration. Since the focusing analysis is performed only in this section, no optimized search procedures are required. The source time is determined in a second processing step after the source location. Initial tests with 2-D homogeneous media indicate the high potential of the method. Since the maximum of the image section is distinct even very weak events can be detected.
The problem of earthquake localization is one of the most basic problems in seismology. It is stated as follows: Given a set of arrival times and a velocity model, determine the origin time and the coordinates of the hypocenter of the event. This definition inherently assumes that the arrivals of an event are visible on a certain number of stations of the observing array. Moreover, the arrivals have to be identified in the seismograms prior to the actual localization of the event. This not only requires the correct identification of the onset of the arrivals, but also the proper correlation of the event among the different stations of the network. For an overview on techniques using picking of events see, e.g., Thurber and Rabinowitz (2000) for absolute location and, e.g., Waldhauser and Ellsworth (2000) for relative location. Recently, Gajewski and Tessmer (2005) introduced a localization method based on reverse modeling, which does not require any picking of events. Another advantage of this method is the focussing of energy in the back projection process. This allows to image very weak events, which could not be identified in the individual seismogram of the recording network. Back projection can also be performed with high computational efficiency by stacking the amplitudes of seismic traces along diffraction traveltime curves. This approach can be easily implemented in a target-oriented way by just considering the area where events are expected. We will first introduce the methodology of the stacking approach, which is different to stacking seismic reflection data since the excitation time of the source is not known. Numerical 2-D examples using the correct and erroneous velocities illustrate the potential of the method.
In a passive experiment, the network records the events emitted during a certain time period. The start of the recording is denoted by t1 and the end of the recording is t2. The excitation time of the event as well as the location in the subsurface are initially not known.
