The idea of construction of algorithm and results of developed software on the determination of microearthquake locations due to frac job is given. Influence of anisotropy on the accuracy of frac event locations is briefly discussed. Methodology of determination of horizontal component of geophones is developed based on perfshots and earlier located frac events data. Produced software works in real time and in automated regime.
Building of efficient program applications for hydraulic-fracture- induced microseismic event detection and location from observations in an offsetting monitor well, remains a topical problem in micro-seismic mapping technologies. Here, the performance criteria are as follows:
(1) a high degree tolerance of the extracted arrivals and the resulting event locations, to various interferences, including the background noise, multiple reflections, multiple P- and S-wave arrivals as a result of multi-step time histories of dilating cracks, as well as to unfavorable source-receiver geometries;
(2) a reasonable CPU time per micro-seismic event;
(3) the data processing should assume no attendance except at the geophone orientation calibration and seismic velocity model building stages.
Signal Extraction
The algorithms and methodology that we developed for automated processing of the 3C seismograms, including that in a real time mode, base themselves on some analogies with several distinctive features of visual perception, where isolation of an object from the surroundings may be regarded as being a result of clusterization of instantaneous foci of the attention as the eye cans the points within the range of vision. Specifically, the visual snap points describe trajectories with a tendency to form clusters in the vicinity of the object of interest.
In an effort to model the visual analysis of the seismograms, we associate the snap point clusters with the respective sub-domains in a suitably chosen pattern space, among the dominant patterns of which may be cited triaxial
within a working time window.
By analogy with the visualization procedure where a great many pair-wise comparisons of adjacent fragments of a 3- component seismogram are performed, we consider a special-purpose transformation of the original seismogram as a possible formalization of the signal extraction process:
Let at (f; t) be the sum of time-dependent scalar quantities, f, from a 3-C seismogram,
, taken over time samples t,…, t + t. In particular, in the case f = ||
(t)||2 the at(f; t) above is the sum of squared amplitudes of the observed seismic wave motion within the current time window [t, t + t ] (an energy estimate within constant factor).
with t = t
p
being the time window size as a monotonically increasing function of control parameter p varying over an interval [p1, p2], thus defining systems of nested time windows within a working time domain (0, T).
Note that the m(f; t) functional can be regarded as
f(t)
due to a seismic wave arrival.
If F = {fn (t), n = 1, .., N} is a sequence of scalar quantity estimates from the 3C-recording
(t) of a seismic event, then a certain amount of information on the waveform dynamics variation rate at the point in time t can be obtained by a suitably weighted summation of the abovementioned m(f ; t) over f Î F:
