Page 243 - Fundamentals of Gas Shale Reservoirs

MONITORING PASSIVE SEISMIC EMISSIONS WITH SURFACE AND SHALLOW BURIED ARRAYS 223
• For many areas such as the Marcellus Shale Fm., recorded traces must be analyzed for all time windows over
there is substantial topography and greater variation in the time interval of interest (e.g., over the pumping time of a
the near‐surface velocity. Buried geophones in the frac stage). Each time window must be focused (flattened)
Marcellus are typically on the order of 30 m (100 ft) for every depth voxel in the volume of interest. The work­
deep. For a near‐surface velocity of 3700 m/s (12,000 flow and algorithms used for SET are the same as is used for
ft/s), the delay between the up‐going and down‐going prestack depth migration (PSDM) in surface reflection
signal is 17 ms. This delay is such that the ghosting on seismic data. The primary difference between PSDM and
the buried geophone interferes with the waveform of SET is that PSDM uses two‐way travel times for focusing
the signal and can lead to detection and location accu­ while SET uses one‐wave travel times.
racy problems for the buried geophones. The ghosting SET is performed as follows. First, the study volume is
also causes notches in the amplitude spectra which divided into voxels, and ray tracing is used to compute the
impacts moment tensor inversions. To improve travel time from the center of every voxel to every receiver
correction of such ghosting, shallow buried arrays often (Fig. 10.12). Typical voxel sizes are 8–15 m (25–50 ft) on
have multiple receivers at different levels in each hole. edge. Second, the traces from all receivers are aligned in
time (flattened) using the one‐way travel‐time computations.
In summary, surface array data must be filtered to eliminate This process aligns the traces as if they had all been emitted
the coherent noise that is inherent to surface array record­ from the voxel of interest. In the last step, a short time
ings, and buried array data must be deghosted to improve window (typically 100 to a few hundred milliseconds) is
location accuracy, focal mechanism solutions, and moment stepped over the data through the time interval of interest
tensor inversions. and a measure of seismic activity is computed within the
MEQs recorded with downhole sensors typically have window by comparing all of the flattened traces. At each
strong amplitude in the frequencies above 100 Hz and show time step, the window is moved only a portion of its length
very broad band amplitude spectra (Fig. 10.11). For surface so that overlapping windows are computed ensuring that all
recordings, the emissions from the rock movements at activity is captured. Different workers may use any of various
depth must travel through the entire section of the earth to measures of seismic activity including stacking, semblance,
reach the geophones at the surface. This propagation path coherence, and cross‐correlation methods. This procedure is
and distance results in attenuation and scattering, and there repeated for every voxel resulting in a multidimensional data
is a loss of the higher frequencies. Because higher frequencies volume whose dimensions are the X,Y,Z coordinates of
attenuate more rapidly, these frequencies are lost. The the depth voxels, the time step, and the activity measure.
dominant pass band of the rock column is below 60 Hz. Consequently, a depth volume of activity is generated at
However, frequencies below 15–20 Hz are often so contam­ each time step. For a single frac stage that lasts for 2.5 hours,
inated with surface noise that they must be filtered out. there will be more than one hundred thousand depth volumes,
Considering both low‐frequency surface wave noise and and millions for the entire well job.
attenuation of higher frequencies during wave propagation, An important feature of SET is that it images total seismic
the signal available for analysis at the surface is typically in activity within each time step, and thereby captures much
the 20–60 Hz band. As described in the following sections, more energy than conventional seismological methods that
a complete suite of microseismic products can be generated locate discrete MEQs only. Imaging total trace energy cap­
using only this frequency band, although only larger magni­ tures both more MEQ energy and seismic energy from other
tude MEQs are detected. phenomena that are ignored by conventional methods, such
as LPLD activity (Das and Zoback, 2013a, b) and EDS
(Sicking et al., 2014).
10.5.2 Seismic Emission Tomography
SET captures more MEQ energy because seismological
This section discusses processing of surface and near‐ methods can only be used on MEQs that are sufficiently
surface passive seismic data. SET is the fundamental under­ large and clear to identify as distinct events and that have
pinning of the entire processing chain. SET is used for the clear, identifiable P‐wave and S‐wave arrivals. Earthquake
detection and location of MEQs, focal mechanism analysis, frequency versus size distributions are linear when plotted
imaging cumulative seismic activity, and constructing on log–log plots (i.e., log of earthquake frequency vs. mag­
fracture images from the cumulative activity volumes. nitude; magnitude is a logarithmic measure of signal
amplitude). The slope and intercept of the line, respectively,
10.5.2.1 Overview SET is the method of choice for detect­ termed b‐value and a‐value, may vary with location and
ing and locating seismic emissions from the subsurface. failure type, but the distribution is linear. Large, clear earth­
Seismic emissions from rock movements in the subsurface quakes are consequently less abundant than smaller, less dis­
arrive at sporadic times and from various locations. In order tinct earthquakes that cannot be utilized for conventional
to detect these emissions and locate them accurately, the seismology. These smaller earthquakes combined emit more