- Directional resolving power. Because the sources of interest
are relatively weak, distinguishing them from the background will be more
difficult than with conventional seismic data. Nearby surface noise sources
could be as strong or stronger than the sources we are looking for,
depending on local conditions.
In conventional seismic profiling, the source is sufficiently strong that,
in most cases, such out of plane energy is weak enough to be ignored.
But with weaker sources, our experiments must be designed so that
we can determine the direction of arrival of an incoming wave (and hence
process to enhance or suppress it).
A single
line of receivers is inadequate for this purpose. Better is a 2-D or even
3-D array of geophones. If three-component phones are used, a 1-D
array offers some directional resolving power, but as Martin Karrenbach
and I showed (Karrenbach and Cole, 1989), a 2-D array
is still far superior to a 1-D array when three-component phones are used.
Did our previous experiments have sufficient directional resolving power?
Drill bit experiment: NO. A 1-D array of single-component phones was
used, and nearby surface noise sources were a problem.
Passive experiment: YES. A 2-D array was used.
- Knowledge of geology. Some knowledge of the subsurface geology
can obviously be quite helpful. In the drill-bit case, knowledge of the
velocity model allows us to compute the expected moveout trajectory of
drill-bit energy. In a passive experiment, some knowledge of subsurface
structure might allow us to focus our attention on structures that are
likely to scatter a significant amount of energy, and a velocity model
would again be useful in determining moveout trajectories.
Assessment of previous work:
Drill bit experiment: YES.
Passive experiment: NO.
- Receivers at depth. Geophones located on the surface pick up
surface-traveling noises such as wind noise and ground roll from
nearby sources.
In Kostov's drill-bit experiment, three geophones were located in a shallow
well near the borehole. Kostov used these buried phones as a reference
(relying on their insensitivity to surface noise sources) in a correlation
procedure that reduced the influence of surface-traveling noise on his data.
Assessment of previous work:
Drill bit experiment: YES. However, three channels was an
insufficient number. Kostov was unable to combine them effectively
because there were so few.
It would be better to have a downhole array, which
could be beam steered along with the surface array to determine
arrival direction of incident waves.
Passive experiment: NO.
- Adequate spatial sampling. This is a concern in any seismic
experiment, but I note it here to emphasize that the sampling must be
adequate in all dimensions of the array.
Otherwise events will be aliased.
The required sampling interval in any dimension depends on the
frequency of the incident energy and on its moveout.
High frequencies and energy traveling closer to the horizontal
direction necessitate a smaller receiver spacing.
Given a maximum frequency of
interest fmax and a minimum apparent velocity expected
for incoming energy vmin the spatial sampling interval
required in one dimension is given by:
|  |
(1) |
Assessment of previous work:
Drill bit experiment: YES. (But only in 1-D)
Passive experiment: YES. In fact, our geophone spacing (25 feet)
could have been larger. A minimum apparent velocity of 1500 meters/sec
and a maximum frequency of 50 Hz gives a maximum allowable spacing of
15 meters.
However, we were also limited in the overall size
of our array, so we chose a finer spacing.
- Sufficient aperture size. Again, this is a concern in any
experiment. Here we may need the aperture to be sufficiently large for a
number of reasons. If we want to look at scattered energy, our array needs
to be large enough to see the moveout of such scattered energy, which becomes
small as the scatterer becomes distant.
Assessment of previous work:
Drill bit experiment: YES. (But only in 1-D)
Passive experiment: YES. Our array was 500 meters in diameter.
This should allow us to resolve scatterers within about 1 kilometer
of the array center. A larger array would be able to see scattering
from farther away. Figure 1 shows the moveout across the
array (in milliseconds) for scattered waves as a function of distance
from the center of the array. For an array that is 600 meters in
diameter, there is still 20 msec of moveout if the source is at a
distance of 2 kilometers. Thus I feel confident in claiming that our
500 meter array is large enough to see scattering from within 1 kilometer.
Nikolaev and Troitskiy (1987)
perform a
similar scattering analysis using the NORSAR array, which is about 110 km
across.
They look for scattering from structures deep in the crust.
However, they use lower frequencies. If we wish to look at
typical exploration frequencies,
we need to sample more finely in space, hence our
arrays must be more compact and our targets much closer.
