Structural interpretation - 3-D Seismic Interpretation - page 98

where V int is the interval velocity in a layer with rms velocities V a and V b to its top
and base, and T a and T b are the corresponding normal incidence times. The formula
does not take account of ray-bending effects, and gives incorrect results for dipping
interfaces (with dips larger than about 7 ◦ for cases modelled by Al-Chalabi). Reflector
curvature also biases velocities, and is not easy to correct for.
Anisotropy arises as an issue because the velocities determined from seismic pro-
cessing are, broadly speaking, horizontal velocities through the ground; for depth con-
version, we of course need to have vertical velocities. Many rocks, however, exhibit
anisotropy, with horizontal velocities larger than vertical ones. This may be intrinsic
to the rock or an effect of small-scale interbedding of faster and slower lithologies. It
is possible to measure anisotropy (e.g. from long-offset VSP data; Armstrong et al. ,
1995) , but often no direct measurements are available and it has to be inferred from the
comparison of seismic and well data.
In general, migration velocities are closer to true velocities in the ground than are
stacking velocities, because of the removal of structure effects and much of the statics
effect; anisotropy remains a serious factor, however, and it is not possible to use a
migration velocity directly for accurate conversion from time to depth. However, if we
are looking for a way to interpolate velocities between wells, migration velocities can
be useful. We can compare the actual well velocities in a particular formation with
the migration velocities measured at the well locations, and so estimate a correction
factor; if anisotropy does not vary much laterally within the formation, then it should
be possible to use a single correction factor for it across the entire area.
All these comments apply as much to 2-D as to 3-D seismic data. The main benefit
of 3-D data is that the velocity field will have been densely sampled in space. It is
likely to be of better quality than velocities derived from isolated 2-D lines because
of the opportunity to spot mistakes by plotting out sections through the velocity cube
(e.g. horizontal slices), and because it is densely sampled it can easily be smoothed
to remove the effects of statics. However, it is still only really suitable as a way of
interpolating between wells.

3.3.4 Lateral shifts

Sometimes the accurate lateral positioning of events in the seismic dataset is very
important. An example might be the case of planning a well to drill into a fault block on
the upthrown side of a major fault. There may be a need to drill as close to the fault as
possible, perhaps to ensure maximum drainage of a reservoir compartment, but it will
be crucial to drill on the correct side of the fault and not accidentally on the downthrown
side, which might be outside the hydrocarbon accumulation altogether.
Accurate lateral positioning depends mainly on the quality of the seismic migration
process; for modern surveys, any uncertainty in the surface positions of shots and
receivers is negligible by comparison. It is important to realise that migration may