You are on page 1of 8

Seismic modeling

The section is split into two parts:


1. Modeling techniques with use the vertical description of the subsurface
provided by well logs, such as GEOGRAM and SYNTHETIC VSP.
2. Ray tracing techniques which use a 2-D model of the subsurface and
compute of a response of a limited number of reflections.

GEOGRAM
The following Fig. 3-54 shows a typical GEOGRAM presentation from the Gulf
of Suez. The familiar evaporate section are present and these caused difficulty in
interpreting the surface seismic section below the salt. The first problem for the
geophysicist is to tie the depth derived well logs into the seismic time scale.

Fig. 3-54. Composite GEOGRAM presentation.

The depth scale is plotted together with the way time scale, and on the right hand
side the Sonic and Density logs are displayed with the reflection coefficient series.

The first set of traces on the left (primaries only) corresponds to the convolved
reflection coefficient series. It simulates an ideal surface seismogram where the
initial seismic pulse is propagated without being affected by transmission losses,
and being reflected only once.

The second set of incorporates the transmission coefficient, and the third set of
traces is the filtered reflectivity series. The physical reflection process is simulated
by taking into account the transmission losses and multiples.

The primaries and transmission losses are removed from the third set of data to
obtain the fourth set.

Several features are apparent from this GEOGRAM processing:

1. The primaries are very rapidly attenuated by the transmission


coefficients, below 1.7 S.
2. The set of traces "multiples only" is very similar to the set of traces
"primaries and multiples".
3. Most of the primary reflections are present on the "multiples only"
display.
4. These logs are typical of the Gulf of Suez with a thick evaporates layer
( 1.6 to 2.05 S), the top of which corresponds to the onset of strong
attenuation.

The top part of this section (1.6 to 1.75 S) comprises alternately high velocity strips of
salt or anhydrite and low velocity strips of shale, which produce a "cyclic" reflection
coefficient series.

The seismic pulse which propagates through the subsurface is the sum of a direct
pulse which is rapidly attenuated by the transmission coefficients, and of short-term
multiples which are delayed by an additional path.
Short term multiples are so numerous, due to the many strips of salt that they interfere
constructively so that the amplitude of the resultant wave packet may be greater than
the direct pulse, by several orders of magnitude. This is the reason why the two set of
traces, "multiples only" and "primaries and multiples", are so similar.

The main reflectors present on the "primaries only" traces can be seen on the
"primaries and multiples" display but not exactly at the same time. It can be seen by
comparing the strong horizon at 1.97 to 1.98 S, corresponding to a layer of anhydrite
on both sets of traces, that is delayed by 20 mS correspond to the delay between the
onset of the seismic pulse (corresponding to the true transit time) and its maximum
(corresponding to the energy packet).

As seen in the Fig. 33-5 the signal character is very similar between the displays and
most of the reflectors can be identified down to 2.2 S. Below 2.2 S the existence of
the reflectors seen on the surface seismic questionable.

The GEOGRAM presentation with primaries and multiples, as seen in Fig. 3-56
matches the seismic section very closely, and is considerable improved particularly
below 2.0 S. It can be seen that on the surface seismic most of the deep reflections are
distorted or destroyed by residual multiples that are not removed in the processing.

It is interesting to note in Fig. 3-55 that the reflection coefficient measured in the well
(primaries only) between 2.1 and 2.24 S are very similar to those seen on the seismic
section towards the right of the well. The strong horizon seen at 2.65 S on the seismic
section appears to be a multiple.

Fig. 3-55. GEOGRAM presentation with primaries only superimposed upon the
seismic section at the position of the well.

Fig. 3-56. GEOGRAM presentation with primaries and multiples, superimposed on


the seismic section at the position of the well.

SYNTHETIC VERTICAL SEISMIC PROFILE

The SYNTHETIC VSP is similar to the GEOGRAM technique in that it relies on the
same assumptions. The surface is described as a series of homogeneous and isotropic
layers, each having a constant acoustic wave velocity and a constant density.

The Sonic and Density logs, once corrected for anomalies such as invasion and bad
hole effects are converted along a time scale to provide the description of the model.

The propagation of a compressional wave at normal incidence can then be simulated.


At each interface between two layers having the same travel time, a downgoing wave
Dj imparts its energy into a reflected wave Uj and a transmitted wave Dj+1 given by
the following equation:

Uj = DjRj + (1+Rj) Uj+1


Dj+1 = (1- Rj) Dj- Rj Uj+1
Fig. 3-57. Reflection and transmission of waves at an interface.

The reflection coefficient R, is given by the usual relationship between the velocity V
and density ρ of two adjacent layers:

Rj =

In order to compute the upgoing and downgoing waves in a recursive manner for all
layers, two additional assumptions are mode:
1. The reflection coefficient of the top interface has a value very close to
1 so that no energy is transmitted across the boundary into air.
2. The last layer is assumed to be infinite towards the bottom and no
energy comes back from below the well.

Fig. 3-58. SYNTHETIC VSP response to unity energy input from surface.

Starting from the upper left hand position of the diagram, each individual ray path is
computed. If, for example, we want to estimate the energy reaching the surface at a
time equal to 4 ∆t, we must compute all the ray paths contained within the triangle
ABC. If we want to estimate the energy appearing at time 10 ∆t at a geophone buried
in layer Z2, we must compute all the path contained within the trapezium ABCD.

GEOGRAM is obtained by computing the upgoing wave within the first layer.

The VSP response at a given geophone position is obtained by computing both the
upgoing wave and the downgoing wave which are traveling in the layer corresponding
to this level.

When recording actual VSPs we generate a wave whose initial energy spreads out on
a spherical wavfront (in the case of a homogeneous medium). Consequently the signal
amplitude recorded by the downwhole geophone decrease as the inverse of the
distance from the source. When dealing with a layered medium the phenomenon is
even more complex because the radius of curvature of the wavefront is altered
crossing each interface, thereby generating an overall front which is no longer
spherical.

Fig.3-59. SYNTHETIC VSP of model shown on left hand side of diagram.

A newly developed algorithm has been included in the SYNTHETIC VSP software
providing an approximation to the spherical divergence.
In the top part of the figure the waves is plotted with primary events only (each ray
path is reflected only once). On the bottom part the more general case is shown.
As for the GEOGRAM processing, the impulse response has to be filtered within a
limited frequency band in order to match real seismic data such as surface recorded
sections or VSPs.

SYTHETIC VSP Application


Applications of the SYNTHETIC VSP are the same as for GEOGRAM. However, in
addition, it provides modeling of the amplitude of the acoustic waves, and enables
determination of the exact depth where amplitudes are generated.
In order to demonstrate this technique, a field example from the Gulf of Suez will be
used.
Sonic and Density log are input in the same way as for GEOGRAM processing.
It can be seen that the initial pulse produced from surface widens progressively as it
propagates, and short period multiples combine to constitute a long tail. The major
part of the wave train's energy is no longer contained within the first pulse, whose
amplitude decrease rapidly due to transmission losses, but rather by the multiples.

It conformed to the wave propagation theory and may be used as input data to test
seismic software, and its ability to process actual data and provide the interpreter with
the true reflection coefficient series.

The wavefield exhibits strips corresponding to intervals between strong reflectors.


The blank zone in Fig. 3-61 between 0.5 and0.58 S in the vicinity of the direct arrival
curve corresponds to an homogeneous salt layer which does not produce ant
reflections, both its velocity and density being constant.

Fig. 3-62. SYNTHETIC VSP, filtered upgoing wave with primaries and multiples.
The wave seen in Fig. 3-62 is filtered by a Richer wavelet with a central frequency of
14 Hz. This wave, which is corrected to two way time, can be matched to the surface
seismic section and used to convert from a time to depth scale.
The first trace on the right hand side of the display represents the upgoing wave as
would be recorded at the shallowest logging depth; it is equivalent to the conventional
GEOGRAM display with all order multiples.

The major advantage of the SYNTHETIC VSP over the GEOGRAM technique is that
it allows the interpreter to determine where the multiples come from. The arrows
indicate multiple events which are clearly created by the top of the salt.
The base of the evaporate layers create multiples which interfere with the genuine
reflections generated below. This interference of multiples with true reflectors is so
strong that, when considering the trace computed at the top of the available logs, it is
impossible to see the reflected energy.
Fig.3-36 is comparable to a real VSP upgoing waves assuming that the VSP
processing was able to remove all the multiples.

Fig. 3-63. SYNTHETIC VSP, upgoing waves with primaries only. The GEOGRAM
traces corresponding to the two extreme cases seen on the SYTNETIC VSP, no
transmission effects.

Considering events in the vicinity of the direct arrival curve is equivalent to


GEOGRAM with primaries only. The extreme right edge of the plot corresponds to a
GEOGRAM processing which would have been computed with a spherical
divergence approximation. This presentation points out the interfaces which are
strongly absorbers; for example the top of the evaporate absorbs an amount of energy
roughly equivalent to that which the plot can represent.
Because it models the propagation of the seismic signal as a function of depth, the
SYSTHETIC VSP enables us to understand the phenomena giving rise to signal
distortions.

Fig. 3-64 shows a compression of the SYNTHETIC VSP with actual VSP, the
example shown is also from the Gulf of Suez, and is same one as discussed in the
section on GEOGRAM technique.

Fig. 3-64. SYNTHETIC VSP with primaries and multiples compared to the actual
VSP.

Conclusions
1. The SYNTHETIC VSP is a extension of the GEOGRAM technique.
2. It helps the interpreter to understand how the multiples are generated,
and provides him with a powerful tool for analyzing both surface seismic and
borehole seismic.
3. Under the standard assumptions of an horizontally layered surface, it
may help in developing and testing new processing algorithms such as
velocity filters and deconvolution.

RAY TRACING MODELLING

When dealing with tectonized environments such as the Gulf of Suez,


geophysicists find that the usual CDP stack processing is incorrect because of
highly diffracting layers. Ray tracing technique are often used for converting data
into depth.
Well seismic methods such as VSPs are essential uni-dimensional. When recorded
through complex structures, the geometry can also become extremely
complicated, and in this case interpretation is greatly assisted by making use of a
modeling capability which takes into account 2-D geometry.

Modeling VSP
A conventional VSP is recorded by placing a seismic source on surface close to the
borehole, and sampling the seismic wavefield at various geophone locations along he
borehole axis.
If the horizons under consideration are horizontal and the downgoing and upgoing
wave fronts passing by the geophone are propagating vertically, then the seismic
echoes recorded along the borehole are from reflectors at the borhloe axis. In the
previous section we have seen examples of models where the VSP response exhibits
upgoing events symmetrical with respect to the downgoing events.

Fig. 3-65. Structural model with an unconformity.

In this section we will assume a formation model, as for example in Fig. 3-65 and use
ray tracing modeling to investigate the seismic wave paths.
The model chosen has a dip angle of 30˚, and it is assumed that each layer is
thoroughly defined by its acoustic velocity and density. The parameters used in this
example are shown in the figure. Although shear waves, pressure waves and mode
conversions can be handled by the software, only the pressure waves have been
modeled for the purpose of this example.
The succession of diagram seen in Fig. 3-66 shows how each horizon can be treated
separately in order to visualize the reflection process more clearly. Some of these
diagrams show that parts of the reflectors may be "hidden" due to the diverging
income rays. In each of these examples the blind zone corresponds to the pinch-out at
the unconformaity.
It is seen that the reflecting points are no longer confined to the well axis but that they
tend to "migrate" up-dip. Consequently the arrival time corresponding to these
reflections is not symmetrical with the direct transit time.

This departure from symmetry must be taken into account during the processing by
using a velocity filter which allows for time moveout in the up-going wave energy
with respect to its ideal direction.

From the curvature of the upgoing events, in the vicinity of the first arrival curve, one
can estimate the dip of the reflectors, using the interval velocity just above the
reflector.

Fig. 3-66. acoustic ray tracing simulating a VSP with 40 geophone positions, using
the model seen in Fig. 3-65.

It can be seen that the echo coming from the dipping reflector cannot be recorded by
the shallowest geophones because normal incidence us reached earlier than one would
expect from the depth model.

The upgoing reflections are not horizontal and their time of arrival decreases as the
reflecting point moves updip. In the limiting case, when the geophone is at ground
level, both source and geophone are the CDP time of a conventionally processed
surface seismic section.

Fig. 3-67. Waveforms computed from the model, filtered with a Ricker wavelet.

Fig. 3-68. The upgoing wave corrected to two way time as it would be after
processing of an actual VSP.

Fig. 3-69. VSP modeled from horizontal layers for comparison.

Fig. 3-70. Upgoing wave in two way time corresponding to VSP in Fig. 3-69.

The effect of the dipping horizons in the model seen in Fig. 3-69 can be evaluated by
comparison with Figs. 3-69 and 3-70 in which the VSP was obtained with the deepest
horizons and intersecting the well at the same depth. The upgoing reflections are flat
and horizontal at a time corresponding to the vertical ray path.

In other words, when looking at an upgoing reflection close to the first arrival, we are
considering a migrated time (the ray path is vertical down to the reflector) whereas
when looking at an upgoing wave seen by a geophone close to surface, we are looking
at an unmigrated time (the ray path is normal to the reflector).
The consequence of this is that when matching a VSP corridor stack or a GEOGRAM
display to an unmigrated seismic section, the interpreter may be led either to squeeze
the well seismic trace or to stretch the surface seismic trace in order to match them.

Fig. 3-71. unmigrated surface trace recorded at the well position and the time
migrated trace.

The VSP upgoing wave corrected to two way time provides a link between the
migrated trace and the unmigrated trace.

The SYNTHETIC VSP computed by 2-D ray tracing provides the interpreter with a
tool for a better understanding of actual VSPs and allows him to confirm this model.

Modeling offset shooting

The offset shooting method requires computation of the ray paths associated with
each proposed source position, and geophone configuration prior to its processing. In
the previous section it was seen that a model of the surface may be built using mainly
the vertical velocity profile from the VSP previously recorded in the well.

In order to optimize the layout geometry for the acquisition phase, it is advisable to
model the survey in advance, by means of a ray tracing technique.

Fig. 3-72. Shows the position of the well with respect to a model of a faulted block
typical of the Gulf of Suez. The well is 4000m deep, and the fault approximately
500m from the borehole.

The layers have interval velocities ranging from 2500m/s for the shallowest one to
4000m/s for the deepest. The second layer corresponds to salt having a velocity of
5000m/s.

As seen in Fig . 3-73 the ray tracing allows us to visualize the way in which the upper
surface of the fault plane will be illuminated, and consequently to determine the
optimum offset position of the source.

It can be seen that the salt layer behaves as a strong lens which focuses a significant
amount of energy onto a restricted area below the source position. For offset values
equal to 1750 m and less, the downthrown section of the reflector is well covered, and
for an offset equal to 750 m both parts of it are illuminated.

Fig. 3.73. Series of results of modeling the reflection of an acoustic wave from the
first faulted interface. Note the source positions are regularly spaced every 250 m.

In practice the more complicated the structures, the more care must be taken in
selecting the optimum offset positions.

No single source position can provide a satisfactory coverage of the target and, in
particular, the image of the reflectors at the borehole location (given by a VSP) must
be linked to their image further away. From the results of this ray modeling, a near
offset source location such as 500 m (Fig. 3-73 g) would be selected, together with a
far offset position such as 1250 m (Fig. 3-73 d).

Having chosen the two offset position for optimum coverage of the upper faulted
interface; see Fig. 3-74, it is important to estimate how these positions would
illuminate the lower formations, as seen in Fig 3-75 and 3-76.

Ray tracing modeling has proved to be a useful tool for the geophysicist, providing
him with means of optimizing the design of an offset shooting survey in such a way as
to illuminate the horizons of the model that he has postulated. It also allows him to
experiment with other possible models.

The complex structures of the Gulf of Suez lend themselves to this approach.

Fig. 3-74. Ray paths corresponding to the near and far offset positions chosen, for
optimum illumination of the upper faulted interface.

Fig. 3-75. Coverage of the middle interface with source positioned at 500 and 1250 m.

Fig. 3-76. Coverage of the lower interface with source positioned at 500 and 1250 m.

You might also like