Download as pdf or txt
Download as pdf or txt
You are on page 1of 10

INTEGRATED SEISMIC RESERVOIR CHARACTERIZATION AND MODELING:

A GULF OF MEXICO 3D CASE HISTORY

Paper submitted for GCSSEPM 1996 Research Conference

REBECCA B. LATIMER AND PAUL van RIEL

Jason Geosystems, 1155 Dairy Ashford Rd Suite 705, Houston, Texas 77079

ABSTRACT

A quantitative, detailed description of reservoir architecture and properties which honors all available well log, seismic and
geologic information can strongly improve the economics of reservoir development and enhance production. Spatial distribution of
reservoir properties can be determined by stepping through a workflow which starts where standard workstation seismic and geologic
interpretation ends. In order to obtain the most accurate and detailed results, one must design a multidisciplinary workflow that
quantitatively integrates well log, petrophysical and seismic information. The main steps are:

• Seismic well calibration;


• Wavelet Estimation
• 3D seismic lithology inversion;
• 3D geologic modeling, utilizing interpolated well control;
• Detailed interpretation of reservoir stratigraphic units using a seismic lithology inversion cube merged with the geologic model;
• Estimation of reservoir parameters such as average porosity and net pay from the lithology cube.

Following this workflow through a Gulf of Mexico 3D case study shows how the reservoir’s stratigraphic and lithologic
properties are more accurately determined than would be possible with standard seismic interpretation and seismic attribute analysis.
The seismic lithology analysis allows for delineation of depositional systems tracts, which can be superimposed upon the seismic
interpretation. The method also identifies lithologic changes away from well control. These results are directly related to reservoir
properties and are used to support the optimization of the development drilling program.

INTRODUCTION

It is critical to obtain an accurate description of the reservoir architecture and lithology variations within the
reservoir units in order to optimize field development. In the majority of reservoir development projects, the description
of the reservoir away from well control is determined from 3D seismic data. The standard approach to incorporating
seismic information, after structural interpretation, involves:

• Detailed sequence stratigraphic analysis of the reservoir units to determine the depositional history, reservoir
architecture and facies distribution.
• Analysis of seismic data attributes, in particular seismic amplitude, to determine variations in reservoir lithology,
morphology and possible connectivity between reservoirs.

In the procedures above, the available well information provides calibration points and the lateral variations in the
seismic attributes provide the details on the distribution of reservoir properties away from well control.

The seismic data, being band limited in nature, limits the resolution. Due to this inherent limitation, the reliability of
the above standard seismic interpretation approach suffers due to wavelet effects and tuning. As a result:
• Seismic amplitude variations may largely be due to seismic event tuning (waveform interference) rather than
lithology changes.
• Layer thicknesses measured from the seismic are no longer equivalent to true layer thicknesses.
• Variations in the rock properties surrounding the reservoir units can cause changes in seismic amplitudes, thereby
masking those caused by changes within the reservoir units.
• Seismic amplitude attribute analysis can not distinguish between the lithology variations in the reservoir units and
those in the surrounding rocks.

The objective of seismic lithology inversion is to address these problems by converting the seismic data to lithology
data. The advantage is that lithology data represents layer properties and not seismic reflection interface properties. The
full seismic trace information is used in deriving seismic lithology data, as opposed to attribute analysis which only uses a
very limited part of the seismic trace information.

Several lithology parameters can be reconstructed from seismic data; the most common of which is acoustic
impedance (the product of density and interval velocity). Acoustic impedance is directly related to the seismic reflections
and should show lithologic variations where they are present and resolvable in the seismic data. Velocity and density, and
therefore acoustic impedance, are impacted by porosity, sand/shale ratio and fluid fill. In fact, through careful
petrophysical analysis, acoustic impedance can be directly related to such reservoir parameters, so that reservoir
attributes such as net pay or porosity can directly be derived from the acoustic impedance data. This makes acoustic
impedance a highly useful parameter in reservoir characterization.

This study focuses on the use of acoustic impedances to improve detailed reservoir interpretation and to improve the
accuracy with which lateral variations in reservoir lithology can be determined. In order to analyze the effectiveness of
the acoustic impedance method, it is applied in the study of a Gulf of Mexico field. Results are compared to those
obtained with the standard seismic interpretation approach using seismic amplitude attributes.

GEOLOGICAL OVERVIEW

The study area consists of an approximately 24 sq. km area on the west flank of a faulted anticline, represented by
the enclosed 3-D survey. The study area is located in the Plio-Pleistocene flexure trend in the Gulf of Mexico. The
reserves are trapped within five zones in the Middle Pliocene section, two main reservoir zones (Blue and Green) and 3
thinner, secondary reservoir zones (Pink, Orange and Red).

The entire section is thought to consist of two main sequences (Mayall et. al., 1992). The upper sequence is
composed of the lowermost Green progradational unit and the upper-most Red transgressive package. The upper-most
reservoir (Red) is thin and is thought to be composed of low density grain flows (turbidity deposits). The Green sand
package was deposited as overlapping packages of clinoforms in a prograding shelf margin deltaic sequence. The
reservoir is partially fault controlled and contains both stratigraphic and structural components. The reservoir is
continuous wherever the sands are in contact, e.g. overlapping.

The lower-most sequence consists of a lower (main) progradational unit (Blue) and two upper transgressive turbidite
sections (Orange and Pink) which separate the main Blue and Green sands. The Orange and Pink packages cover a larger
interval than the massive sands but contain thin sands and thus are difficult to track. The sand/shale ratios are lower in the
turbidite packages and the boundaries are not as clear. These turbidite units are used to add to the net pay as upside
potential. The lower zone (Blue layer) is thought to be deposited as a prograding package of delta front slumps and delta
mouth bars. The layer appears to have more distinct sand lenses, separated by larger shale interbeds. Connectivity of
reservoirs is not apparent in this zone. Both the Green and Blue sands appear to be elongated in the north-east to south-
west direction which is parallel to the paleo shelf margin. Due to the stratigraphic variability, the necessity of detailing
the reservoir architecture and lateral lithology variations is paramount in exploiting the reserves.

Gulf of Mexico 3D Case History November 16, 1996 Page 2


SEISMIC INTERPRETATION AND AMPLITUDE ANALYSIS

As discussed in the introduction, seismic data is inherently limited in resolution. The bandwidth of this seismic data
set is 6-50 Hz. This limits the reliability of the standard seismic interpretation when the sands are thin - less than 30 feet.
These thin sands are present in the turbidite layers (Red, Pink and Orange) and are found in the overlapping zones of the
clinoforms in the Green and Blue sands. Additionally, the standard approach is limited when there are variations in the
properties of the rock surrounding the reservoir sands. Both of these are common problems, and also occur in the study
area, which is illustrated in Figures 1, 2, 3, and 4.

Figure 1: Basemap showing the input interpretation for the green horizon and the
location of the arbitrary seismic transect through wells 3, 3ST1, 5, and 4.

Gulf of Mexico 3D Case History November 16, 1996 Page 3


Figure 2: Arbitrary seismic line through wells 3, 3ST1, 5, and 4. Horizon
interpretation and impedance well log curves are overlain.

Figure 2 shows a seismic line through wells 3, 3ST1, 5, and 4. The log curves shown are acoustic impedance
(product of the measured density and velocity) logs. In this reservoir, low acoustic impedance values correspond to
reservoir sands with pay. The seismic data is reverse polarity, so that a seismic trough corresponds to a negative
reflection coefficient, i.e., the start of a low acoustic impedance zone. The most negative (high) amplitudes are shown by
the yellow colors on the seismic amplitude maps of the Green and Blue sands shown in figures 3 and 4.

Figure 3: Seismic amplitude map corresponding to the Green sand


trough event. The anomalous amplitudes (most negative) are
yellow. Main fault zones are seen as trending SW-NE

Gulf of Mexico 3D Case History November 16, 1996 Page 4


Figure 4: Seismic amplitude map corresponding to the Blue sand trough
event. Anomalous amplitudes (the most negative) are
yellow. Main faults trend SW-NE.

Two interpretations have been made on the seismic data set. The first interpretation was made by tracking the trough
on the seismic for the Green sand and the Blue sand. This methodology was difficult due to the fact that the sands are
discontinuous and therefore could not be “tracked” over the entire extent of the data set. It was, however, sufficient for
amplitude extraction.

A second seismic horizon interpretation was undertaken for use with the inversion analysis. During this interpretation
a “point and click” method was utilized which bracketed the pay zones. This interpretation is only intended to bracket the
sand bodies, and not to interpret the boundaries.

As can be seen from the seismic line in figure 2, in the upper Green zone all 3 logs have acoustic impedance lows,
corresponding to high amplitude seismic events. However, in the lower (Blue) zone, well 3ST1 does not have an acoustic
impedance low or significant pay sand, yet the seismic still indicates a relatively strong amplitude event.

This problem can more clearly be seen on seismic amplitude maps, shown in Figures 3 and 4, for the Green and Blue
zones respectively. The seismic amplitudes have been extracted from the “tracked” trough in each reservoir zone. For
both the Green and Blue zones, the amplitude extractions in Figures 3 and 4 show a “patchy” appearance. Areas north
and south of the main fault zone appear equally as prospective.

RESERVOIR CHARACTERIZATION USING ACOUSTIC IMPEDANCES

In order to address the situation where the seismic amplitude attributes are giving false indications of pay zones, a
lithology method based on acoustic impedance is applied. The method requires a multidisciplinary approach which
focuses on integrating well log, horizon interpretation and seismic information. The workflow for such a
multidisciplinary, quantitative reservoir analysis includes the following steps:

• Seismic interpretation of main zones of interest;

Gulf of Mexico 3D Case History November 16, 1996 Page 5


• Seismic well calibration;
• 3D seismic lithology inversion conditioned by the well data and the initial seismic interpretation bracketing the
sands, in order to create a lithology cube;
• Optionally, creation of 3D geologic model by geologically interpolating acoustic impedance well log data in each
geologic unit. The low frequency component of this model is often a viable alternative for the low frequency
component generated by the lithology inversion procedure.
• Revision and detailing of the interpretation of reservoir stratigraphic units by using the lithology cube as the mapping
medium, instead of the seismic data, and in conjunction with well and geologic control;
• Extraction of reservoir properties such as porosity and net pay from the interpreted lithology cube.

The initial seismic interpretation does not require any detailed interpretation at the reservoir level, only the
interpretation of main zones of interest. Detailed interpretation is done on the 3D lithology cube, which is more efficient
and more accurate than interpreting reservoir level detail from the seismic data. In this way, better results are achieved
due to the ease of interpreting on geologically significant lithological layers. This procedure allows for a more accurate
final interpretation with reduced “cycle” time.

The key reason for the success of this approach is that lithology data represents a layer property and not a seismic
reflection interface property. Also, lithology cubes can be generated with broader bandwidth than the seismic data, which
is particularly helpful in addressing tuning problems.

The inversion step is based on a constrained sparse spike inversion technique (Debeye and VanRiel, 1990). The
constraints are set using well control and guided by the initial interpretation. This method reconstructs the low frequency
information missing in the seismic data. Often the lowest portion (typically 0 - 3 Hertz) of the low frequency component
can be more reliably obtained from a geologic model. In this case the lowest frequency band of the inversion result is
replaced by that of the model.

When creating the geologic model, the horizon interfaces, together with any interpreted faults, are considered the
boundaries for the internal layering when creating the geologic model. Based on a user-defined description of the
stratigraphic relationships within the layer, the acoustic impedance well log information is interpolated within the three
dimensional volume. The very low frequency component of this geologic acoustic impedance model may then be merged
with the results from the constrained sparse spike inversion analysis.

The end result of the modeling and inversion is an acoustic impedance lithology cube with broader bandwidth than
the original seismic data. Introduction of the additional bandwidth reduces tuning. Additionally, effort is directed at
constructing a full lithology cube, including information surrounding the reservoirs, so that variations in the rocks
surrounding the reservoirs are also accounted for.

In many cases a petrophysical analysis of the available wells can establish relationships linking acoustic impedance
variations to changes in porosity, sand/shale ratio, interval velocity, etc. Using these relationships, average porosity and
net pay maps can be generated for each of the reservoir units. All acoustic impedance samples within the reservoir layers
are used in these calculations. As a result, these reservoir property maps are typically more reliable than those derived
from seismic attributes, because all of the available information is utilized.

ANALYSIS USED IN CASE STUDY

The seismic to well calibration step was completed as an iterative process which combines log editing and
deterministic wavelet estimation around the wells. The wavelets at each well were found to be very similar, close to zero
phase, and the data bandwidth was determined to be 6-50 Hertz.

The constrained sparse spike inversion method was utilized in order to invert the seismic. The main constraints were
set on horizons bracketing the reservoir. Within the reservoir zone, the constraints were left open to capture the expected

Gulf of Mexico 3D Case History November 16, 1996 Page 6


lithologic variations. The output of constrained sparse spike inversion was conditioned by replacing the very low
frequencies (0-3 Hertz) of the inversion result by the low frequencies of the acoustic impedance geologic model derived
by interpolation of well control.

RESULTS

The acoustic impedance transect in figure 5 has been derived utilizing the method described above. It corresponds to
the same arbitrary line as the seismic section in figure 2. Several features are of interest.

Figure 5: Arbitrary acoustic impedance transect through wells 3, 3ST1, 5, and 4. This transect is along the
same profile as in figure 2. Horizon interpretation and impedance log curves are overlain. Note
the good tie of the seismic derived acoustic impedances with the well control.

At the level of the Green sands, the large seismic amplitudes correspond to zones of low acoustic impedance, i.e.
reservoir sands. However, when viewed in detail, there are clear differences: The low acoustic impedance log curves
closely match the low values of acoustic impedances for figure 5. On the seismic line, figure 2, the pay zones are offset
from the amplitudes, due to seismic reflection alignment with bed boundaries, and possibly due to tuning. The acoustic
impedances form a more accurate basis for reservoir interpretation.

As stated previously, acoustic impedance is a layer property. Therefore, compared with seismic amplitudes, it
becomes very easy to interpret the reservoir units. Cut-off values for acoustic impedances versus, for example, net pay
may be determined for each sand interval distinguishing the reservoir from the surrounding shale. Auto-tracking this cut-
off value is easy. Also note that by using a cut-off value the lithology data clearly shows where reservoir units terminate.

In the Blue layer, there is no development of significant low acoustic impedances around well 3ST1. This is most
evident on figure 5. However, the amplitude analysis alone did show an anomalous amplitude at well 3ST1 (figures 2 and
4) which may have caused the well to be deepened in that zone. An inversion analysis prior to drilling could have
changed the well plan for well 3ST1. Further analyzing figure 5 shows the cause of the anomalous seismic amplitude to
be a change in the background acoustic impedance of the rocks surrounding the Blue sands.

Gulf of Mexico 3D Case History November 16, 1996 Page 7


Restricting a volume view to a particular geologic layer and using transparency can highlight the reservoir quite
effectively. Figure 6 shows a rendering of the impedances corresponding to sand for the Green interval.

Figure 6: The acoustic impedance volume shown in figure 5 is loaded into the 3D
viewer. The viewing angle is set to be from the northeast. The volume is
vertically restricted to the green sand event, and then the impedance values
higher than those corresponding to sand are made transparent, revealing
details of the reservoir continuity and geometry.

The analysis of the results can be further expanded by comparing reservoir property maps derived from the acoustic
impedances to seismic amplitude maps. Figures 7 and 8 show net pay maps for the Green and Blue zone respectively,
which have been derived using curves relating acoustic impedance to net pay. The yellow corresponds to higher net pay.
These yellow areas should be compared to the corresponding seismic amplitude maps in Figures 3 and 4. The following
important differences are seen:

• The most obvious difference between the net pay map for the Green zone and the seismic amplitude map for the
same zone is that the net pay map is showing lineations due to the depositional fabric. The amplitude map is
haphazard and appears “freckled”.
• Seismic amplitudes are overestimating the prospective areas due, most likely, to tuning. It is difficult to distinguish
the areas with the greatest potential from the seismic amplitude maps. However, the net pay maps clearly
differentiates potential. The area south of the main fault is more attractive than the area to the north.

The net pay map for the Blue zone clearly shows large areas where the reservoir sands are poorly developed. The seismic
amplitude map, however, indicates a significantly larger area characterized by relatively high seismic amplitude. As
discussed above, well control verifies the accuracy of the net pay maps derived from acoustic impedances. These maps
can be used to prevent drilling to unproductive deeper Blue zone targets.

Gulf of Mexico 3D Case History November 16, 1996 Page 8


Figure 7: Net pay map for the Green sands. High net pay is yellow.

Figure 8: Net pay map for the Blue sands. High net pay is yellow.

CONCLUSIONS

Gulf of Mexico 3D Case History November 16, 1996 Page 9


Seismic horizon interpretation, well log detail, internal layer stratigraphic complexity, and petrophysical analysis
have been quantitatively combined to create a more thorough understanding of a Gulf of Mexico 3D oil field.

In the case of seismically thin layering and/or lateral variations in the rocks encompassing reservoir units, the
standard approach of seismic interpretation involving reflection tracking and seismic amplitude attribute analysis breaks
down. Due to thin layer tuning, layer thickness measured from the seismic no longer represents true layer thickness.
Additionally, the variations in seismic amplitude due to lithology changes in the reservoir units are also distorted due to
tuning effects.

The drawbacks of the standard approach to seismic interpretation can often be overcome by working with integrated
acoustic impedance data rather than seismic data. The key reason for this is that acoustic impedances are a layer property,
whereas seismic data are a layer interface property.

A highly effective approach to generating acoustic impedance volumes from seismic data involves 3D constrained
sparse spike inversion and 3D geologic modeling. In this approach the geologic model is used to stabilize the very low
frequency component of the inversion result. When well log data is available, a petrophysical analysis can be used to
establish quantitative relationships between acoustic impedance and other lithologic parameters such as porosity, fluid fill
and interval velocity. Through these relationships it then becomes possible to directly calculate such reservoir parameters
as average porosity and net pay. As illustrated in the case history example, maps of these acoustic impedance based
properties are clearly more reliable than maps based on seismic amplitudes.

Using the proposed integrated methodology leads to a significantly more accurate reservoir description than possible
using conventional seismic amplitude based methods. This superior description can be used to support the optimization
of the development drilling program.

REFERENCES

Brown, A.R., 1986, Interpretation of three-dimensional seismic data: The American Society of Petroleum Geologists.

Badgett, K.L., Hill, P.L., Mills, W.H., Mitchell, S.P., Vinson, G.S. III, Wilkins, K. L., 1994, Team Combines
Technologies to Target Horizontal Wells in Gulf of Mexico Oil Field: Oil & Gas Journal, v. 92, no. 11, p. 44-49.

Debeye, H.W.J., and Van Riel, P., 1990, Lp-norm Deconvolution: Geophysical Prospecting, v. 38, p. 381-403.

Lindsay, J.F., Prior, D.B., and Coleman, J. M., 1984, Distributary-Mouth Bar Development and the Role of Submarine
Landslides in Delta Growth, South Pass, Mississippi Delta: AAPG Bulletin, v.68, no. 11, p. 1732-1743.

Mayall, M. J. , Yielding, C. A., Oldroyd, J. D., Pulham, A. J., and Sakuri, S., 1992, Facies in a Shelf Edge Delta - An
Example from the Subsurface of the Gulf of Mexico, Middle Pliocene, Mississippi Canyon, Block 109: AAPG Bulletin,
v. 76, no. 4, p. 435-448.

Gulf of Mexico 3D Case History November 16, 1996 Page 10

You might also like