Seismic interpretation - book
|Series||Geophysical References Series|
|Title||Digital Imaging and Deconvolution: The ABCs of Seismic Exploration and Processing|
|Author||Enders A. Robinson and Sven Treitel|
|Store||SEG Online Store|
Geologists and geophysicists interpret seismic images, in the form of maps and cross sections, so they can choose the most favorable drilling sites for new oil wells — either wildcats or field-extension wells. The word interpretation is subject to many meanings. In the final analysis, however, interpretation involves a person’s exercise of judgment on the basis of geologic and geophysical criteria.
In the recent past, the seismic interpreter has decided not to be satisfied with just the images produced in the seismic-processing stage. The interpreter now has access to software that he can use to enhance the pictures provided in the seismic-processing stage.
In this way, the interpreter can produce maps and cross sections that emphasize specific aspects of the geologic features in question. Sheriff and Geldart (1995, Chapter 10) give an excellent treatment of geologic interpretation. Here, we outline only some of the high points.
In a general sense, interpretive processing falls under the heading of image enhancement. The difference between interpretive processing and seismic processing can be summed up as follows: Seismic processing relies greatly on use of the wave equation and all the related methods of dealing with traveling waves. Interpretive processing, although it takes into account its debt to the wave equation, uses a plethora of nonwave-equation techniques. Interpretive processing relies heavily on various deterministic and statistical image-processing methods, with certain seismic features factored in. Interpretive-type software converts the image of the geology provided by seismic processing into an enhanced image that brings out various critical aspects that are valuable to the interpreter. Such enhanced geologic images, which can be displayed on an interpretive visualization computer system, are used for final evaluation of drilling sites for new oil wells as well as for reservoir characterization.
What is the layer-cake model? Reflected seismic energy, as measured at the surface of the earth, makes up the observed reflection seismogram and represents the known information. Geophysical models of the earth can be simple or complex. The simplest is the layer-cake model, which assumes that the earth consists approximately of horizontal, homogeneous, and isotropic layers within each of which the velocity varies smoothly (Figure 1).
What is an event? Look at a crowd of people walking down the street. Their strides have similarities, but the people are not in lockstep. Suppose, however, that small group is marching in step. They stand out. Now look at a record of raw seismic traces. Generally, one trace is somewhat similar to the next. However, suppose that in at a given instant of time (i.e., the arrival time), the traces abruptly line up (i.e., become coherent) so that the peak (or valley) of one trace becomes quite similar in shape and character to the peak (or valley) of the next trace. Such a lineup of traces in lockstep is called an event. An event indicates the arrival of new seismic energy. Generally, the lineup is along a slanting line, the slope of which indicates the stepout time of the arrival of energy. An event can reveal a reflection, refraction, diffraction, or other type of wavefront arriving at a sensor.
The distinguishing features of various types of events are well documented, but by far the most important type of event is a primary reflection. A primary-reflection event is a lineup of traces that indicates the arrival of energy that has traveled a direct path from the surface to a reflecting interface and then back to the surface. An interpreter picks or selects an event on a seismic record — for example, a reflection event. A pick is an event on the trace, along with its time of occurrence (i.e., the arrival time).
What are multiple events? For every primary event, there are many multiple events. A multiple event is one that makes one or more round-trip paths within the subsurface sedimentary layers before it returns to the surface of the earth. A primary path and a multiple path each can produce reflected energy that appears at the same arrival time on a seismogram. A great disadvantage is that a multiple event can appear on a seismic trace at a location at which no primary event exists. If we see such an event and do no further analysis, we could mistake it for a primary event. More generally, many such multiples would interfere with the primaries, masking them and making it impossible to delineate them. Thus, multiples represent a serious kind of signal-generated noise in the interpretation of seismic events.
How are multiples handled? Current seismic processing practices use two main ways to reduce the disadvantageous effects of multiple events. These two processing methods are stacking and deconvolution. Stacking works in the space domain, whereas deconvolution works in the time domain. Stacking is a method of averaging traces over space that reinforces the primaries and cancels the multiples. Deconvolution is a method of averaging traces over time, with a similar goal. Ideally, stacking, deconvolution, and migration should be carried out simultaneously as one overall operation in seismic data processing. For practical reasons, however, these processing operations are carried out separately in such a manner that the overall seismic-data-processing package is robust and stable.
What is a horizon? A horizon is a surface (i.e., an interface) separating two layers of rock. Where such a surface (even though it might not be identified itself) is associated with a reflection that can be followed over a larger area, a map based on such a reflection event is called a horizon map.
How are 3D data displayed? The two horizontal distance coordinates and the vertical coordinate (which can represent either time or depth) define the 3D volume. A 3D volume can be sliced in various ways. For example, the data can be sliced along three orthogonal planes. The vertical plane through a seismic line produces an inline vertical section, the vertical plane through a seismic crossline produces a crossline vertical section, and the horizontal plane for a given vertical coordinate produces a horizontal section, which is either a time slice or a depth slice, as the case may be.
A horizontal slice is made up of the 2D seismic data lying on a flat level plane that cuts through a volume of 3D seismic data (Figure 2). If the vertical component is time, the horizontal slice is a time slice. In other words, a time slice is a display of the seismic amplitudes for a single time arrival. If the vertical component is depth, then the horizontal slice is a depth slice. In other words, a depth slice is a display of the seismic amplitudes for a single depth. Picking a reflection on a time slice is equivalent to mapping a time contour, and successive contours can be mapped directly by following the same event on successive time slices. This result is a structural contour map.
If a fault is picked on a sequence of time slices, a fault-surface map results. The contours generally reveal fine details of the fault. Another way of examining faults is by the coherency cube, which is discussed later in this chapter.
How else can the data be sliced? The 3D volume also can be sliced vertically along diagonal lines connecting wells or in the plane of a deviated well or in zigzag lines to tie two or more existing wells (these are arbitrary lines).
What is a horizon slice? A horizon slice is a display from a 3D data set of the data elements that all lie on the same reflecting horizon. Such a display brings out variations in amplitude (or some other attribute) on the horizon in question. A horizon slice is different from a horizontal slice (or time slice). As Sheriff and Geldart (1995), point out, horizon slices are useful in the study of stratigraphy and reservoir properties.
What are fault slices? Fault slices are made by slicing through a 3D volume parallel to a fault surface but displaced a small distance (say, 25 to 50 m) into the downthrown and upthrown fault blocks. This displacement avoids the distortions often present very near the fault itself. Let one fault slice be made on the upthrown block and another be made on the downthrown block. The differences observed between these two fault slices indicate variations that occur along the fault surface.
|Previous section||Next section|
|The seismic reflection method||Porosity|
|Previous chapter||Next chapter|
Also in this chapter
- The seismic reflection method
- Absorption loss and transmission loss
- The wave equation
- Wave velocity
- Velocity analysis
- Seismic tomography
- Appendix C: Exercises