One possible hierarchical relationship between various attribute categories.
The term "seismic attribute" has wide connotations within the petroleum industry and every geophysicist seems to have a different definition. In reality, all geophysicists working with seismic data, work with seismic attributes all of the time. The Oxford English Dictionary defines an attribute as a "quality ascribed to any person or thing."
The authors' definition of seismic attributes is equally broad: all information obtained from seismic data, either by direct measurement or by logical or experience-based reasoning.
Though the seismic technology employed today for the exploitation of hydrocarbon resources is complex, the basic work flow has remained unchanged since the earliest days of single-channel analogue recording. Let's review this work flow, and attempt to place seismic attribute methods within this general process.
- Stage 1 - Data acquisition: This is the 'experiment.' We excite the earth, using an energy source on or close to the surface, and measure the acoustic returns from sub-surface features using detectors.
- Stage 2 - Data reduction: The raw data acquired from the field needs to be "processed" before extracting meaningful information about stratigraphy, lithology, fluid saturation, and a host of other properties of use to engineers. These steps increase signal-to-noise, and position ("image") data within a 3D framework.
- Stage 3 - Data interpretation: This step transforms our processed seismic data to geological and engineering properties.
The attribute we most often analyze is seismic amplitude. Our processing steps in Stage 2 build a measurement of the "zero-offset P-wave reflection coefficient," an attribute commonly called "seismic amplitude." However, there are many other measurements or attributes we can extract from our seismic dataset. Some of these attributes may be extracted as part of our data-processing steps (RMS-velocity and horizon-time, for example). Others are derived later, in Stage 3. Of particular interest to geophysicists are attributes which can be closely related to rock and fluid properties using the science of 'rock-physics.' These attributes include:
- AVO - how seismic amplitude changes with respect to source-receiver offset
- Impedance - derived from our seismic amplitude information
- Attenuation ('Q') - how frequency attenuation varies within our seismic data-set
- Velocity - derived from our various imaging algorithms.
There are other classes of attributes which are important for interpretation, though they cannot be so easily related to rock and fluid properties. These attributes include:
- Instantaneous phase and frequency (derived via the Hilbert transform) - these can provide the interpreter with valuable information about stratigraphy.
- Coherency type attributes (semblance, edge-detection etc.) - these algorithms measure how seismic properties vary spatially within our seismic data-set.
A geophysicist today has a bewildering array of seismic attributes available to aid in the search for oil and gas. We have developed a simple set of categories within which all current attribute methods may be placed:
- Instantaneous - attributes computed sample by sample. Examples include complex trace attributes, such as instantaneous frequency and phase.
- Wavelet - attributes computed at the peak of the seismic trace envelope. Examples include mean-frequency and apparent polarity.
- Physical - attributes which respond more or less directly to physical properties: trace envelope, which responds to acoustic impe-dance contrasts, and average velocity.
- Geometrical - attributes that measure how physical attributes vary as a function of space. Examples include coherency and dip-curvature.
- Reflective - attributes that correspond to interface properties, such as AVO measurements.
- Transmissive - attributes relating to the physical properties between interfaces. These include NMO velocities, attenuation ('Q').
The key to seismic reservoir characterization is finding a reservoir property model which honors both seismic and log data.
So, how do we use all of these interesting seismic attributes to help us find more oil and gas? First answer is simple: we don't use them all. We need only employ those which are specific to the task in hand. A second and related consideration is a more practical one. How can we interpret not one seismic attribute volume, but many which might be developed for a given survey or prospect? To answer these questions, we need to introduce the concepts of classification and calibration.
Classification is a statistical process which involves combining several attribute data-sets together to yield a new attribute, based upon the discriminating features of the individual attributes. Such statistical methods are data-driven, and offer a rapid method for detecting anomalies or objects within seismic data-sets. Many such methods are available. Most use one or more forms of neural network or pattern recognition algorithms to locate and separate seismic facies based on attribute response.
Calibration is a more rigorous process, and requires us to introduce external data-sets to quantify our seismic attribute measurements. Our seismic method yields only relative values. To calibrate our attributes to represent rock and fluid properties, we need to integrate our seismic data with attributes derived from borehole information. This process is generally known as "seismic reservoir characterization."
There is a close analogy between seismic attributes and wireline logs. Seismic attributes are mathematical tools we use to measure various seismic properties. Each seismic attribute responds separately to differences in subsurface geology. Similarly, we employ various electrical, acoustic, and magnetic borehole logging tools. Each tool responds differently to various properties such as lithology and saturation.
Interpretation work flows
Concurrent with the development of attribute technology, we have seen the rapid development of 3D visual interpretation systems. Such software tools are essential elements in any modern interpretation method. It simply isn't possible to efficiently analyze and interpret multiple seismic attribute data-sets within conventional "wiggle-based" methods.
The marriage of advanced attribute technology with sophisticated 3D visual interpretation systems offers still further opportunities for improvements in speed and accuracy. Using attribute technology for seismic reservoir characterization, we can convert our basic seismic amplitude data, which represents acoustic properties, to new "attributes" which are closely related to engineering properties, such as fluid saturation and porosity. Interpreters working with seismic-attribute derived geologic volumes can more rapidly locate and refine prospects.
Seismic attributes are essential tools within any seismic reservoir characterization method. What else do we need? Seismic reservoir characterization requires us to calibrate our seismic data to some known property value. We obtain these known values from careful analysis of borehole information.
Rock physics holds the key. It allows us to integrate seismic attributes with measurements made using wireline logging tools. Indeed, an excellent definition of seismic reservoir characterization is "building a rock and fluid model consistent with both seismic data and log data."
Seismic attributes have very much 'come of age.' They have evolved from being interesting mathematical transforms to being essential elements in our approach to deriving geologic and engineering properties from seismic data.
Dr. Tury Taner is Chief Geophysicist and Dr. Joel Walls is Chief Petrophysicist with Rock Solid Images.