Seismic Attributes – some recollections

Early attributes

Few geophysicists can recall the days when a seismic record meant an 8 inch wide, 4 foot long banner of paper carrying 24 seismic traces. Each trace of each seismic event was carefully picked and the time of the reflection inked by hand. Even at that early time seismic attributes formed a useful adjunct to interpretation. Perhaps the most elementary was to grade the quality of each reflection. A label of G, F, or P, was inked along with the reflection time, later to be represented with a solid, light, or dashed line on a hand plotted cross section. In some cases the dreaded NR was used to signify recordings so poor that key reflections could not be distinguished from the unruly noise saturating the recording.

Demonstrating the true doodlebug spirit of turning adversity into advantage, Ben Rummerfield (1954) correctly predicted in one area that the NR gaps corresponded to faulting. This predecessor of semblance was perhaps the first documented use of seismic attributes to find oil.

Another common attribute still very much in use today, was reflection character. Experienced geophysicists learned to identify key reflections by the often difficult to describe minute variations in the shape of a reflection. Some could recognize a favorite event from a distance, much in the manner of a mother penguin identifying her chick from thousands of seemingly identical specimens in a rookery. The same geophysicists also often recognized which subtle changes in reflection character were often associated with a potential reservoir.

Analog recording of seismic data

Analog recording brought the ability to remove normal move-out and to display continuous reflection sections. Although not properly an attribute, data display often can be m o re revealing of anomalies than are many attributes. Continuous section displays certainly made it easier for the interpreter to observe continuity of character and to detect changes thereof. Perhaps more importantly, the scale change resulting from compressing several tens of recordings into a single continuous cross section revealed many subtle changes, such as pinchouts and low angle unconformities, not readily seen before.

In a forerunner of some modern attributes, Frank Reiber developed a camera that recorded only that portion of the signal within a narrow slot, of width equal to trace spacing on the display. The slot straddled the zero crossing line of the trace. On the output, a very high amplitude reflection appeared as a pair of almost parallel lines. The slope of the lines flattened as reflection amplitude or reflection frequency decreased. The slope of the wave form at the zero crossing was meant be an indicator of amplitude. Unfortunately, most of the other attributes were lost in the process and, in common with the modern calculated attribute of wave slope, it was not uniquely descriptive of amplitude or frequency since both determine wave slope.

Amplitude is a trusted and easily perceived attribute because it is a direct function of the contrast in the property of rock velocity (and density), at seismic reflection boundaries. Amplitude and reflection time are the only quantifiable single trace attributes specifically related to geology, but in spite of all the many advances made by reproducible recording and display, amplitude could be measured only in a relative sense.

The tremendous dynamic range of amplitudes occurring over the few seconds of seismic signal life unfortunately far exceeded the ability of any recording medium of the time to contain it. Therefore some reacting feedback circuits continually pummeled and squeezed the incoming signal in order to cram it in to the confines of the recorder. United Geophysical Company, among others, tried to preprogram the gain control, but such things as hole to hole difference in shot impulse often still required several repeat shots with the gain modified each time to fit. Shot holes refused to cooperate, since response characteristics changed with each shot or the hole simply collapsed.

Digital recording of seismic data

All these problems disappeared with the advent of digital recording and its ability to measure the absolute amplitude of each seismic signal sample. Often, knowing the absolute amplitude was not as important as simply defining the normal amplitude of a given reflection. On plotting a display section the average amplitude could be color scaled to some low level gray, such that any appreciable increase in amplitude of the reflection would stand out as a distinct dark line. The display procedure was termed a Bright Spot section, and in some areas was an immediate success. A sudden increase in the precisely measured reflection amplitude along a boundary often provided a direct indication of gas in the section. From this grew the science of DHI, Dire c t Hydrocarbon Indicators, comprising those attributes that can detect pore space filled with hydrocarbons rather than water.

Complex trace analysis

Digital recording and digital computers provided the right combination of power to measure other signal properties, such as frequency, and phase. Some of the leaders of the time, notably Turhan Taner, showed further that the complex signal analysis techniques, such as provided by the Fourier transform, yielded a plethora of precisely measured statistical properties of seismic traces.

Other workers measured almost every conceivable geometrical and physical property of the trace and of groups of traces, often without any reference to how the attribute related to geology. The number and variety of attributes proliferated wildly as developers discovered one more way of torturing the data to extract still another elusive factor.

Conrad Schlumberger, many years before, recorded geophysical properties of the geologic column by drawing an instrument up a borehole to measure the geologic property sensed by the tool. Resistivity, the first tool, measured low values in the presence of water, (usually salty in the subsurface), and high in the presence of hydrocarbons. This marvelous device was the Bright Spot of its time; a direct hydrocarbon indicator. Other tools such as Self Potential, to measure electrolyte geochemistry, soon followed. Later, Radioactivity, Sonic Velocity and Density tools followed, to name a few. It was soon found that the combined response of several tools could be factored to provide a complete and fairly reliable description of lithology, porosity and fluid content.

Some eager but perhaps less experienced processors sought to use combinations of attributes to emulate the results of well log analysis, but in their rush may have failed to recognize that each well logging tool records a different property of the geologic section while seismic trace attributes all merely measure different properties of the same seismic trace; there is no new geologic information in any attribute derived from it.

Invariably, attributes of a single seismic trace provide little information useful to the explorationist. It is only when some other measured variable, usually distance, is added, that changes in the attribute with location become significant, and the attributes become useful. Even so, rarely do seismic attributes inform the user what is happening to the geology, only that something is happening.

More attributes

Notable exceptions are AVO, which brings geometry into the analysis for gas, and is backed by solid mathematics. Shear wave traces add another variable that can be measured and related to geology. Velocity may require depth information but is quantitative and closely related to geology. Seismic Trace Inversion yields acoustic impedance, which often is closely related to a sonic log but without some method of replacing the lost in recording very low frequencies that build the characteristic shape of a sonic log, simple inversion of a seismic trace may lead to false conclusions.

Semblance, and its many family members, is best applied to a 3D seismic cube, but can be a marvelous aid to mapping faults and geometric anomalies buried in the data. A semblance cube can be sliced in almost any direction and still provide useful information. On the other hand slices through many attribute cubes where structure is present are useless for most practical purposes, unless the slice is directed to follow a key horizon.

In the end, during the early furor of discovery, many attributes were tested but few were chosen. Those that could be quantified, that could be related directly to geology, and were widely applicable to most situations soon found a place in the interpreter’s tool kit. Many others were found to be limited in scope, difficult to relate to geology, or less effective than others, or did not meet the crucial test of cost/benefit. Some attribute displays were found to provide little more than different color band displays of the same phenomenon.

Use and abuse of attributes

There is the matter of misuse of attributes. Modern workstations provide almost limitless flexibility in scaling and coloring displays. More that one dry hole has probably been drilled on a bright spot produced by cranking up the amplitude scaling of some weak amplitude anomaly to present evidence more favorable that justified by the data alone. Even the hoary old problem of amplitude recovery today still depends on the skill of the processor. Often applied for a final display is the AGC routine, which immediately takes the data back to analog days.

Many attributes, such as those that require a window of samples, are too coarse or too vague in their definition to be useful for small features, while the response of larger anomalies can often be recognized directly. In spite of the many advantages of 3D seismic surveys, the distortions of phase and frequency produced by smashing together a number signals arriving from several different directions and sometimes from anything but a common reflecting point disagree with some of the basic attribute assumptions.

For these and many other reasons, interest in the use of the full panoply of attributes waxed and waned, with only a limited number in regular use. Researchers and developers quietly moved on to new fields of investigation and the more aggressive sales people began the search around for the next wonder tool.

Encouragingly, interest has recently revived somewhat. Point source and single point recording may help improve phase, amplitude and frequency response. For some attributes, although it may be difficult to explain the relationship between geology and the attribute response, enough empirical evidence will have accumulated to show that a relationship indeed does exist. My money would go towards increasing the bandwidth of seismic signals as the elixir that would go farthest toward solving the greatest number of technical problems facing seismic practitioners. However, if some forecasters are correct, about the time everything there is to know about attributes is known, we will have run out of oil to find.