A measurement derived from seismic data, usually based on measurements of time, amplitude, frequency, and/or attenuation. Generally, time-based measurements relate to structure, amplitude-based ones to stratigraphy and reservoir characterization, and frequency-based ones (while often not clearly understood) to stratigraphy and reservoir characterization. Attenuation measurements are usually very uncertain. Measurements are usually based on stacked or migrated data, but prestacked data are used in determining stacking velocity (q.v.), AVO (amplitude variation with offset, q.v.), and other attributes. Because there are many ways to arrange data, attributes constitute an open set, and because they are based on so few types of measurements, attributes are generally not independent. Attributes are useful to the extent that they correlate with some physical property of interest. The primary usefulness of attributes is that they sometimes help one to see features, relationships, and patterns that otherwise might not be noticed.
Seismic measurements usually involve appreciable uncertainty and do not relate directly to any single geologic property. With so many geologic variables, correlation with a particular property in one situation is apt to not hold in another situation. Attributes generally respond to a variety of geologic situations and a geologic change may mean a change in the correlation. The problem is determining the limits to an observed correlation, especially when we do not understand the underlying physics—How wide ranging is a correlation valid? During a Direct Detection Symposium in 1973, Miller Quarles presented numerous processing schemes to enhance hydrocarbon signatures; in response to a question about the ‘‘scientific basis of all these attributes,’’ he responded, ‘‘We don’t know yet, but remember, [we] invented them.’’ Unfortunately we still do not understand how to relate most seismic attributes to geologic causes and situations.
Among the ways we calculate attributes are smoothing and averaging over windows of various sizes, finding residuals, peak values, measuring the distribution within a window (mean, median, kurtosis, percent greater/smaller than a threshold, sums, residuals, scatter, etc.), continuity, edges, smoothness, linearity or curvature, gradients or other derivatives, absolute values, polarity changes (zero-crossings), peak-trough differences, etc. Relations may be measured over windows (spectra, correlation, semblance, covariance), etc.
Attributes can be measured along a single trace or throughout a volume or in other ways. The first attributes identified as such were the 1D complex-trace attributes of envelope amplitude, instantaneous phase, instantaneous frequency, and apparent polarity (see complex-trace analysis) and acoustic impedance (or velocity) determined by inversion (q.v.). Attributes may be measured along a defined (picked) surface (horizon attributes) such as amplitude extraction, dip magnitude, dip azimuth, artificial illumination, and coherence (q.v.). Hydrocarbon indicators (q.v.) are attributes. Attributes can be combined to make new attributes. Transformations of attributes are sometimes given physical-property names (porosity, fluid saturation, lithology, stratigraphic or structural discontinuity, etc.), usually based on local crossplots or local correlations with borehole-log or other measurements; they may be reasonable approximations locally but they are apt to give erroneous values under different circumstances. See Figures A-23 and A-24, Brown (1999, chap. 8), and Chen and Sidney (1997).