Why Spectral Decomposition

Why Spectral Decomposition Since the development of the coherence cube (Bahorich and Farmer, 1995), seismic attributes
have played an important role in detecting faults and fractures in the earths crust. Other coherence algorithms (e.g. Marfurt et al., 1998; Gersztenkorn and Marfurt, 1999) have been applied to improve the quality of images of faults and fractures. Curvature attributes (Roberts, 2001) are widely now used to enhance the resolution of faults and fractures. However, all these investigations are based on full or dominant-spectrum seismic data. Spectral decomposition (Partyka et al., 1999) provides a novel means of utilizing seismic data, and a method called time-frequency 4-D cube was developed to extract seismic information from seismic data based on the different geologic scales in different frequency bands. Using these methods, discrete frequency coherence cubes (Sun et al., 2010) can be applied to detect faults and fracture zones that cannot be easily detected using the full spectrum coherence data. Application of Spectral Decomposition and Curvature Faults and fractures occur on many scales in the Earth. Therefore, we can detect faults and fractures found at different scales using different frequency bands. Spectral analysis shows that the bandwidth of seismic data of the study area is about 10-55 Hz. Therefore, 10 discrete frequency bands are calculated by increasing frequency sampling by 5 Hz.
Spectral decomposition is a methodology to reveal incised channels on large fans that may incise slope and basin toe in deepwater. This may compare favorably against the standard methods of amplitude, phase and frequency extraction and coherency attributes as it displays the channel image changes over desired windows of the frequency spectrum. Spectral Decomposition may also reveal valuable insight into the depositional history of a region. The spectral decomposition attribute transforms data recorded in space-time to spectral domains, for example using discrete Fourier transformation, and is useful in quantifying seismic waveform variations and thickness within reservoir zones of interests (Partyka and Gridley, 1997). It may often be the case that standard windowed attributes extracted are tuned to the response frequency of the dominant wavelet, which may result in subtle features being masked by this response. Once the spectral decomposition is performed, unsuspected features are revealed that may contribute further insight and understanding with regards to reservoir mapping,delineation of stratigraphic and structural features such as channel sands and complex fault systems.
Sweetness attribute is obtained from reflection intensity and instantaneous frequency attributes, and is helpful in delineating subtle discontinuities such as pinch-outs and channels in stratigraphic traps. It can also be useful for distinguishing shales from sands and will be used in conjunction with spectral decomposition. It attempts to capture the ratio between amplitude and the square root of frequency. Sands tend to be associated with high amplitude and lower frequencies, hence co-blending Sweetness with Variance could give a good control of the extension of stratigraphic features, such as channels and pinchouts.
Spectral decomposition
The Spectral Decomposition method replaces the single input trace with a gather of traces corresponding to the spectral decomposition of the input attribute. The input to spectral decomposition is a seismic volume. The output is several volumes, each one representing a different frequency band. Creating spectral decomposition attributes enables you to illuminate the structures with different frequency bands to see if any of them gives you better resolution.
When doing interpretation it is a common practice to look at instantaneous attributes such as envelope or phase volumes. Spectral decomposition gives you several such volumes. At a specific frequency band certain size structures are more visible due to tuning effects, etc. This means that for example instead of looking at one phase volume for a cube of stacked data, you are able to view several of them and see if any single one shows a structure better.
Spectral decomposition is a form of wavelet transform. The wavelets are short temporal band limited sine and cosine wavelets. In the Frequency Increment mode Gabor wavelets with equally spaced central frequencies are designed and applied to the data, whereas in the octave increment mode, Gabor- Morlet filters are designed with equally spaced central frequencies in octave-frequency domain, hence their representation in thefrequency axis will be unequally spaced.
In spectral decomposition you use complex traces.Basically a single trace is convolved with the first real wavelet to get the real trace for the first frequency, and the input trace is convolved with the first wavelet of the imaginary part to get the imaginary trace. Then a complex trace attribute is constructed, such as envelope, phase, etc. This step is repeated for each filtered wavelet operating on the same trace and obtain band limited traces. (This is very similar to filtering the input trace with several band limited filters and creating a new trace from each filter.) In spectral decomposition the input is a single trace and the output is a series of traces each with a different frequency band. The differences between the FFT type of attributes and these is that FFT attributes give a representation of the wavelet, meaning that when you do FFT of an input trace, you get the spectrum of the source wavelet (assumption is that the reflection coefficients have white spectrum). When you take the FFT of a time limited area, because you did not include enough reflections (short in time) to make the white spectrum assumption, you have the spectrum that is a representation of the reflections coefficients. Also, because you have a limited time window, your resolution in the frequency domain is limited (the shorter the window in time, the wider the frequency band). That is why we refer to frequency bands, rather than frequencies
Geometrical Attributes Spatial discontinuities, such as faults and channels systems,may best be highlighted by attributes such as coherence or variance (Bahorich and Farmer, 1995; Marfurt et al., 1998; Pedersen et al., 2002. Methodologies and workflows now exist to enable the extraction of superior and higher fidelity discontinuity information from these attributes, or from the original seismic data.During the exploration phase, the interpretation of the seismic data may focus on regional or major tectonic elements or fault systems, or alternately at appraisal and development stage, may focus on fault systems for understanding reservoir geometry and connectivity, which may ultimately affect hydrocarbon recovery. Ant Tracking is a proprietary tool for extracting fault information at any scale from seismic data, or may be extracted from a spatial discontinuity attribute volume, such as variance. The outputattribute dataset highlights fault surface features with greater clarity than may be directly interpreted from the original seismic data.

Why Spectral Decomposition

Uploaded by

Copyright:

Available Formats

Why Spectral Decomposition

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Why Spectral Decomposition

Uploaded by

Copyright:

Available Formats

Why Spectral Decomposition Since the development of the coherence cube (Bahorich and Farmer, 1995), seismic attributes

You might also like