OpendTect User Documentation version 3.2
Prev		Next

Spectral decomposition

Name

Spectral decomposition -- Attribute that returns the amplitude spectrum (FFT) or waveletcoefficients (CWT)

Description

Spectral Decomposition unravels the seismic signal into its constituent frequencies, which allows the user to see phase and amplitude tuned to specific wavelengths. It is a useful tool for "below resolution" seismic interpretation, sand thickness estimation, and enhancing channel structures. The user can choose between two types of transform:

FFT the Fast Fourier Transform. The FFT requires a time-gate and a step-size between the analyzed frequencies. This step can be interpreted as the frequency resolution.
CWT the Continuous Wavelet Transform. The CWT requires a wavelet type.

When choosing the CWT, you can set the wavelet type:

Morlet
Gaussian
Mexican Hat

In FFT only, the signal within the time-window will be transformed into frequency domain. The given step determines the output resolution, if necessary zeros will be added to acquire this resolution. The amplitude spectrum is calculated for the requested frequency. The time-window slides from top to bottom to cover the complete signal. In an ideal situation, the time-window encompasses one seismic event, which may be a superposition of multiple geological events which interfere in the seismic trace.

The CWT is defined as the sum over the signal multiplied by a scaled and shifted wavelet. The wavelet is shifted along the signal and at each position the correlation of the wavelet with the signal is calculated. The result is called a waveletcoefficient. The given frequency corresponds to the central wavelet frequency. The step determines the output resolution, which is especially interesting when evaluating this attribute.

Spectral Decomposition (CWT) applied to a horizon. Notice that each index frequency is describing the specific parts of channels (NNE-SSW oriented). Also thinner and thicker parts of horizon along channels are highlighed clearly.

Spectral Decomposition is usually applied relative to a mapped horizon, but OpendTect allows the user to decompose any (sub-)volume or slice through the data. In case the user would like to decompose a horizon, then the following work-flow can be applied;

Load the horizon and click in the tree to ensure that the horizon is the active element
Go to the attribute-set window
Select Spectral Decomposition from the list of attributes
Choose FFT or Wavelet Decomposition and set the time-window or the wavelet and set the 'step'
Press the "Evaluate attributes" icon
Specify the range of frequencies to 'calculate' (these are computed with the specified 'step' in the previous window)
Use the slider to 'movie-style' inspect the different components
Switch the toggle "Save slices on Accept" to on. This toggle can only be switched on for horizons because horizon decomposition can be time-consuming. With the toggle switched on, the slices are stored as horizon slices for later inspections
Press "Accept" to save the current frequency. Note that the attribute-set window now reflects the frequency that will be computed by this attribute
Give the attribute a name and add it to the list. When the user wants to apply this attribute in a later phase, only the specified component will be calculated
In order to re-examine the stored frequencies, go to "horizon" in the tree and press the right-hand mouse button. Choose "Select Attribute" from the pop-up menu. Then select " Surface Data" and select the requested components. Now, the user is able to scroll through the components by using Page-Up and Page-Down buttons.

Prev	Home	Next
Similarity	Up	Velocity Fan Filter