subsurfaceAI
Different Methods for Spectral Decomposition in AttributeStudio
Published:
Author: subsurfaceAI

One of the most popular features in AttributeStudio is Spectral Decomposition, which includes five different methods for decomposing the seismic signal into constituent frequencies.  The table below compares the methods for Spectral Decomposition available in AttributeStudio (by the way, this table is part of the AttributeStudio User Guide):

Comparison of Spectral Decomposition Methods in AttributeStudio
Spectral Decomposition Method Windowing Approach Advantages and
Disadvantages
Discrete Fourier Transform (DFT)
  • Fixed-time window
  • User specifies length of time window
  • Fixed-time window
  • User specifies length of time window
Continuous Wavelet Transform (CWT)
  • Moving, scalable time window
  • Window size automatically changes with frequency
  • Allows adaptive sampling of trace, which preserves some events better than DFT
  • Cannot adequately resolve low-frequency events that are closely spaced in time domain
Time-Frequency Continuous Wavelet Transform (TFCWT)
  • Moving, scalable time window (like CWT)
  • Window does not average neighboring frequencies
  • Generates a real time-frequency map, with higher resolution than CWT for a time–frequency spectrum
  • Computationally intensive
S-Transform (ST)
  • Moving, ƒ-dependent time window
  • Window length is decided by the frequency (has a more rigorous relationship with ƒ than CWT or TFCWT)
  • Results are similar to TFCWT, with higher resolution than CWT for a time–frequency spectrum
  • Computation time is less than TFCWT
Continuous Wavelet Packet Transform (CWPT)
  • Moving, scalable time window
  • Window size automatically changes with frequency
  • Has good frequency resolution
  • Computation time is comparable to CWT