|
<< Click to Display Table of Contents >> Navigation: Modeling > Colop |
Colop (Color Inversion Operator) designs a spectral shaping operator that bridges the frequency content of seismic data and a well-derived acoustic impedance spectrum. The module reads both spectra from ASCII files, fits a power-law trend to the well impedance spectrum using linear regression in log-log space, and computes a frequency-domain residual operator that describes how the seismic spectrum must be reshaped to match the well impedance trend. This residual operator is then transformed to the time domain, tapered with a Kaiser window, phase-rotated by a user-supplied angle, and written out as a convolution-ready wavelet.
Use this module as the operator-design step in a colored inversion workflow. After running Colop, apply the resulting wavelet to the seismic data using a convolution module to convert seismic amplitudes into a broadband relative impedance volume. The module produces a suite of QC displays — including the normalized seismic spectrum, the regression-based well trend, the raw and tapered operators, and the final time-domain wavelet — so that the operator design can be fully inspected before application.
This module reads its primary inputs from external ASCII spectrum files specified in the Parameters section rather than from the project database. No seismic gather inputs are connected in the standard sense; all spectral data are loaded from the files described below.
Path to the ASCII file containing the seismic amplitude spectrum. The file must contain two whitespace-separated columns: frequency (Hz) and amplitude in decibels (dB). The module converts the dB values to linear amplitudes internally. This spectrum is typically obtained by averaging the amplitude spectra of a representative set of seismic traces over the zone of interest. Accepted file extensions are .txt and .dat. At least three frequency-amplitude pairs are required.
Path to the ASCII file containing the well-derived acoustic impedance amplitude spectrum. Like the seismic spectrum file, it must contain two columns: frequency (Hz) and amplitude in dB. This spectrum is computed from one or more well logs (typically the product of P-wave velocity and density) and represents the target spectral shape that the colored inversion operator will imprint onto the seismic data. Accepted file extensions are .txt and .dat.
Path to the output ASCII file where the computed spectral operator will be saved. This file records the frequency-domain representation of the shaping operator and can be used for archiving or further inspection. Accepted file extensions are .txt and .dat.
The beta (shape) parameter of the Kaiser window applied to the time-domain operator. After the residual spectrum is inverse-Fourier-transformed into the time domain, a Kaiser window is multiplied sample-by-sample to taper the edges of the operator and suppress truncation artifacts (ringing). A higher beta value produces a narrower, more aggressive taper that suppresses side-lobes more strongly but also reduces the operator's effective bandwidth; a lower value yields a less tapered, wider operator. The default value is 70. Adjust this parameter if the QC tapering display shows excessive ringing or if the operator spectrum appears over-smoothed.
Minimum normalized seismic amplitude below which the residual operator is set to zero. When the normalized seismic spectrum falls below this level at a given frequency, the module does not attempt to compute a shaping ratio at that frequency, preventing division by near-zero values that would create unstably large operator amplitudes. The value is expressed as a fraction of the peak normalized seismic amplitude, so a value of 0.2 (the default) suppresses the operator at all frequencies where the seismic amplitude is less than 20% of its peak value. Increase this value if the seismic data has a limited usable bandwidth or if low-amplitude noise at the band edges is causing operator instability.
Constant phase rotation, in degrees, applied uniformly to the entire residual spectrum before the inverse Fourier transform. This parameter compensates for a known or estimated phase of the seismic wavelet so that the output colored inversion result is in phase with the well impedance log. The default value of -90 degrees is appropriate for a 90-degree phase wavelet (common for zero-phase seismic converted from a minimum-phase source). Adjust this value based on wavelet phase analysis of your data; valid range is typically -180 to +180 degrees.
The number of time samples to retain in the final output wavelet, centered on the peak of the time-domain operator. After the inverse FFT and Kaiser windowing, the module extracts a symmetric window of this length from the center of the full-length operator and saves it as the output wavelet gather. The default is 100 samples. The minimum allowed value is 1. Longer operators capture more of the operator's tails and may improve convolution accuracy, but also risk including noise at the margins; shorter operators are more stable but may truncate meaningful energy. Choose a length that captures the main lobe and immediate side-lobes visible in the Color inversion operator QC display.
The sample interval of the output wavelet gather, in seconds. This controls the time resolution at which the wavelet is stored and should match the sample rate of the seismic data to which the operator will subsequently be applied. The default value is 0.002 s (2 ms). Note that the operator's internal time-domain sample interval is derived from the input seismic spectrum's maximum frequency (Nyquist), and the output gather is stored with this parameter's value.
Selects whether the computation runs on the CPU or a CUDA-capable GPU. For this module, which processes small spectral arrays rather than large seismic volumes, CPU execution is typically sufficient and recommended.
Controls whether the job is dispatched to a remote processing node in the distributed computing environment. Because Colop operates on small ASCII spectrum files and produces a single wavelet, distributed execution is not typically required.
The minimum number of data chunks sent to each processing node in a distributed job. This setting is relevant only when distributed execution is enabled.
When enabled, restricts the number of CPU threads used on remote processing nodes to the value set in the Number of threads parameter. Use this to reserve resources on shared cluster nodes.
An optional text string appended to the job name when submitting to a distributed processing queue. Useful for identifying and tracking multiple concurrent Colop runs in the job scheduler.
When enabled, allows the user to specify which CPU cores the process is bound to using the Affinity parameter below. Useful in shared-resource environments to prevent interference with other jobs.
Specifies the CPU core affinity mask when Set custom affinity is enabled. Only active if Set custom affinity is turned on.
The maximum number of parallel CPU threads allocated to this job. Because Colop processes a single spectrum rather than a large ensemble of gathers, this setting has minimal impact on run time and can be left at its default.
When enabled, this module is bypassed entirely during workflow execution. All downstream modules that depend on its outputs will receive unmodified pass-through data. Use this to temporarily disable the Colop step for testing purposes without removing it from the workflow.
The final colored inversion operator stored as a seismic gather with a single trace. This is the primary output of the module and represents the time-domain spectral shaping filter — phase-rotated, Kaiser-tapered, and truncated to the requested operator length. It is ready to be applied to the seismic data via a convolution module to produce a relative impedance volume. The time axis of this wavelet is centered at time zero, and the sample interval equals the Output sample rate parameter.
A QC scatter plot of the well impedance spectral data points in log10(frequency) vs. log10(amplitude) space. Each point represents one frequency sample from the input impedance spectrum file. This display allows the user to assess the distribution of well spectral data and verify that the linear regression is a reasonable fit.
The best-fit regression line overlaid on the well impedance scatter plot (see above). The line is computed by linear regression on the log10-log10 spectrum, yielding the power-law trend (slope and intercept) that describes how acoustic impedance amplitude varies with frequency. This trend is used to define the target spectral shape for the shaping operator.
The normalized seismic amplitude spectrum plotted against frequency. Amplitudes are divided by the peak value so that the spectrum ranges from 0 to 1. This curve shows the usable bandwidth of the seismic data and helps the user select an appropriate Threshold value.
The well-derived power-law trend spectrum evaluated at the same frequency grid as the seismic spectrum. This curve represents the target shape — the spectral amplitude the seismic data should have after applying the Colop operator. Comparing this curve against the normalized seismic spectrum reveals the degree of spectral mismatch that the operator must correct.
The raw (pre-tapering) residual operator spectrum: the ratio of the well trend spectrum to the normalized seismic spectrum at each frequency, normalized to a peak of 1. Frequencies where the seismic amplitude falls below the Threshold are set to zero. This curve shows the frequency-domain shaping required before any time-domain tapering is applied, and is useful for assessing the severity of the spectral correction.
A flat horizontal line plotted at the Threshold value across the full frequency range. This is displayed alongside the normalized seismic spectrum to make it easy to see which parts of the spectrum fall below the threshold and are therefore excluded from the operator computation.
The final time-domain colored inversion operator, displayed as amplitude vs. time (in seconds). This is the central QC output: it shows the shape, symmetry, and length of the operator that will be convolved with the seismic data. The operator is centered at time zero, phase-rotated by the Phase parameter, and truncated to the Number of samples of the operator. Inspect this display to confirm the operator has a well-defined main lobe and that side-lobes are adequately suppressed by the Kaiser taper.
The amplitude spectrum recovered by forward-Fourier-transforming the Kaiser-windowed time-domain operator. This display allows direct comparison of the tapered operator's frequency content against the raw residual spectrum, showing how much the Kaiser window has smoothed the operator in the frequency domain. Use this alongside the raw operator display to verify that the tapering has not excessively reduced the bandwidth of the output wavelet.