|
<< Click to Display Table of Contents >> Navigation: Scaling > Signal spectrum conversion |
Signal spectrum conversion reshapes the amplitude spectrum of every trace so that it matches a user-defined target spectrum. The module reads two ASCII spectrum files — an input (source) spectrum describing the current data and an output (target) spectrum describing the desired result — and computes a shaping filter whose frequency-domain response is the ratio of the target to the source. This filter is then applied trace by trace via convolution in the time domain.
Typical applications include equalising data from different vintages or acquisition systems, matching a processed dataset to a reference volume, and correcting systematic spectral imbalances caused by the near surface or the acquisition source. The module supports a difference display for before-and-after comparison. Use the Init operator custom action to pre-compute the shaping filter and inspect it in the graphics views before running the full dataset.
The primary seismic dataset to be spectrally shaped. Connect the data flow from the preceding processing module here.
Handle to the SEG-Y file associated with the primary dataset, propagated through the sequence.
Trace header table for the primary dataset, passed through unchanged.
The current gather from the processing sequence. Each trace in the gather is filtered independently by the precomputed spectral shaping operator.
Stack line passed through to downstream modules.
Crooked-line geometry definition for 2D surveys, passed through.
3D bin-grid definition, passed through.
Pre-sorted trace index, passed through.
Path to the ASCII file defining the current (source) amplitude spectrum of the data. The file should contain frequency–amplitude pairs (in dB) at evenly or unevenly spaced frequencies. This spectrum is used as the denominator of the shaping filter. It can be obtained from the Fast spectral analysis or Frequency spectrum f-t modules.
Path to the ASCII file defining the desired (target) amplitude spectrum. The shaping filter is computed as the ratio of this target spectrum to the input spectrum, band-limited by the Min frequency and Max frequency parameters. Typically this file represents the spectrum of a reference dataset or a manually designed target shape.
Length of the time-domain shaping filter (convolution operator), in seconds. Default: 4.0 s. A longer operator can represent more complex spectral shapes with higher frequency resolution but increases computation time. For smooth, gradually varying spectral corrections a length of 0.1–0.5 s is usually sufficient. Increase the length only if the required spectral correction varies steeply with frequency.
Sample interval of the computed shaping filter, in seconds. Default: 0.004 s. Set this to match the sample interval of the seismic data. A finer sample rate increases the Nyquist frequency of the operator and is required when processing data with a sample interval finer than 4 ms.
Lower frequency limit of the spectral shaping operator, in Hz. Default: 0 Hz. The operator is set to unity (no correction applied) below this frequency. Increase this value to protect low-frequency content from being altered by noise or instability in the source spectrum at the low end.
Upper frequency limit of the spectral shaping operator, in Hz. Default: 250 Hz. The operator is tapered to zero above this frequency. Set this to the Nyquist frequency of the data or to the usable bandwidth of the seismic signal, whichever is lower, to avoid amplifying high-frequency noise.
When enabled, g-Platform automatically connects compatible data items from the preceding module. Disable to wire connections manually.
Determines how the module responds when NaN or infinity values are detected in the input data. Fix replaces bad values with zero. Notify logs a warning and continues. Continue silently proceeds.
When enabled, the module computes and outputs a difference gather (input minus output) in addition to the shaped gather. The difference gather is available on the Gather of difference output port and can be used to inspect the noise or the spectral change introduced by the correction.
Selects the compute device. CPU is the default and available on all systems.
Options for distributing processing across multiple cluster nodes.
Minimum number of gathers dispatched to each node as a work unit.
Caps the number of threads per cluster node to avoid resource contention.
Text appended to the distributed job name for identification in cluster queues.
Enables manual CPU core affinity for processing threads.
CPU core or core range to which threads are pinned. Active only when Set custom affinity is enabled.
Number of parallel CPU threads. Set to 0 to use all available cores.
When enabled, the module is bypassed and input data passes through unchanged. Useful for A/B testing of the spectral correction.
The primary output dataset with spectrally shaped trace amplitudes.
SEG-Y file handle propagated downstream.
Trace header table, unchanged from the input.
The spectrally shaped output gather, ready for further processing or saving.
Stack line passed through unchanged.
Crooked-line definition passed through unchanged.
Bin grid passed through unchanged.
Sorted trace index passed through unchanged.
The difference gather (input minus shaped output), available when Calculate difference is enabled. Connect this to a display or save module to inspect the energy removed or added by the spectral shaping filter.
Reads the Input spectrum and Output spectrum files, computes the spectral shaping filter, and displays it in the Graphics views (Operator in time domain and Operator in frequency domain). Run this action before executing the full sequence to verify that the operator looks correct. If the source or target spectrum files have not been set, the action will report an error.