|
<< Click to Display Table of Contents >> Navigation: Multiples > Radon multiple attenuation by velocity |
This module attenuates multiple reflections in NMO-corrected CMP gathers using a high-resolution parabolic Radon transform guided by an external velocity model. After NMO correction, primary reflections align near zero residual move-out while multiples retain significant residual curvature. The module decomposes each gather into parabolic events in the tau-p domain using an anti-aliasing least-squares method, identifies events whose velocity is below a user-defined percentage of the primary velocity as multiples, and subtracts them from the input to produce a primary-only gather.
The velocity boundary between primaries and multiples is defined as a time-varying table of velocity scale factors. An optional horizon constraint can restrict where attenuation is applied, protecting primaries near a specific geological boundary. QC windows provide velocity semblance and Radon spectrum displays before and after filtering.
The seismic data container. Input data should be geometry-assigned CMP gathers sorted by CMP/offset or CMP/absolute offset. The gathers must not be NMO-corrected before input; NMO is applied internally using the velocity model.
The input CMP gather to be processed. Connect to the output gather of the preceding module in the workflow.
The external RMS velocity model used to perform NMO correction before the Radon transform and to define the primary-velocity reference curve. A well-estimated velocity model is critical for good separation between primaries and multiples in the tau-p domain.
A time-varying table of velocity scale factors that defines the boundary between primaries and multiples. Each row specifies a two-way time and a corresponding factor (e.g., 0.85 means 85% of the primary velocity). Events with velocity below this percentage of the primary velocity at that time are treated as multiples and attenuated. Multiple rows allow the threshold to vary with depth, accommodating surveys where the primary/multiple velocity contrast changes significantly with time.
When enabled, the module generates additional visualisation windows showing the Radon transform spectrum, velocity semblance before and after multiple attenuation, and NMO gather comparisons. These windows are valuable for tuning the velocity factor table and Radon parameters, but require additional computation. Disable this option for production processing to improve speed. Default: enabled.
Parameter group controlling the range and resolution of the parabolic Radon transform used to decompose the gather into tau-p components.
The minimum residual move-out curvature (in seconds) at the maximum offset used in the Radon transform. Negative values model events with inverse curvature (rare in practice). Set this to a small negative value to capture near-zero-curvature events. Default: -0.5 s.
The maximum residual move-out curvature (in seconds) at the maximum offset. Set this large enough to encompass the residual move-out of the strongest multiples. Larger values include more of the multiple energy but may also capture steeply-dipping primary events. Default: 1 s.
The sampling interval (in seconds) of the Radon transform, i.e., the curvature step between successive parabolas. Setting this approximately equal to the dominant wavelet period gives the best resolution without over-sampling. Smaller values increase computation time. Default: 0.008 s.
The width of the cosine taper (in seconds) applied in the P dimension of the Radon domain around the P min, P mid, and P max boundary values. Tapering avoids abrupt transitions that would introduce artefacts in the output. Default: 0.05 s.
The width of the cosine taper (in seconds) applied in the time (tau) dimension at the start and end of the attenuation window. Use this to avoid edge artefacts at the top and bottom of the attenuated zone. Default: 0.2 s.
Parameter group controlling the NMO and velocity semblance computation used internally for the Radon transform and QC displays.
The maximum allowable NMO stretch (as a percentage) before a sample is muted. High stretch at far offsets and shallow times distorts the wavelet; setting a stretch limit protects the gather from excessive NMO distortion. Default: 10%.
The number of time samples over which the velocity semblance is smoothed in the time direction before display. Smoothing reduces noise in the semblance panel and makes picking easier during QC. Default: 10 samples.
The replacement velocity (in m/s) used in the NMO operator for accurate time-depth conversion. This is typically set to the near-surface velocity or the water velocity for marine data. Default: 0 m/s (disabled).
The time window length (in seconds) used for normalising the velocity semblance before display. Normalisation improves the visibility of semblance peaks at all depths. Default: 0.1 s.
The lowest velocity (in m/s) displayed on the QC velocity semblance panels. Set to cover the range of both multiple and primary velocities at the target depths. Default: 1000 m/s.
The highest velocity (in m/s) displayed on the QC velocity semblance panels. Default: 6000 m/s.
The velocity sampling interval (in m/s) of the semblance grid. Smaller steps produce a finer semblance but increase computation time. Default: 50 m/s.
The length of the time window (in seconds) used to smooth the semblance display. Smoothing reduces noise in the panel and makes it easier to identify primary and multiple velocity trends. Default: 0.05 s.
Parameter group with optional pre- and post-processing controls for the Radon filter.
The minimum frequency (Hz) of the band-pass filter applied before the Radon transform. Use this to suppress very low-frequency energy that might bias the multiple model. Leave at 0 to process the full bandwidth.
The maximum frequency (Hz) of the band-pass filter applied before the Radon transform. Use this to limit the transform to the signal band and exclude high-frequency noise. Default: 100 Hz.
A small stabilization factor (as a percentage of the maximum spectral amplitude) added to the least-squares inversion to prevent numerical instability during the inverse Radon transform. Increase this value slightly if the output contains high-frequency artefacts. Default: 0.1%.
When enabled, an AGC (Automatic Gain Control) is applied before the Radon filter to equalise amplitude levels and improve separation of primaries from multiples. After filtering, the AGC gain is reversed so that output amplitudes reflect the true input levels. Default: enabled.
The time window length (in seconds) used for the AGC amplitude equalization. Shorter windows equalise more aggressively; longer windows preserve more of the natural amplitude variation. Only active when Use AGC is enabled. Default: 0.5 s.
An additional protection zone (in seconds) above the first-arrival time defined by the velocity model. Within this zone the attenuation is suppressed to avoid removing shallow primary energy near the first-break mute boundary. Default: 0 s (no extra protection).