|
<< Click to Display Table of Contents >> Navigation: MultiFocusing > Data Enhance - 3D CO-MF - ext |
Data Enhance - 3D CO-MF - ext is an extended version of the 3D Common-Offset Multi-Focusing (CO-MF) data enhancement module. It improves the signal-to-noise ratio of 3D pre-stack seismic gathers by performing multi-channel partial stacking using a pre-computed CO-MF image database (.kdb file). Each output trace is reconstructed from a weighted combination of nearby CO-MF image traces, based on spatial proximity and offset similarity, using the Multi-Focusing principle to correctly account for wavefield curvature.
Compared to the standard 3D CO-MF module, this extended variant adds several advanced capabilities: flexible input/output geometry via on-disk trace header files (TVOD mode), an optional dual-interpolator regime for near-offset and far-offset zones (the "multi-AB" technique), an azimuth aperture filter to restrict which raw traces contribute to each output trace, offset-dependent decimation for large surveys, and a spatial fold-constraint map. These extensions make it suitable for large-scale 3D surveys where fine control over trace selection, memory efficiency, and output geometry is required.
The module supports multi-threaded CPU execution, GPU acceleration, and distributed (cluster) processing. It writes results incrementally to a .gsd output file and can resume interrupted runs using the Append write mode. Before running, use the Show MF bins for calculation custom action to preview the set of bins that will be processed with the current parameter settings.
When set to false (default), the input geometry is supplied via a SEG-Y data handle and in-memory trace headers. When set to true, the input geometry is read from a TVOD on-disk trace vector, which is required for very large surveys where the full trace header table does not fit in memory. Enabling this option shows the Input traces data handle and Trace handler read bulk size inputs, and hides the SEG-Y data handle.
When set to false (default), the output bin geometry is defined by in-memory trace headers connected via the Output trace headers input. When set to true, output geometry comes from an on-disk TVOD handle, suitable for surveys too large to hold in memory. This option controls which of the output geometry connectors is active.
Path to the output file where enhanced seismic traces will be written (.gsd format). This file is created or extended during processing. If the file already exists and contains previously computed chunks, the module can continue from where it left off when Write mode is set to Append.
The pre-computed 3D Common-Offset Multi-Focusing image database (.kdb file). This file is generated by a dedicated MF database creation module and contains the spatially interpolated MF wavefield representations for each CO gather. It is the primary source of enhancement weights used during processing. This input is mandatory.
Connects to the SEG-Y file or project data item that holds the raw pre-stack seismic data to be enhanced. This handle provides both the trace geometry and the amplitude data. Visible only when Use input headers from disk is false.
In-memory trace header table describing the input geometry (shot/receiver/CMP positions, offsets). Visible only when Use input headers from disk is false.
On-disk TVOD (trace vector on disk) handle for reading input trace geometry. Used instead of the SEG-Y data handle when Use input headers from disk is true. Suitable for surveys too large to load into memory.
In-memory trace header table describing the desired output geometry, i.e., the set of bins and offsets to reconstruct. Visible only when Use output headers from disk is false.
On-disk TVOD handle for reading output geometry. Used instead of the in-memory output trace headers when Use output headers from disk is true.
Optional picking corridor item. When connected, restricts the enhancement to a time-offset corridor defined by interactive picks on the MF semblance panel. This allows focusing the enhancement on a selected event or reflection zone rather than processing the full time axis.
Number of traces to read in a single I/O operation when reading from a TVOD on-disk trace vector. Default: 200000. Visible only when Use input headers from disk or Use output headers from disk is true. Larger values improve throughput at the cost of memory; reduce if memory pressure is a concern.
Optional 2D map (GMatrixItem) of near-surface velocity (m/s) for use with the "By min offset map" border type. When connected, the time boundary between the two interpolator zones varies spatially according to this map, accommodating lateral variations in the near-surface velocity field.
When enabled, the module applies Normal Moveout (NMO) correction to the output enhanced traces. Use this option when the downstream processing step requires NMO-corrected gathers rather than pre-NMO data. Default: false (output gathers are not NMO corrected).
Enables a mute that mimics the MF-domain mute applied during database construction, ensuring that the enhancement is applied only in the same time-offset region used to build the .kdb file. Default: false. Enable this if artifacts appear at early times or at offsets outside the MF aperture.
A group of parameters controlling how the Multi-Focusing wavefield image is constructed from the CO-MF database. These include the number of summation directions, the angular range (from/to angle in degrees), signal-to-noise enhancement mode, correlation threshold, angle distance selection, and radial distance selection. The default angular sweep covers the full azimuth range and uses a single summation direction. Adjust the angular parameters if the survey has strongly directional noise or acquisition footprint.
Selects how the early-arrival mute is defined. Use mute factor (default) applies a constant-offset mute based on the Mute factor value. Use mute function applies a time-variant mute function defined elsewhere in the project. The mute prevents the algorithm from using the direct-wave zone where the MF approximation is not valid.
The mute onset time in milliseconds applied to suppress the direct-wave zone when Mute type is set to "Use mute factor". Default: 1000 ms. Minimum: 0 ms. Set this value to the approximate two-way travel time to the shallowest reflector of interest.
Controls how the Common-Offset aperture is defined when selecting MF image traces for summation. XY aperture (default) selects traces based on spatial distance in the X-Y plane. Offset aperture selects traces based on offset distance. Use Offset aperture for sparse, azimuthally uniform surveys; use XY aperture when spatial coverage is the primary concern.
When enabled, output traces that could not be reconstructed (because no suitable MF or raw traces were found within the search apertures) are set to zero and excluded from the output. Default: false. Enable this to avoid writing meaningless noise traces in sparsely covered areas of the survey.
A group of parameters defining how CO-MF image weights are spatially interpolated to the output bin locations. The method can be set to Triangulation (faster, no smoothing) or Kriging (slower but better for irregular geometries).
Selects the spatial interpolation algorithm used to map CO-MF weights onto each output bin. Triangulation (default) uses a Delaunay triangulation and is fast and exact at known locations. Kriging provides smoother results and handles irregular grids better. Choose Kriging for surveys with non-uniform MF bin spacing.
The variogram model used when Interpolation method is Kriging. Default: Exponential. The Exponential model is appropriate for most seismic surveys. Spherical models are better for data with a hard range limit; Gaussian models produce the smoothest interpolation and suit slowly varying fields.
The spatial correlation range (m) for Kriging interpolation of MF weights. Default: 100000 m. This defines the distance beyond which MF bin locations are considered uncorrelated. In practice, set this to approximately the diameter of the survey or larger to ensure all available MF bins contribute to each output location.
The number of nearest MF bins to use in the Kriging system when interpolating weights. Default: 15. Increasing this value improves interpolation stability at the cost of computation time. Reduce it for faster processing on dense regular grids.
A container grouping the MF trace selection and raw data trace selection sub-groups. These parameters jointly control which MF image traces from the .kdb database and which raw input seismic traces are considered when reconstructing each output trace.
Sub-group controlling the selection of Multi-Focusing CO image traces from the .kdb database for each output bin. Key parameters here govern the spatial and offset search windows within which the algorithm looks for contributing MF traces.
Maximum lateral distance (m) from an output bin centre to a CMP in the MF database for that CMP to be considered as a contributing MF bin. Default: 500 m. Range: 0–2000 m. Increasing this value brings in more MF traces from farther away, improving coverage in sparse areas but potentially smearing lateral resolution.
Maximum offset distance (m) from the target output trace offset to a CO-MF trace offset for that CO-MF trace to be selected. Default: 500 m. Range: 0–100000 m. This parameter limits which offset classes in the MF database are used for each output trace; set it to cover the expected range of offset variation within your survey.
The maximum number of MF database bins (CMPs) that may contribute to the reconstruction of a single output trace. Default: 6. Minimum: 1. Higher values allow the algorithm to draw from a larger neighbourhood, which can improve SNR but increases computation time. Reduce this value for faster processing on dense, regular surveys.
When enabled, MF database bins are only used if the output trace offset exceeds the MF minimum offset to operate threshold. Default: false. Enable this to prevent enhancement at very short offsets where the MF approximation may produce artifacts.
The minimum source-receiver offset (m) at which MF-based enhancement is applied. Default: 100 m. Range: 0–2000 m. Active only when Use min offset constraint is enabled. Traces with offsets below this value will not be enhanced by the MF database and will pass through unchanged or be zeroed.
When enabled, if no MF database bins are found within the primary search aperture for a given output bin, the algorithm will fall back to using alternative MF bins from a wider area. Default: false. Enable this to improve coverage in areas with gaps in the MF database, at the risk of slightly reduced spatial accuracy.
Activates a dual-interpolator regime in which two separate sets of CO-MF weights are computed and combined: one for times before a defined border (shallow zone) and another for times after it (deep zone). This allows different MF aperture configurations to be used at different depths. Default: false. Enabling this option reveals the Border type, A and B border time, V0 border time, Use taper, and Taper window parameters.
Determines how the time boundary between the two interpolator zones is computed when Use multi-ab is enabled. Constant (default) uses the fixed value from A and B border time for all bins. By min offset map computes the boundary time spatially using the connected V0 map and each trace's offset, making the boundary laterally variable to account for a non-flat surface or varying near-surface velocity.
The fixed two-way travel time (s) at which the boundary between the two interpolator zones is placed when Border type is Constant. Default: 0.5 s. Range: 0–2000 s. Set this to the approximate two-way time of the boundary between your shallow and deep zones of interest.
The near-surface replacement velocity (m/s) used to convert minimum offset to a time boundary when Border type is "By min offset map". Default: 2000 m/s. Range: 100–200000 m/s. This value should correspond to the average near-surface P-wave velocity in the survey area.
When enabled, a taper window is applied at the boundary between the two interpolator zones to avoid abrupt transitions. Default: true. Disable only if testing the effect of the zone boundary without smoothing.
The half-length of the taper applied at the zone boundary (s). Default: 0.1 s. Minimum: 0.01 s. Larger values produce a smoother transition but reduce the effective extent of each zone. Active only when Use taper is enabled.
Sub-group controlling which raw input seismic traces are gathered around each output bin for amplitude reconstruction. These parameters define the spatial and offset search radii applied to the actual seismic data (not the MF database). Correct configuration here is critical for gathering enough input data without including traces from geometrically unrelated CMPs.
The maximum number of raw input traces allowed to contribute to a single output trace when Max fold type is Constant. Default: 50000. This is effectively an upper bound on the super-gather size. Reduce it to limit memory use in high-fold areas of a 3D survey.
An integer stride factor applied to raw traces when building the super-gather. Default: 1 (use every trace). Setting this to 2 uses every other trace, to 3 every third trace, and so on. Use this to reduce processing time and memory consumption on very dense surveys where the full fold is not needed to achieve the desired enhancement.
The minimum offset (m) at which the decimation factor begins to be applied. Default: 0 m. Traces with offsets below this value are always used (not decimated). This allows near-offset traces to be preserved at full density while far-offset traces are thinned out for efficiency.
Maximum lateral distance (m) from the output bin centre to a raw input trace midpoint for that trace to be included in the super-gather. Default: 150 m. Range: 0–100000 m. This defines the spatial aperture of the raw data selection. Increase it if the survey has large bin spacing or to include more traces for smoothing; decrease it to tighten lateral resolution.
Maximum distance (m) between the source-receiver midpoint of an input trace and the corresponding MF trajectory point (CS-CR distance) for that trace to be eligible. Default: 250 m. Range: 0–100000 m. This is a stricter geometric criterion than the CMP distance and accounts for the MF non-hyperbolic moveout geometry. Adjust together with Max distance to CMP to control the effective aperture.
The total azimuth range (degrees) within which raw input traces are accepted. Default: 360 degrees (all azimuths). Reducing this value restricts the enhancement to traces whose source-receiver azimuth falls within the specified window centred on the nominal output trace azimuth. Use narrower apertures to suppress azimuthally anisotropic noise or footprint on wide-azimuth 3D surveys.
When enabled, the azimuth aperture window is applied symmetrically about the nominal azimuth, accepting traces from both the forward and reverse azimuths simultaneously. Default: false. Enable this for surveys where reciprocal shot-receiver pairs are present and should be treated equivalently.
Optional 2D spatial matrix (GMatrixItem) that provides a laterally variable scale factor for the CMP aperture radius. When connected and the FZ matrix uses type is set to a mode that uses the CMP channel, the local factor from this map is multiplied by the aperture value, allowing the effective search radius to vary across the survey. Useful when the geology or acquisition geometry produces spatially variable fold.
Selects which aperture parameter(s) are scaled by the FZ factor matrix. Default: "cmp and cs-cr". none disables FZ scaling entirely. cmp scales only the CMP search radius. cs-cr scales the source-receiver midpoint distance criterion. cmp and cs-cr scales both simultaneously.
A container grouping parameters that govern the output fold map computation, which can be generated via the Create output fold map custom action. The fold map shows how many raw traces contribute to each output offset class, which is useful for quality-control before running the full enhancement.
When enabled, the fold map computation applies the per-offset density and minimum fold constraints defined in the Offset density table. Default: false. Enable this to produce a constrained fold map that reflects the actual trace-selection behaviour during enhancement.
A table of (Offset, Density, Min fold) triplets that defines the desired fold density at each offset class for the output fold map. Default offsets: 200, 500, 1000, 1500, 2000, 2500, 3000, 3500, 4000 m; default density: 1; default minimum fold: 20. Adjust these values to match your survey's nominal fold at each offset class. Used only when Use constraints is enabled.
An integer factor that controls the time-sample stride used internally during the enhancement computation. Default: 1 (process every sample). Values greater than 1 reduce the effective sample rate during processing, which can accelerate the computation at the cost of temporal resolution. Use only for rapid preview runs or when the data is heavily oversampled relative to the signal bandwidth.
When enabled (default: true), the interpolation and stabilisation time windows adapt dynamically to the local signal characteristics, allowing the enhancement to be more aggressive in high-SNR zones and more conservative in noisy zones. Disable this option if you want strict control over the window length via the Interpolation window length and Stabilization window length parameters.
The half-length of the time window (s) used for computing the correlation-based weights between MF image traces and the raw data during enhancement. Default: 0.010 s (10 ms). Range: 0–1000 s. Longer windows produce more stable weight estimates at the cost of temporal resolution. Set this to approximately half the dominant period of the signal.
The half-length of the time window (s) used for computing the regularisation (stabilisation) term that prevents instability in the weight inversion when signal energy is low. Default: 0.010 s (10 ms). Range: 0–1000 s. Set equal to or slightly larger than the Interpolation window length for most surveys.
Controls how results are written to the output file. Direct (default) creates a fresh output file and overwrites any existing content. Append resumes a previous run by writing only the bins that were not yet completed, allowing interrupted jobs on large surveys to be restarted without losing already-computed output.
A container grouping parameters that restrict processing to a spatial sub-area of the survey defined by inline and crossline ranges. Setting all four values to -1 (default) processes the full survey extent as defined by the output geometry. Use these parameters to limit processing to a test patch or to split a large job into geographic sub-regions for parallel submission.
The first inline number to include in the calculation. Default: -1 (no lower limit; process from the first available inline). Set a positive value to skip inlines before this number.
The last inline number to include in the calculation. Default: -1 (no upper limit; process to the last available inline). Set a positive value to stop processing after this inline.
The first crossline number to include in the calculation. Default: -1 (no lower limit). Use together with the inline limits to define a rectangular patch on the survey grid.
The last crossline number to include in the calculation. Default: -1 (no upper limit). Use together with the other area limits to define the processing patch.
An advanced group containing experimental and diagnostic parameters. These settings are intended for specialist use and should not normally be changed in production workflows.
Switches between two internal GPU execution paths. Default: false (uses the established GPU path). Enable this option only when directed by support to test a newer GPU kernel implementation. When false, the standard GPU code path is used regardless of the Execute on setting.
The maximum number of traces that can be held in GPU memory at one time during a single processing chunk. Default: 4096. Increase this value if the GPU has ample VRAM and the super-gathers are large; decrease it if GPU out-of-memory errors occur. Active only when GPU execution is selected.
Optional path to a text file (.txt) listing inline/crossline pairs that should be recalculated even if they are already marked as completed in the output file. This is used to force recomputation of specific bins, for example after correcting input data for a subset of the survey, without reprocessing the entire volume.
Parameters controlling the in-memory cache used for reading SEG-Y data during processing. Adjusting these settings can improve I/O throughput on large surveys where disk access is the bottleneck.
Selects whether the enhancement computation is run on the CPU or GPU. GPU execution can provide significant speedups for large super-gathers, but requires a compatible NVIDIA GPU and sufficient VRAM. Use CPU mode when no GPU is available or when the GPU VRAM is insufficient for the chosen fold and window sizes.
Options for distributing the computation across multiple cluster nodes. When enabled, the survey is partitioned into chunks that are sent to remote worker nodes for parallel processing. This is the recommended mode for processing large 3D surveys that would take impractically long on a single workstation.
The number of output bins processed in each distributed task chunk. Larger bulk sizes reduce overhead from task management but may cause load imbalance if some chunks take much longer than others.
When enabled in distributed mode, limits the number of threads each remote worker node may use. This is useful when the cluster nodes are shared with other jobs and full CPU utilisation would impact other users.
An optional text string appended to the distributed job name, making it easier to identify this specific run in the cluster job queue when multiple enhancement jobs are submitted simultaneously.
Enables manual CPU core affinity settings, allowing the process to be pinned to specific CPU cores. Leave disabled unless there is a specific performance reason to restrict core usage.
Specifies the CPU core mask or core list when Set custom affinity is enabled. Consult system documentation for the correct format.
The number of CPU threads used for parallel processing on the local machine. Set to the number of physical CPU cores for best performance. Reduce if other processes need CPU resources simultaneously.
Allows custom shell scripts to be executed automatically before and after the module runs. Useful for automating file management, sending notifications, or triggering downstream processing steps.
Path to a script that is executed before the enhancement starts. Use this to prepare input data, check disk space, or log the start of the job.
Path to a script that is executed after the enhancement completes. Use this to compress output files, move results to a project directory, or send a completion notification.
When enabled, this module is bypassed entirely during pipeline execution. Use this to temporarily disable the enhancement step without removing it from the workflow, for example when comparing results with and without enhancement.
The enhanced pre-stack seismic gathers are written directly to the file specified by Output file name (.gsd format). There are no connector-based output data items; all results are file-based. The output fold maps generated by the Create output fold map action are displayed in the map view and are not written to a separate file. The Show MF bins in storage action displays the set of MF database bins in the map view for quality control.
Opens the .kdb storage file and displays the locations of all MF database bins in the map view. Use this to verify that the storage file covers the intended survey area and to check MF bin density before running the enhancement.
Recalculates and displays the set of output bins that will actually be processed given the current parameter settings, including the calculation area limits and the raw-data availability check. Run this action before submitting a large job to confirm that the expected bins are in scope and to catch any misconfiguration.
Exports the MF CO-MF parameters for the currently selected bin to a SEG-Y file. A file-save dialog will appear to choose the output path. This is a diagnostic action used to inspect the MF image parameters for a specific location.
Computes and displays a fold map showing the number of raw input traces that contribute to each output bin at each offset class, using the current Fold map params and Distance for trace selection settings. Use this as a quality-control step to verify that sufficient fold is available across the survey before committing to a full processing run.