|
<< Click to Display Table of Contents >> Navigation: MultiFocusing > Data Enhance - 2D CO-MF (ext) |
Data Enhance - 2D CO-MF (ext) applies Multi-Focusing (MF) wave-field reconstruction to enhance or interpolate pre-stack seismic gathers along a 2D line using common-offset (CO) focusing operators. The module reads CO-MF focusing weights from a pre-computed storage database (KDB file) generated by the Engine - 2D CO-MF step, then uses those weights to re-synthesize each output gather trace by coherently stacking contributions from neighboring raw input traces. The result is a set of enhanced or regularised pre-stack gathers with improved signal-to-noise ratio and more complete offset coverage.
The "(ext)" variant extends the standard Data Enhance - 2D CO-MF workflow with additional controls: a time-variant maximum fold table that limits the number of contributing traces as a function of two-way travel time, support for offset-based aperture mode, optional NMO correction of the output gathers, and a dynamic correlation window. The output is written to a GSD file that can be used as input to subsequent migration or stacking steps.
Supports CPU and GPU execution and can be distributed across multiple nodes for large 2D datasets. The calculation area can be restricted to a subset of output bins using the First/Last bin range parameters.
Controls how input trace geometry is supplied. When set to false (default), trace headers are read from an in-memory trace header item connected via the Input trace headers connector together with a SEG-Y data handle. When set to true, input geometry and amplitudes are read directly from a trace-vector-on-disk (TVOD) file handle, which is suitable for very large datasets that do not fit in memory. Toggling this flag changes which input connectors are visible.
Controls how the output trace geometry (the list of bins and offset positions that will be reconstructed) is specified. When false (default), output bin positions are taken from the Output trace headers connector. When true, the output geometry is read from a TVOD disk file, enabling reconstruction onto a large or irregular output grid without loading all headers into memory.
Path to the CO-MF database file (.kdb) produced by the Engine - 2D CO-MF module. This file contains all pre-computed MF focusing weights, time-quantization tables, and geometry for each output bin. The database must match the seismic data in terms of sample count; the module will report an error if the sample counts differ. This is a mandatory input — processing cannot proceed without a valid storage file.
Path and file name for the enhanced output gathers, written in GSD format (.gsd). The GSD format is g-Platform's native binary seismic format and stores the reconstructed pre-stack data for use in downstream modules. When Write mode is set to Append, results are added to an existing file rather than overwriting it.
Connector for the input seismic data in SEG-Y format. Used when Use input headers from disk is false. The raw pre-stack SEG-Y file provides the amplitude data that is read and redistributed by the MF focusing operators to synthesise the output gathers.
Trace header collection for the input dataset, containing the bin coordinates and offset information for every raw input trace. Used together with the SEG-Y data handle when Use input headers from disk is false. The module uses these headers to locate which input traces fall within the aperture of each output bin.
TVOD (Trace Vector On Disk) file handle for reading input seismic data directly from disk. Used when Use input headers from disk is true. This mode is recommended for 2D lines with very large trace counts where loading all headers into memory is impractical. The bulk read size parameter controls how many traces are loaded per read operation.
Trace header collection that defines the target bin positions and offsets for the reconstructed output gathers. Used when Use output headers from disk is false. Each bin listed in this header collection will receive a reconstructed gather in the output GSD file.
TVOD disk file handle that provides the output geometry (bin and offset positions) when Use output headers from disk is true. The module reads all bins from the TVOD structure and uses them as the reconstruction targets.
Number of traces read from disk in a single I/O operation when using TVOD disk mode. Default: 200,000 traces. Larger values reduce the number of disk seeks but increase memory usage. Only visible when Use input headers from disk or Use output headers from disk is enabled.
Optional MF picking item that can be used to supply a picking corridor or constraint for the MF wave-field reconstruction. Connect this item if a CO-MF picking result has been produced and you wish to constrain the enhancement to focus only on energy near picked events.
Optional 2D map (GMatrixItem) providing a spatially varying near-surface velocity V0, used to refine the MF operator moveout within each output bin. When connected, the V0 value at each bin location is looked up from the map and passed to the CO interpolator. If not connected, V0 defaults to the value stored in the KDB database. Connect this item when significant near-surface velocity lateral variation is expected along the 2D line.
When enabled, the enhanced output gathers are NMO-corrected using the velocity information stored in the KDB database, so that reflections appear as flat events as a function of offset. Default: false. Enable this option when the downstream processing step requires NMO-corrected gathers (for example, before AVO analysis or pre-stack attribute extraction).
When enabled, applies a mute pattern that emulates the aperture boundary behaviour of the Multi-Focusing operator, tapering amplitudes near the edges of the MF aperture to suppress edge effects in the reconstructed gathers. Default: false. Use this option to suppress artefacts along the edges of the MF imaging aperture when they are visible in the output gathers.
A group of parameters that control how the MF wave-field image is formed from the available focusing directions and weights. These parameters are shared across all MF data enhance modules.
Directions — number of azimuthal directions used to form the MF image stack. Default: 1. For a 2D line a single direction is sufficient in most cases.
From angle / To angle — angular range (degrees) over which MF contributions are accepted. Default: -90° to 90°. Narrowing this range restricts the reconstruction to a particular dip sector, which can reduce noise from steeply dipping events.
SN enhance — when enabled, applies additional signal-to-noise enhancement weighting during image formation. Default: false.
Correlation threshold — minimum normalised correlation coefficient required for a MF contribution to be accepted. Default: 0.10 (10%). Increase this value to reject weakly correlated contributions and suppress incoherent noise in the output.
Min angle distance selection — minimum angular separation (degrees) between selected MF directions. Default: 1°. Increasing this value enforces greater diversity between selected contributing directions.
Min radius distance selection — minimum spatial radius separation (m) between selected contributing MF bin positions. Default: 100 m. This prevents the same spatial region from being over-represented in the reconstruction.
Selects how traces are muted before reconstruction to suppress far-offset noise or stretching artefacts. Default: Use mute factor. Use mute factor applies a constant time-offset mute defined by the MuteFactor parameter. Use mute function applies a user-defined time-variant mute function stored in the KDB database. Choose Use mute function when a detailed mute was defined during the engine step and should be re-applied consistently to the enhanced output.
Time delay (ms) applied as a constant mute to input traces before they contribute to the MF reconstruction. Default: 1000 ms. Active only when Mute type is set to Use mute factor. Increase the value to mute less aggressively (accept more far-offset data); decrease it to suppress more of the far-offset wavefield that may be affected by NMO stretch.
Defines how the common-offset aperture is measured when selecting contributing raw traces for each output trace. Default: XY aperture. In XY aperture mode, traces are selected based on their midpoint (CMP) distance from the output trace position. In Offset aperture mode, traces are selected based on their absolute offset value, which is better suited to datasets with irregular source-receiver geometry. Use Offset aperture for land data with irregular shooting geometry or for lines with significant azimuth variation.
When enabled, output traces that contain only zeros (i.e. no raw input traces fell within the aperture) are removed from the output gather rather than written as dead traces. Default: false. Enable this option to reduce the size of the output GSD file and to avoid dead-trace overhead in downstream processing steps.
Container grouping all aperture distance parameters that control which input traces are selected for each output bin reconstruction. This group is divided into two sub-groups: MF Trace Selection (controlling aperture into the MF database) and Raw Data (controlling aperture into the raw seismic input).
Sub-group defining spatial search distances into the CO-MF database. These distances determine which pre-computed MF bins from the KDB file contribute to the reconstruction of each output trace.
Maximum lateral distance (m) between the output bin centre and a MF database bin CMP position for that MF bin to be used in the reconstruction. Default: 100 m; valid range: 0–2000 m. Set this to approximately one CMP spacing. A value that is too small may leave some output bins without MF support; a value that is too large may introduce contributions from distant structural regions.
Maximum offset difference (m) between the target output trace offset and a CO-MF database trace offset for that database trace to be considered a valid contributor. Default: 1000 m; valid range: 0–100,000 m. This parameter defines the offset search radius in the KDB for each reconstructed output trace. Larger values allow contributions from a wider offset range stored in the database.
When enabled (default: true), the offset comparison between the output trace and MF database traces uses the absolute value of the offset, so that positive and negative offset traces are treated equivalently. Disable this option only if the database contains signed offsets and directional asymmetry must be preserved.
Sub-group defining how raw input seismic traces are selected for each output bin reconstruction. These parameters limit the number and spatial extent of raw traces gathered from the input dataset before the MF weights are applied.
Maximum lateral distance (m) between the output bin centre and the midpoint of a raw input trace for that trace to be included in the reconstruction aperture. Default: 150 m; valid range: 0–100,000 m. This is typically set to 1–2 CMP spacings. Increasing this value broadens the aperture, which can improve fold at the cost of including more spatially distant traces.
Maximum lateral distance (m) from the output shot/receiver position to the corresponding position of a raw input trace. Default: 150 m; valid range: 0–100,000 m. This distance is checked independently for the source side and the receiver side. Use together with Max Distance To CMP to control the full spatial aperture from both midpoint and source/receiver perspectives.
When enabled, the source-to-source and receiver-to-receiver distances are used symmetrically when selecting raw input traces, so that a trace is included only if both its source and its receiver are within the respective aperture distances. Default: false. Enable this option to enforce strict symmetry in the aperture selection, which is useful for data with regular geometry where cross-spread contamination should be minimized.
When enabled (default: true), the correlation window used during MF image formation adapts in length based on the local signal content, widening the window in low-energy zones and narrowing it in high-energy zones. When disabled, fixed windows of lengths defined by Interpolation window length and Stabilizations window length are used. Disable this option if you observe inconsistent amplitude balancing and prefer a uniform window length throughout the trace.
Half-length of the time window (s) used to compute the MF correlation and weighting for each output sample. Default: 0.010 s (10 ms). The total window is twice this value. Shorter windows provide sharper temporal resolution but may be unstable for low-frequency content; longer windows provide more stable weights but may blur thin-bed events. Adjust this parameter together with Stabilizations window length to balance resolution against stability.
Half-length of the time window (s) used for amplitude stabilisation of the MF weighting coefficients, suppressing division by near-zero values in noise-dominated zones. Default: 0.010 s (10 ms). Increase this value if the output gathers show high-amplitude noise spikes caused by division instabilities in weak-signal zones.
Controls how results are written to the output GSD file. Default: Direct. In Direct mode the output file is created from scratch (or overwritten after a user confirmation prompt). In Append mode, newly computed chunks are added to the end of an existing file. Use Append when restarting a partially completed run or when processing the line in segments. The module automatically skips bins that are already marked as computed in the output file, enabling efficient restart after interruption.
Container grouping the bin range limits that restrict processing to a subset of the output geometry. Useful for testing on a short section of the line, for segmented processing, or for re-running only a specific portion of a 2D line.
Index (crossline number) of the first output bin to process. Default: -1 (no lower limit — start from the beginning of the output geometry). Set to a positive crossline number to begin processing partway along the line. Bin indices start from 1.
Index (crossline number) of the last output bin to process. Default: -1 (no upper limit — process to the end of the output geometry). Set to a positive crossline number to stop processing before the end of the line. Use together with First bin to calc to define a processing window for test runs or for splitting the line into segments across multiple jobs.
Parameters controlling the internal SEG-Y read cache. Increasing the cache size (GB) can improve performance when repeatedly accessing the same seismic traces from a large input file on a slow disk. Thread timer and expiration time settings control how long cached data is retained.
Selects whether the MF reconstruction is performed on the CPU or on available NVIDIA GPU devices. GPU mode uses parallel CUDA kernels and can significantly reduce processing time on long 2D lines with dense trace sampling. GPU mode requires compatible NVIDIA hardware with sufficient video memory for the super-gather and weight buffers.
Options for distributing the computation across multiple compute nodes in a cluster. When enabled, the bin calculation workload is partitioned into chunks and dispatched to remote nodes. This is recommended for very long 2D lines where single-node processing would be too slow.
Number of output bins processed in a single distributed computation chunk. Larger bulk sizes reduce scheduling overhead but increase memory requirements per node. Tune this value based on the available memory on each compute node.
Maximum number of CPU threads used per distributed compute node. Limit this value if other jobs are running concurrently on the same nodes or if memory constraints require fewer parallel threads.
Optional text suffix appended to the distributed job name to distinguish it from other concurrent jobs in the cluster queue. Useful when running multiple Data Enhance jobs simultaneously on the same cluster.
When enabled, allows the thread-to-CPU-core affinity to be set manually via the Affinity parameter. Useful on NUMA systems to pin processing threads to cores with local memory access, improving cache efficiency.
CPU core affinity mask, active only when Set custom affinity is enabled. Specifies which CPU cores the processing threads are bound to.
Number of parallel CPU threads used for the MF reconstruction on the local node. Increase this to the number of available physical CPU cores for best performance. When GPU mode is selected, this value controls the number of GPU devices used in parallel.
Optional script execution hooks. Specify shell scripts or commands to run automatically before or after the module executes.
Path to a script executed before the module starts processing. Use this to set up output directories, check disk space, or stage input data.
Path to a script executed after the module completes. Use this to move output files, trigger downstream jobs, or send notification alerts.
When enabled, the module is bypassed entirely during flow execution. Use this to temporarily disable the enhancement step without removing it from the processing flow.
The module writes a single output file: a GSD-format pre-stack seismic file specified by the Output file name parameter. This file contains the MF-reconstructed gathers for all output bins within the specified calculation area. Each gather holds the synthesised traces at the offsets defined in the output trace header geometry. The GSD file can be connected as input to migration, stacking, or attribute extraction modules in the processing flow. After completion, the file's metadata information is automatically registered in the project.