|
<< Click to Display Table of Contents >> Navigation: Multiples > SRME 2D/3D (obsolete) |
This module is deprecated. Use the current SRME 2D/3D module for new projects.
SRME (Surface-Related Multiple Elimination) is a data-driven multiple attenuation method that models water-bottom and surface-related multiples by cross-convolving the recorded wavefield with itself. No velocity model is required: the algorithm uses the input seismic data and the geometry of sources and receivers to predict the multiple wavefield. The predicted multiples are then adaptively subtracted from the input to recover the primary-only record.
Two multiple-modeling modes are available: Water depth models surface multiples using the sea-floor reflection and a water-layer replacement velocity; Horizon Velocity model uses a user-supplied horizon and velocity field for more accurate multiple prediction in complex water-bottom environments. After multiple prediction, an optional adaptive subtraction step minimises energy leakage from the subtraction by matching the predicted multiple amplitude and phase to the actual multiple energy.
Reference to the SEG-Y file object containing the full pre-stack dataset. The module reads all traces from this file during multiple prediction to build the cross-convolution model.
Sorted trace header index for the full dataset. This index is used to efficiently locate the traces needed for multiple modeling for each output CMP gather.
The BinGrid defining the inline/crossline survey geometry. Required for 3D SRME to determine which source-receiver combinations contribute to the multiple model for each CMP bin.
Path to the output GSD file where the demultipled seismic data will be written. Click the browse button to choose the location. The GSD format is g-Platform's native seismic data container.
Minimum source-receiver separation tolerance used when searching for contributing trace pairs during multiple prediction, in meters. Default: 25 m. Trace pairs whose midpoint distance falls below this threshold are accepted as contributing to the multiple model. Smaller values produce a more accurate but slower model.
Maximum source-receiver separation tolerance for trace pair selection, in meters. Default: 100 m. Trace pairs with midpoint distance up to this value are included in the multiple model. Increasing this parameter allows more contributing pairs but may degrade prediction accuracy if the wavefield varies rapidly with position.
Maximum inline search aperture for trace pair selection during multiple prediction, in meters. Default: 1500 m. This limits the inline extent of the cross-convolution summation. Larger apertures include more data and can improve the completeness of the multiple model, but increase computation time.
Maximum crossline search aperture for trace pair selection, in meters. Default: 500 m. For 2D surveys, set the crossline aperture to zero or a small value. For 3D surveys, this aperture controls the out-of-plane contribution to the multiple model.
When enabled, extrapolates the wavefield to near offsets where traces are missing (near-offset gap). Default: disabled. Enable when the survey has a significant near-offset gap that would otherwise cause incomplete multiple prediction at short source-receiver separations.
Water-layer replacement velocity used in the Water depth multiple prediction mode, in m/s. Default: 1530 m/s. Set this to the acoustic velocity of the water column in your survey area. Typical seawater values range from 1480 to 1550 m/s. Only used when Multiple Type is set to Water depth.
Selects the source of water-bottom information used for multiple prediction. Water depth uses a constant replacement velocity and an optional Bottom Depth matrix to define the sea-floor reflection. Horizon Velocity model uses an interpreted horizon and a spatially varying velocity model for more accurate multiple prediction in areas of complex or variable bathymetry.
Container group providing the data inputs for the selected Multiple Type mode and controlling mute application. Input horizons connects the picked horizon object (active in Horizon Velocity model mode). Bottom Depth matrix provides a gridded water-bottom depth map (active in Water depth mode). Mute Input Data applies a mute to the input traces before multiple prediction. Mute Output Data (default: enabled) applies a mute to the predicted multiple model before subtraction. Mute taper (default: 26 ms) sets the taper length applied at the mute boundary. Correction Type selects whether horizon-based travel-time correction is applied (Use Horizon Correction) or not (NONE).
Container group for advanced processing options. Clip Small values (default: enabled) applies amplitude clipping to suppress very small values in the predicted multiple model. Clip Small values of Amplitude (default: 0.01 %) sets the clipping threshold as a percentage of the maximum amplitude. Accuracy is function of offset (default: disabled) enables offset-dependent accuracy distance scaling. Saving mode (for debugging) controls whether the output file is written with append or direct (overwrite) mode.
Container group controlling the adaptive subtraction filter used in the interactive visualization. This filter matches the predicted multiple model to the actual multiple energy before subtraction to minimise over- or under-subtraction artefacts. The following sub-parameters are available:
Horizontal window (default: 15 traces) — number of traces in the lateral window used to estimate the adaptive filter. A wider window gives a more stable but less spatially adaptive filter.
Vertical window (default: 40 ms) — time window length used to estimate the adaptive filter. Should be long enough to encompass the wavelet of the multiple.
Min vertical shift (default: 0 ms) and Max vertical shift (default: 40 ms) — range of time shifts over which the adaptive filter is searched to account for timing errors in the predicted multiple.
Step vertical shift (default: 16 ms) — step interval for the shift search.
Subtraction type — solver algorithm for the adaptive subtraction: LSQR (least-squares, default), LSQR advance, or FISTA (fast iterative shrinkage-thresholding for sparse solutions).
Lamda (default: 0.0001 %) — regularisation parameter for the adaptive subtraction solver. Increase this value to add more regularisation when the subtraction produces instabilities.
Container group limiting multiple attenuation to a sub-area of the survey. The area can be defined by inline/crossline range or by sequence number range, depending on the Calculation area mode setting. All range parameters default to -1 (no limit). Set these to restrict processing to a test area before running the full dataset.
SEG-Y data caching parameters controlling memory use during random-access trace reading. Larger cache sizes improve performance when the multiple prediction requires frequent access to many traces from different parts of the dataset.
Selects whether the multiple prediction computation is performed on the CPU or GPU.
Options for distributing the SRME computation across multiple nodes in a cluster environment.
Minimum number of gathers in each distributed computation chunk.
Maximum number of CPU threads per distributed processing node.
Optional text suffix appended to the distributed job name.
When enabled, allows specifying a custom CPU core affinity mask for the processing threads.
CPU affinity mask. Only active when Set custom affinity is enabled.
Number of CPU threads used for parallel multiple prediction. Set to the number of available physical cores for best performance.
Options for executing external scripts before or after this module runs.
Path to a script that is executed before this module begins processing.
Path to a script that is executed after this module finishes processing.
When enabled, this module is bypassed and the data passes through unchanged.
Interactive display showing the raw input CMP gather for the currently selected inline/crossline position. Used for QC comparison with the demultipled output.
Interactive display showing the input gather after any muting or preprocessing applied prior to multiple modeling. Useful for verifying the mute settings before running the full prediction.
Interactive display of the predicted multiple model for the selected gather. Inspect this panel to assess the quality of the multiple prediction before committing to full-volume processing.
Interactive display of the demultipled output gather for the selected position, showing the input minus the predicted multiples after adaptive subtraction.
Displays the sequence number of the currently selected CMP gather in the interactive display.
Displays the inline number of the currently selected CMP gather.
Displays the crossline number of the currently selected CMP gather.
Triggers multiple prediction for the currently selected CMP gather and updates the Selected GatherOut Multiples display. Use this action to interactively preview and tune the multiple prediction for a representative gather before running full-volume processing.
Applies adaptive subtraction to the predicted multiples for the currently selected gather and displays the result in the Selected GatherOut panel. Use this action to optimise the adaptive subtraction parameters (window sizes, solver type, regularisation) before committing to full processing.