|
<< Click to Display Table of Contents >> Navigation: Migration > Kirchhoff PreSDM - file in/out - migration |
Note: This module is deprecated and is no longer recommended for use. It is preserved for compatibility with existing processing flows. For new projects, use the current Kirchhoff PreSDM variants available in the Migration group, such as Kirchhoff PreSDM - file in/out - migration TT or Kirchhoff PreSDM multifile.
This module performs Kirchhoff Pre-Stack Depth Migration (PreSDM) using a file-based input/output workflow. It reads seismic shot or common-midpoint gathers and a precomputed travel-time table from SEG-Y files, applies the Kirchhoff summation operator, and writes the migrated depth image — organized as Common Image Gathers (CIGs) — directly to an output file. The migration can run in 2D or 3D mode and supports both CPU and GPU execution for large datasets.
The Kirchhoff algorithm works by summing seismic amplitudes along diffraction surfaces defined by the travel-time table. Each output depth sample is formed by stacking all input traces whose travel times (source plus receiver) match the depth point location. The migration aperture controls which input traces contribute to each image point, and anti-aliasing filters prevent spatial aliasing artifacts from steep dips or coarse sampling.
The output is a set of offset-domain CIGs, with user-defined offset range and increment. These gathers are used for velocity model quality control: flat events in the CIGs confirm a correct velocity model, while residual moveout indicates velocity errors that require tomographic updating.
The SEG-Y file handle for the input seismic data to be migrated. This is typically a pre-processed dataset of shot gathers or common-offset gathers in the time domain. The file must have geometry (source and receiver coordinates) correctly set in the trace headers.
The trace header collection associated with the input seismic data. The headers define which traces belong to each gather, their shot and receiver coordinates, and their offset values. These headers are used to map each input trace to its correct position in the migration aperture.
The SEG-Y file handle for the precomputed travel-time table. The travel-time table stores, for each image point on the output depth grid, the two-way travel time from source and receiver positions. It is generated by a separate ray-tracing or eikonal solver step using the velocity model before migration begins.
The trace header collection for the travel-time table file. These headers describe the spatial indexing of the travel-time table, enabling the migration engine to look up arrival times for the correct image point locations.
An optional second travel-time table SEG-Y handle for near-offset arrivals. When the Use near offset TT parameter is enabled, this near-offset table provides more accurate travel times for short-offset traces, which may behave differently from the far-offset arrivals captured in the primary travel-time table.
The trace header collection associated with the near-offset travel-time table. Used only when the near-offset TT option is active.
The bin grid defining the spatial positions of output image points. This geometry determines which CMP locations will be included in the migration result. Typically this is the survey bin grid established during geometry loading.
The path and filename for the output migrated volume. The file is written in GSD format (g-Platform Seismic Data). Each image point location produces an offset-domain CIG stored at the corresponding bin position in this output file.
An optional delta-model (DM) velocity perturbation model for amplitude correction. This advanced input is hidden by default and is not used in standard migration workflows.
Enables the primary DM model correction. Hidden by default (default: disabled). Enable only when a delta-model is available and has been configured.
An optional second-order DM perturbation model. Hidden by default and not required for standard migrations.
Enables the secondary DM4 model correction. Hidden by default (default: disabled).
The minimum source-receiver offset (in meters) included in the output Common Image Gathers. Default: 0 m. Traces with offsets smaller than this value are excluded from the migration. Increasing this value can suppress near-offset noise or guided waves that contaminate shallow image points.
The maximum source-receiver offset (in meters) included in the output Common Image Gathers. Default: 3000 m. Set this to match the maximum usable offset in your dataset. Using far offsets improves deep imaging and provides longer lever arms for velocity analysis, but may introduce noise if the signal-to-noise ratio is poor at those offsets.
The offset bin width (in meters) used to organize the CIG output. Default: 50 m. All input traces with offsets falling within the same bin are summed into a single output offset trace. Smaller increments produce denser, more detailed gathers but increase output data size and run time. Larger increments speed up computation and reduce storage but may smooth out AVO or velocity analysis detail.
The maximum lateral radius (in meters) from each image point within which input traces are considered for the Kirchhoff summation. Default: 3000 m. A larger aperture captures steeply dipping reflectors and improves imaging of complex structures, but increases computation time because more input traces must be read and summed for each image point. The aperture should be at least as large as the maximum lateral displacement expected from the target reflectors. An aperture that is too small will clip steep dips and reduce resolution; one that is too large wastes computation on traces that contribute only noise.
Selects whether the migration operates in 3D or 2D mode. Default: disabled (2D mode). Enable this option for 3D surveys. In 2D mode, the aperture search is restricted to the inline direction. In 3D mode, the search extends in both inline and crossline directions, allowing the algorithm to handle true 3D wavefield geometry.
The interpolation method used on the CPU to look up travel-time values between grid nodes. Options: Bilinear (default) or Nearest. Bilinear interpolation produces smoother travel-time surfaces and better image quality at the cost of slightly more computation. Nearest-neighbor lookup is faster but may introduce staircase artifacts in the image, especially for coarsely sampled travel-time tables.
The interpolation method used when the migration runs on a GPU. Options: Texture linear, linear (default), or hyperbolic. Texture linear uses the GPU hardware texture sampler for fast bilinear interpolation. Linear performs software bilinear interpolation in shader code. Hyperbolic applies a hyperbolic moveout approximation, which can be more accurate for certain velocity models but requires additional computation. This parameter has no effect when executing on CPU.
Controls whether and how the migrated amplitude is normalized by the number of traces that contributed to each output sample. Options: none (default), trace, or sample. When set to none, no normalization is applied and amplitude is a simple stack. Trace normalization divides by the number of contributing traces at each image point (corrects for variable fold). Sample normalization divides on a per-sample basis, accounting for fold variations with depth. Fold normalization is recommended in areas of irregular geometry or varying coverage to avoid amplitude artifacts at survey edges.
The number of samples in the depth axis of the output migrated volume. Default: 1000. Together with the Depth ratio parameter, this determines the total depth extent of the migration output. The actual depth range is: Depth samples × Depth ratio. Increase this value to extend the image deeper, at the cost of more output data.
The depth interval between consecutive output samples, in meters. Default: 5 m. This controls the vertical resolution of the output volume. With the default settings (1000 samples × 5 m), the output depth range is 0–5000 m. Choose a depth ratio that provides adequate vertical sampling for your target: approximately one-quarter of the dominant wavelength at the target depth is a common guideline.
When enabled, the migration uses a separate near-offset travel-time table (connected via the Input SEG-Y time table handle (near) input) for short-offset traces. Default: disabled. Enable this when the near-offset travel times differ significantly from the far-offset table, for example in surveys with strong topography or shallow velocity contrasts, to improve shallow image accuracy.
The near-surface replacement velocity in m/s, used to account for topographic static corrections. Default: 1500 m/s. This value is applied when computing travel-time adjustments from the acquisition surface to the datum. Set this to the average near-surface velocity in your survey area. For marine surveys, 1500 m/s (the velocity of water) is typically appropriate. For land surveys, use the velocity of the near-surface weathering layer (often 300–1000 m/s).
A dimensionless scaling factor that controls the aggressiveness of the anti-aliasing filter applied during Kirchhoff summation. Higher values apply a stronger low-pass filter to traces before summation, reducing spatial aliasing on steeply dipping events. Lower values preserve high-frequency content but may allow aliasing artifacts on steep dips. Tune this parameter if you observe dip-dependent ringing or loss of steep-dip signal in the migrated image.
The reference spatial step (in meters) along the X (inline) direction used for anti-aliasing filter design. Default: 15 m. Set this to the inline receiver spacing of your acquisition geometry. The anti-aliasing filter cutoff frequency is inversely proportional to this step: a smaller step allows higher frequencies to pass before filtering, while a larger step applies a more aggressive cut.
The reference spatial step (in meters) along the Y (crossline) direction used for anti-aliasing filter design. Default: 15 m. Set this to the crossline receiver spacing. For 2D surveys, this parameter is not applicable and can be left at the default.
The reference elevation (in meters) of the flat datum plane to which static corrections are applied before migration. Default: 1100 m. The datum should match the reference elevation used when computing the travel-time table. For land surveys with significant topography, the datum is typically chosen above the highest surface elevation. For marine surveys, the datum is usually sea level (0 m).
The maximum frequency (in Hz) preserved in the migration output. Default: 120 Hz. Frequencies above this limit are removed during the anti-aliasing step. Set this to the highest meaningful signal frequency in your data — typically the upper limit of your recording bandwidth. Reducing this value suppresses high-frequency noise in the migrated image but also limits resolution.
The maximum incidence angle (in degrees) at which energy is accepted for summation at each image point. Default: 90 degrees. Restricting this angle limits the dip of reflectors that can be imaged and rejects traces arriving at very oblique angles where ray-theory approximations break down. Reducing the angle limit (for example to 60 degrees) can improve image quality in complex areas by rejecting multiples and other noise that arrives at steep angles. The value of 90 degrees effectively disables angle muting.
An integer factor that thins the input trace population before migration. Default: 1 (no decimation). A value of 2 uses every second trace, reducing computation time by roughly half at the cost of lower fold and potentially noisier output. Use this for quick parameter tests or preview migrations before running the full-density production job.
A container group holding advanced storage options for distributed processing. Expand this group to access the Use local storage and Seismic filename parameters described below.
When enabled, each remote processing node writes its intermediate results to local disk storage rather than to the shared network location. Default: enabled. This can significantly accelerate distributed migration by avoiding network I/O bottlenecks. Disable this only if shared storage access is faster than local disk access on your processing cluster. When using local storage, specify the per-node file path using the Seismic filename parameter below.
The file path on each processing node's local disk where temporary seismic data is staged during distributed migration. Default: /path_to_file.sgy. Update this path to a valid local scratch directory that is accessible on all compute nodes before running a distributed job. Only relevant when Use local storage is enabled.
A container group with advanced settings for spatial interpolation of the output image points (pickets). These parameters control how output image locations are mapped from the input trace geometry. Expand this group to access the parameters below.
The bin spacing (in meters) along the X (inline) axis used for interpolating output image points. Default: 20 m. Valid range: 1–1000 m. This should match the inline bin dimension of your survey grid. Smaller values produce a denser output grid but increase computation and storage.
The bin spacing (in meters) along the Y (crossline) axis used for interpolating output image points. Default: 20 m. Valid range: 1–1000 m. Set this to match the crossline bin dimension of your survey grid.
A constant border (in meters) added around the output image grid boundary. Default: 100 m. This margin ensures that image points near the survey edge receive contributions from nearby traces outside the nominal grid boundary, reducing edge effects in the migrated image. Increase this value if edge artifacts are visible near the survey perimeter.
When enabled, image point coordinates are rounded to the nearest grid node before migration. Default: enabled. Rounding snaps irregularly positioned image points onto a regular grid, which can simplify subsequent post-processing and display. Disable this if you need to preserve the exact sub-grid positions of image points.
The rounding granularity applied to image point coordinates when Round pickets is enabled. Default: 25. Valid range: 1–100000. Coordinates are rounded to the nearest multiple of this value. For example, a round factor of 25 snaps positions to the nearest 25 m grid. Set this to match your survey bin size for consistent alignment of output image points with the survey grid.