|
<< Click to Display Table of Contents >> Navigation: Modeling > Multisource RTM (distributed, 2D) |
The Multisource RTM (distributed, 2D) module performs 2D Reverse Time Migration (RTM) using a multisource blending strategy, with support for distributed execution across a compute cluster. Multiple shot records are simultaneously back-propagated through the depth velocity model, reducing total computation time compared to single-source RTM. To suppress crosstalk noise between blended sources, random phase shifts are optionally applied to each source before blending.
This module is the distributed version of the Multisource RTM module and is designed for large 2D surveys that benefit from spreading the wavefield propagation across multiple cluster nodes. The output is a depth-domain stacked image suitable for structural interpretation.
Link to the SEG-Y file handle providing the pre-stack seismic shot records to be migrated. The data should be pre-processed (denoised, deghosted, surface-consistent statics applied) before being input to RTM.
Link to the trace header vector containing source and receiver geometry information for each input shot record. Correct geometry is essential for the RTM operator to place energy at the correct depth positions.
Link to the depth-domain interval velocity model used for both the forward source wavefield propagation and the backward receiver wavefield propagation. The velocity model should be smooth to avoid numerical dispersion artefacts in the finite-difference scheme.
Reference datum elevation (m) used to align the velocity model with the acquisition surface. Default: 0 m. Set this to the elevation datum of your velocity model if different from mean sea level.
Exponent for a time-domain amplitude compensation function applied to the source wavefield. A positive exponent gradually increases amplitude as a function of time, compensating for geometric spreading. Default: 1. Valid range: 0 to 3.
Exponent for a depth-domain amplitude compensation function applied to the migrated image. Higher values amplify deep image samples to compensate for illumination decay with depth. Default: 1. Valid range: 0 to 3.
When enabled, each source record is phase-randomised before blending, which distributes crosstalk energy incoherently across the image and allows it to be suppressed by stacking multiple super-shots. Disabling this option will cause coherent crosstalk artefacts in the final image. Default: enabled.
Number of individual shot records blended together into one super-shot for simultaneous back-propagation. Default: 1 (no blending). Increasing this value reduces the number of RTM propagation steps by a factor equal to the group size, but also increases crosstalk noise in the image.
Index step between the sources selected to form each multisource group. A larger step increases the spatial separation between blended sources, which reduces coherent crosstalk by distributing interference energy more widely in the image. Default: 1.
Minimum source-index separation between sources assigned to the same multisource group. This prevents closely spaced sources from being blended together, which would produce coherent crosstalk focussed near each source. Default: 1.
Maximum number of super-shot groups to process. Default: 1. Set this to the total number of super-shots needed to cover all input shot records given the chosen group size and step.
When enabled, the RTM image is accumulated (summed) across all multisource groups rather than replacing the image with each new group. Enable this for standard RTM accumulation. Default: enabled.
Small stabilisation constant added to the denominator of the normalised cross-correlation imaging condition to prevent division by zero in low-illumination zones. Default: 1e-9. Increase this value slightly if you observe numerical instabilities in the image at poorly illuminated model edges.
Container group defining the source wavelet shape used in the forward modelling step of RTM.
Dominant frequency (Hz) of the source wavelet. Set this to the dominant frequency of your field data. A higher frequency produces a higher-resolution image but requires a finer modelling grid and smaller time step to avoid numerical dispersion.
Total length (s) of the source wavelet window. Controls how long the wavelet extends in time. Set this to at least 3–4 periods of the dominant frequency.
Shape of the source wavelet used in modelling. Options include Ricker1, Ricker2, AKB, Berlage, Gaussian, GaussianDeriv, MinPhase, Klauder, Ormsby, Spike, Zero, and Unit. Ricker1 is the standard choice for synthetic modelling and produces a zero-phase band-limited pulse.
Time position (s) of the spike within the wavelet window when ImpulseType is set to Spike. Default: 1 s.
Amplitude of the spike wavelet when ImpulseType is set to Spike. Default: 1.
Container group with parameters that control the spatial and temporal discretisation of the finite-difference wave propagation engine.
Spatial grid size (m) used in the finite-difference wave propagation. Default: 5 m. To avoid numerical dispersion, this value should be at most one-fifth of the minimum wavelength (minimum velocity divided by maximum frequency). Smaller values give higher spatial accuracy but increase memory and computation time significantly.
Total number of depth grid points in the modelling volume. Default: 1000. Together with the Model resolution, this defines the maximum depth of the model (Max depth = Number of depth samples × Model resolution).
Width (m) of the absorbing boundary zone added around the modelling grid to damp outgoing waves and prevent artificial edge reflections. Default: 400 m. Increase this value for low-frequency wavefields or if you observe wraparound energy in the output image.
When enabled, applies a free-surface reflection condition at the top of the modelling grid, allowing the simulation to generate surface multiples and ghost reflections in the modelled wavefield. For most RTM imaging applications this option should be disabled to avoid multiple contamination in the image. Default: disabled.
Internal time step (s) used in the finite-difference wavefield propagation. Default: 0.001 s (1 ms). This value must satisfy the CFL stability condition: it must be smaller than the Model resolution divided by the maximum velocity in the model. Reduce this if you observe numerical instability.
Total number of time steps in the wavefield propagation simulation. Default: 4001. Together with the Modelling step, this defines the total simulation time. Increase this value to image deeper reflectors, ensuring the simulation runs long enough for energy to propagate to the target depth and back.