|
<< Click to Display Table of Contents >> Navigation: Tomography > Grid tomography 2D / 3D update |
This module performs grid-based tomographic inversion to update a depth interval velocity model using NIP (Normal Incidence Point) wavefield attributes. It takes a residual M-gather — containing the difference between observed and modeled NIP parameters — together with an existing depth velocity model, and iteratively solves a damped least-squares inversion to produce a refined velocity grid that better explains the observed data. Both 2D (single inline) and full 3D (volume) geometries are supported.
The inversion proceeds through a series of global iterations in which the velocity grid is progressively refined: the grid spacing is reduced by the resolution factor at the start of each subsequent iteration, allowing the inversion to converge from a coarse large-scale model to a fine-detail result. Outputs include the updated depth velocity model, the updated NIP grid velocity object, and a tomographic velocity gather that can be visualised as a depth image.
Note: This module is deprecated. It is retained for compatibility with existing workflows. New projects should use the current Grid tomography 2D / 3D module.
The residual NIP M-attribute gather produced by the preceding tomographic modelling step. Each trace in this gather carries the difference (residual) between the observed and predicted NIP parameter (M) at each bin location and two-way time. The module reads the bin positions from this gather to define the spatial extent of the inversion domain automatically. This gather must be provided before running the inversion.
The initial depth interval velocity model in NIP grid format. This model is used as the starting point for the inversion and is resampled onto the inversion grid at each global iteration. The model must already be in the NIP grid velocity format (produced by a prior Grid tomography step); standard RMS or interval velocity cubes are not accepted here.
Choose whether the inversion is performed in 2D or 3D. Select 2D for single-line (inline) data where no crossline extent is needed; select 3D for full 3D surveys with both inline and crossline coverage. In 2D mode, all Y-direction grid parameters and the crossline regularisation weights (sigPxiy, sigXiy) are not used. Default: 3D.
The ray-tracing step length (in metres) used when computing synthetic traveltimes through the velocity model during the modelling phase of each iteration. Smaller values increase ray-path accuracy at the cost of longer computation time. The default value of 20 m is adequate for most surveys; reduce this value if the velocity model contains very sharp lateral contrasts or if high accuracy is required in shallow, thin layers.
This group controls the iterative least-squares inversion solver and the regularisation (damping) weights applied to each NIP wavefield attribute. Regularisation prevents the solver from fitting noise by penalising large deviations of the inverted attributes from their initial values.
The total number of outer (global) inversion iterations to perform. In each global iteration the velocity grid is refined by dividing the current grid spacing by the resolution factor, and a new local inversion is solved on the refined grid. More iterations allow the tomography to achieve finer spatial resolution, but also increase computation time. Default: 5. Typical values range from 3 to 8.
The number of inner (local) conjugate-gradient iterations performed by the LSQR solver within each global iteration. Increasing this value allows the solver to converge more fully at each grid resolution level. Default: 12. If convergence is slow or the output model shows artefacts, try increasing to 20–30.
Regularisation weight (damping coefficient) applied to the velocity perturbation. A larger value constrains the inversion more strongly, keeping the updated velocity close to the starting model and producing smoother results. A smaller value allows greater velocity updates and can recover finer detail but may also amplify noise. Default: 1000. Adjust this parameter based on the expected magnitude of velocity updates relative to the starting model.
Regularisation weight for the NIP M-attribute (related to the normal curvature of the wavefront). Controls how strongly the inverted M values are constrained towards their initial estimates. Default: 1. Increase this value if the M residuals are noisy and the inversion produces unstable results.
Regularisation weight for the two-way traveltime attribute. This dampens updates to the traveltime component of the inversion system. Default: 1. In most cases the default is appropriate; increase if the inversion shows oscillations in the traveltime domain.
Convergence threshold for the depth component of the inversion. The solver stops iterating when the depth update falls below this value. Default: 0.001. This value is expressed in the same units as the depth model (metres). Reducing this value enforces tighter convergence but does not always improve the quality of the final model if data noise is the limiting factor.
Regularisation weight for the inline component of the NIP slowness vector (Pxi in the inline X direction). Higher values constrain the inline slope of the NIP ray more strongly. Default: 2. Setting this to zero disables the constraint for this component. Used in both 2D and 3D modes.
Regularisation weight for the crossline component of the NIP slowness vector (Pxi in the crossline Y direction). Default: 2. This parameter is active only in 3D mode; it is ignored in 2D.
Regularisation weight for the inline NIP position (Xi in the inline direction). Controls how tightly the inversion constrains the horizontal position of each NIP along the inline axis. Default: 1. Used in both 2D and 3D modes.
Regularisation weight for the crossline NIP position (Xi in the crossline Y direction). Default: 1. Active only in 3D mode; ignored in 2D.
This group defines the spatial extent, reference velocity, and grid spacing of the tomographic velocity grid. The grid is built as a set of nodes (knots) spanning the survey area; the velocity is solved at each node and then interpolated to produce the continuous velocity model.
The near-surface replacement velocity (in m/s). This value is used as the reference velocity at the datum level and anchors the inversion when the Use V0 as constrain option is enabled. Set this to the known or estimated velocity of the weathering layer or replacement velocity used during statics application. Default: 1500 m/s.
When enabled, the v0 value is used as a hard constraint at the datum level, preventing the inversion from updating the near-surface velocity. This is useful when the replacement velocity is well known and should not be perturbed by the tomography. Default: enabled. Disable only if there is genuine uncertainty in the near-surface velocity and sufficient data coverage to resolve it.
The shallowest depth (in metres) at which the tomographic velocity grid begins. Velocity nodes above this depth are not updated by the inversion. Default: 0 m. When the Residual M update Gather is connected, this value is updated automatically from the data. Set manually if you need to restrict the inversion to a specific depth range.
The maximum depth (in metres) of the tomographic velocity grid. Velocity nodes below this depth are not included in the inversion. Default: 8000 m. Set this value to match the deepest reflector of interest. Using a Z end value much deeper than the deepest data will create under-constrained nodes at depth and may destabilise the inversion.
The minimum inline coordinate (in metres) of the tomographic grid. When the Residual M update Gather is connected, this value is automatically populated from the minimum bin position in the data. Default: 0 m. In most workflows, this does not need to be changed manually.
The maximum inline coordinate (in metres) of the tomographic grid. Automatically populated from the maximum bin position in the Residual M update Gather when the data is connected. Default: 20000 m.
The minimum crossline coordinate (in metres) of the tomographic grid. Automatically populated from the data. Default: 0 m. Active only in 3D mode.
The maximum crossline coordinate (in metres) of the tomographic grid. Automatically populated from the data. Default: 20000 m. Active only in 3D mode.
The factor by which the grid spacing is divided at the start of each subsequent global iteration. For example, with an initial X grid step of 3000 m and a resolution factor of 1.5, the grid step after the first iteration becomes 2000 m, then approximately 1333 m, and so on, until the final grid step is reached. A larger resolution factor increases the rate of grid refinement across iterations. Default: 1.5. Typical values are 1.2 to 2.0.
The number of ray paths computed per velocity grid node interval. Increasing this value densifies the ray coverage between nodes, producing a more complete and better-constrained Jacobian matrix for the inversion. Higher values improve accuracy but increase computation time roughly in proportion. Default: 3. Minimum value: 1.
When enabled, the initial grid step sizes (X, Y, and Z) are taken directly from the spacing of the input depth velocity model rather than from the manually specified initial grid step values. This is useful when you want the first inversion pass to exactly reproduce the resolution of the prior model before refining. Default: disabled. You can also trigger this action manually using the Get params from init model button.
The spacing (in metres) between velocity grid nodes in the inline direction at the start of the first global iteration. This defines the coarsest inline resolution used by the inversion. Default: 3000 m. This parameter is hidden when Use grid step from initial model is enabled.
The spacing (in metres) between velocity grid nodes in the crossline direction at the start of the first global iteration. Default: 3000 m. Active only in 3D mode. Hidden when Use grid step from initial model is enabled.
The spacing (in metres) between velocity grid nodes in the depth direction at the start of the first global iteration. Default: 2000 m. Hidden when Use grid step from initial model is enabled.
The minimum allowed inline grid spacing (in metres). The iterative refinement will not reduce the inline grid step below this value, regardless of the number of global iterations and the resolution factor. Default: 500 m. This should be set to a value consistent with the expected spatial resolution of the velocity field and the bin spacing of the survey.
The minimum allowed crossline grid spacing (in metres). The iterative refinement will not reduce the crossline grid step below this value. Default: 500 m. Active only in 3D mode.
The minimum allowed depth grid spacing (in metres). The iterative refinement will not reduce the depth grid step below this value. Default: 500 m. Set this to match the required vertical resolution of the output model.
The datum depth (in metres) used as the reference elevation for the inversion. During processing this value is automatically set to the maximum surface elevation (Z value) read from the input gather. It may also be set manually to a fixed datum if the survey was acquired with a flat datum. The datum defines the upper reference surface from which the NIP depth positions are measured. Default: 0 m (overridden at runtime from data).
This group controls the format and extent of the output depth interval velocity model that is written at the end of each global iteration.
The folder path where intermediate and final output velocity model files are saved. Specify a valid directory path on the local or network file system. If left empty, files are written to the default project output directory. Saving to a local fast disk is recommended to minimise I/O overhead during multi-iteration runs.
The total depth extent (in metres) of the output velocity model. The tomographic velocity gather and the saved velocity grid will span from the datum down to this depth. Default: 8000 m. This value should be equal to or greater than the Z end parameter to ensure the full inversion depth range is represented in the output.
The depth sampling interval (in metres) of the output velocity model. The inverted velocity field is resampled onto a regular depth grid with this interval before being saved and passed to the output data items. Default: 5 m. Use a value commensurate with the final Z grid step — setting Delta Z much finer than the final grid step will not add resolution but will increase the size of the output files.
A gather of depth traces representing the final interval velocity model in the depth domain. Each trace corresponds to one bin location and contains the interval velocity as a function of depth, sampled at the Delta Z interval down to the specified Depth. This gather can be displayed as a 2D depth image using the built-in View2D Depth Image viewer. It is updated at the end of each global iteration so that progress can be monitored.
The updated depth interval velocity model in NIP grid format. This output is equivalent in format to the input Depth velocity and can be passed directly as input to the next iteration of the Grid tomography 2D / 3D or Grid tomography 2D / 3D update module, enabling successive model refinement passes.
The inverted velocity stored in the internal NIP grid velocity object. This item carries the full 2D or 3D velocity grid at the final iteration resolution and is the primary object used for downstream depth imaging or model quality control within the g-Platform depth processing workflow.
Clicking this button reads the grid node spacing from the connected input depth velocity model and populates the Initial X grid step, Initial Y grid step, and Initial Z grid step fields accordingly. It also reads the spatial bounding box from the Residual M update Gather and updates the X start, X end, Y start, Y end, and Datum fields. Use this action to initialise the grid parameters consistently with an existing model before starting a new update pass.