|
<< Click to Display Table of Contents >> Navigation: Modeling > 2D Tomo refraction statics |
The 2D Tomo refraction statics module computes near-surface velocity models and refraction static corrections for 2D seismic lines using first-break tomography. It uses the Fresnel wave-path approach combined with the SIRT (Simultaneous Iterative Reconstruction Technique) algorithm to iteratively update the subsurface velocity field so that modelled first-break traveltimes match the observed first-break picks. The resulting velocity model is then used to calculate static corrections that remove the distorting influence of shallow, low-velocity layers on reflection seismic data.
This module supports both CPU and GPU execution, and can distribute processing across multiple compute nodes for large datasets. Before running the tomographic inversion, first-break picks must be imported using the Import FBPicking action. Once picks are loaded, press Solve to run the inversion and generate the static correction output.
The bin geometry item that defines the spatial layout of CMP bins along the 2D line. This geometry is used to construct the lateral grid of the velocity model and to map the computed statics back to the correct surface positions.
The trace header collection for the dataset. Source and receiver coordinates, picket positions, and offsets are read from these headers to associate first-break picks with the correct shot-receiver pairs during the tomographic inversion.
The SEG-Y file handle providing access to the seismic amplitude data. The gather data is used for interactive quality control display of source, receiver, and bin gathers both before and after static corrections are applied.
This group of parameters defines the spatial extent and grid resolution of the near-surface velocity model. All depth and distance parameters within this group are specified in the project distance units (typically metres or feet).
The velocity assigned to the deepest refractor and used as the initial (starting) value for the entire velocity model before the iterative inversion begins. Set this to an approximate velocity of the competent bedrock or the fastest layer expected within the model depth range. The default value is 6000 m/s. If this value is significantly higher or lower than the true sub-weathering velocity, more inversion iterations may be needed to converge.
The reference elevation to which static corrections are computed. All traveltimes and depths in the velocity model are referenced to this datum level. Set this to the project datum elevation. The default value is 1000 m. The model depth grid starts at this elevation and extends downward by the value specified in Max depth from Datum.
The maximum depth of the velocity model measured downward from the datum. This must be large enough to include the deepest refractor contributing to the first-break picks at the longest offsets used. The default value is 1500 m. Setting this too shallow will truncate the model and cause inversion artefacts at depth; setting it unnecessarily large increases computation time.
The vertical grid spacing of the velocity model. A smaller value produces a finer depth resolution but increases memory use and computation time. The default value is 10 m. This should be set to a fraction of the expected thickness of the shallowest layer you wish to resolve.
The horizontal grid spacing of the velocity model along the 2D line. This controls the lateral resolution of the inverted velocity model. The default value is 200 m. Use a value comparable to or smaller than the dominant lateral variation scale of the near-surface you expect. A finer spacing resolves more detail but increases processing time.
The horizontal grid spacing used for the internal traveltime lookup table that is computed from the velocity model to synthesise first-break times. This can be set finer than Step X depth out to improve traveltime accuracy without increasing the output model grid density. The default value is 25 m.
The lateral smoothing distance applied to the velocity model after each iteration. Smoothing suppresses high-frequency noise and prevents the inversion from fitting pick scatter rather than real geology. The default value is 200 m. Increase this value in areas with sparse or noisy first-break picks; decrease it where lateral velocity contrasts are sharp and picks are of high quality.
The minimum picket (line station) coordinate used to define the lateral extent of the velocity model grid. The model is built over the range from Min picket to Max Picket. If left at the default, the model extent is determined automatically from the trace header data.
The maximum picket coordinate used to define the far end of the velocity model grid. Set both Min picket and Max Picket to restrict the model to a subline or to override the automatic extent derived from the data. The default value of -1 means the maximum is set automatically from the data.
The dominant frequency of the seismic wavelet, in Hz. This value is used to compute the width of the Fresnel wave-path zones through which traveltime sensitivity is distributed. A higher frequency corresponds to narrower Fresnel zones and finer spatial resolution in the inversion, but requires the velocity model grid to be proportionally fine. The default value is 20 Hz. Set this to the central frequency of the first-break signal in your data.
The number of times the velocity model is updated during the inversion. Each update recalculates Fresnel wave-paths through the current model and applies a SIRT correction. The default value is 7. Check the convergence plot after each run; if the error is still decreasing at the last iteration, increase this value. Typical values range from 5 to 15.
When enabled, the inversion is allowed to produce velocity models in which velocity decreases with depth (a velocity inversion). When disabled, the model is constrained so that velocity is non-decreasing downward. Disable this option in areas where a simple, geologically realistic layered model is expected. Enable it only if the geology is known to contain low-velocity zones within the near-surface.
The lower bound on velocity values permitted in the model, in m/s. The inversion will not update any model cell below this value. The default is 1000 m/s. Set this to a physically reasonable minimum velocity for the shallowest material in your survey area (for example, the velocity of very loose, unconsolidated sediment).
The upper bound on velocity values permitted in the model, in m/s. The inversion will not update any model cell above this value. The default is 7000 m/s. Set this to a value slightly above the fastest expected bedrock velocity in the survey area to prevent unphysical high-velocity artefacts.
The minimum source-receiver offset, in metres, from which first-break picks are included in the inversion. Picks at very short offsets may be dominated by the direct wave rather than refractions and can distort the near-surface model. The default value is 0 m. Increase this if short-offset picks are unreliable or represent the direct wave only.
The maximum source-receiver offset, in metres, from which first-break picks are included in the inversion. Picks at very long offsets penetrate deeper into the earth and constrain deeper parts of the model, but they also introduce more uncertainty. The default value is 5000 m. Reduce this if long-offset picks are noisy or if you wish to restrict the model to shallower depths.
The offset increment used when sampling picks across the offset range for the regression-based pick interpolation step. The default value is 300 m. This controls how densely the offset axis is sampled when constructing the interpolated traveltime surface used as input to the tomographic inversion.
The spatial aperture, in metres, used for the regression that interpolates first-break pick times onto the model grid. A larger aperture averages over more picks and produces a smoother traveltime surface, reducing the effect of individual pick errors. The default value is 1000 m. In areas with irregular or sparse shooting geometry, a larger aperture improves stability.
The maximum allowable difference between an observed first-break pick time and the traveltime predicted by the current velocity model, expressed in distance units (equivalent to the spatial tolerance in the model coordinate system). Picks that deviate from the model by more than this amount are treated as outliers and excluded from that iteration. The default value is 300 m. Reduce this value to reject more outliers and enforce a tighter fit; increase it in early iterations or when pick quality is variable.
This group controls the iterative solver that computes residual statics corrections from the final tomographic velocity model. These iterations refine the shot and receiver statics after the velocity inversion is complete.
The number of global solver iterations used to refine the residual statics corrections. Each global iteration passes through all shots and receivers to update the static solution. The default value is 2. Increase this if the residual statics convergence plot shows the solution has not stabilised.
The number of local solver iterations performed within each global iteration when computing residual statics. Local iterations refine the statics solution at the level of individual shot or receiver groups. The default value is 2. Increasing this value can improve accuracy in areas with complex near-surface structure but will increase computation time.
Runs the full tomographic inversion using the current parameter settings and the loaded first-break picks. The module iteratively updates the velocity model and displays the evolving model and convergence curve after each iteration. On completion, the computed statics correction is written to the output statics item. First-break picks must be imported before this action can be executed.
Opens a file browser to load first-break pick times from an external file in .fbpicks format. This action must be performed before running Solve. The picks are matched to the trace headers using source and receiver identifiers, and the matched pairs are used as the observed data constraints for the tomographic inversion.