Sobolev Anisotropic Smooth

Description

Smooths a gather while preserving sharp edges such as fault planes, unconformities or other strong discontinuities. Instead of smoothing uniformly in every direction, the module reduces smoothing across steep local slope changes and applies full smoothing elsewhere, so genuine structural boundaries stay sharp while noise and small-scale roughness are suppressed. It is useful for denoising velocity, dip or attribute volumes without blurring geological boundaries, and for conditioning slope/dip models prior to further interpretation or processing.

Input data

Input gather

The gather (for example a velocity, dip or other attribute section) to be smoothed.

Input slope model

An optional auxiliary gather (such as a dip or edge/slope attribute) used to guide where smoothing should be reduced. When a valid slope model is supplied, edges are detected from it instead of from the input gather itself, which is useful when a dedicated edge/structure attribute gives a cleaner picture of discontinuities than the data being smoothed.

Parameters

Regularization parameter

Controls the overall strength of the smoothing (labeled Lambda). Larger values pull each sample more strongly toward its neighbors, giving heavier overall smoothing away from detected edges; smaller values leave the data closer to its original, noisier state.

Default: 0.05, minimum 0. Increase for noisier input that needs heavier smoothing; reduce toward zero for a gentler, near-pass-through result.

Edge sensitivity parameter

Controls how strong a local slope change must be before it is treated as an edge and protected from smoothing (labeled Kappa). Smaller values make the module sensitive to weak slope contrasts, preserving more, finer detail as "edges"; larger values require a stronger contrast before smoothing is reduced, so only pronounced discontinuities are preserved and more of the data is smoothed uniformly.

Default: 0.05, minimum 0.0000001. Lower it if genuine faults/edges are being over-smoothed; raise it if noise is being mistaken for edges and left unsmoothed.

SIRT Max iteration

The maximum number of iterations the internal iterative solver performs while converging on the smoothed result. More iterations give the solution more chances to converge fully (better accuracy) at the cost of longer run time; fewer iterations run faster but may leave the result short of full convergence, particularly for larger gathers or higher Regularization parameter values.

Default: 300, minimum 0. Increase if the result still looks under-smoothed after the default number of iterations; decrease to speed up processing on large volumes when full convergence is not critical.

Output data

Output gather

The edge-preserving smoothed version of the input gather.

References

Fehmers, G. C. and Höcker, C. F. W., 2003, Fast structural interpretation with structure-oriented filtering: Geophysics, 68, 1286-1293.