From 1-D Velocity Uncertainty to Horizon Confidence: A Monte-Carlo Ray-Tracing Approach

This work introduces a probabilistic scheme for seismic velocity picking. A water-bottom scan fixes the shallow velocity. Below this anchor, each semblance column is normalised to a probability‐density function (PDF). Its spectral contrast governs a weight that blends this “current” PDF with a window-averaged historical PDF; a coherence penalty discourages erratic jumps. Repeating the tempered draw at every node and looping numerous times forms an ensemble whose mean is the picked velocity and whose standard deviation is a smoothly varying uncertainty that narrows in coherent zones and widens where energy is diffuse or conflicting. To propagate these kinematic uncertainties, each depth-migrated horizon is back-ray-traced through the reference model to the acquisition datum, recording exit time and angle. At every exit point a synthetic zero-offset trace carrying a chosen wavelet is planted. For every perturbed velocity realisation, aperture-limited post-stack depth migration around the stored angle re-images those traces, yielding a cloud of horizon nodes that captures vertical and dip-dependent lateral error. This workflow is able to estimate 1D velocity uncertainty and propagate it to 2D horizon uncertainties for geologic scenarios, which do not strongly deviate from plane-parallel layering.