Seismic Velocity Model Updating via Migration Scan Travel Times

The patent introduces migration velocity scans as an intermediary between the noisy seismic data and the velocity model updating process. Instead of directly using residual depth error from noisy gathers, the method creates a suite of migrated images using test velocity models, then uses image quality metrics (signal-to-noise ratio, image power, geological reasonableness) as intermediate indicators to select optimal velocity models. This intermediary approach filters out the noise impact and provides more reliable velocity updates.

The patent changes the parameter space from direct residual depth error measurement to image quality parameters. By evaluating multiple test velocity models and comparing their resulting images using metrics like signal-to-noise ratio, stacked image power, and geological reasonableness, the method transforms the velocity updating problem from a direct measurement task to a comparative optimization task, thereby improving reliability under noisy conditions.

The patent segments the velocity model updating process into distinct stages: generating a suite of test velocity models, migrating data with each model, evaluating image quality, and selecting optimal models. This segmentation allows the complex problem to be broken down into manageable steps, where each stage can be optimized independently. The use of image quality metrics as intermediate evaluation points creates natural breakpoints in the processing workflow.

The patent applies partial action by selecting a subset of test velocity models for detailed evaluation rather than processing all possible models. The method uses image quality thresholds to identify sufficiently good models without exhaustively analyzing every possibility. This partial approach maintains reliability while reducing the computational burden of evaluating the complete parameter space.

The patent implements feedback loops where image quality metrics from migrated images are used to evaluate and select optimal velocity models. The process continuously refines the velocity model by comparing predicted images with observed data quality, using the image power, signal-to-noise ratio, and geological reasonableness as feedback signals. This iterative feedback mechanism makes the complex spatially varying updates more manageable by providing clear evaluation criteria at each step.

The patent replaces direct mechanical measurement of residual depth error with an information-based system using image quality metrics. Instead of relying on direct geometric measurements from noisy gathers, the method uses computational image evaluation and statistical metrics to infer velocity model quality. This substitution transforms a difficult direct measurement problem into a more tractable information processing problem.