Automated 3D Boulder Detection

Please note that Boulder Detection module is licensed separately.

Detecting boulders in subbottom sediments is critical for de-risking offshore wind farm installations and minimizing their costs.

High-resolution (HR) and ultra-high-resolution (UHR) seismic is a suitable geophysical method. If the frequency content of the seismic source is fit for the task, the boulders appear on seismic cubes as diffracted events.

The boulder detection method in RadExPro is based on the detection of diffracted events caused by boulders on 3D seismic cubes.

After migration, even relatively weak-amplitude diffractions are focused into local amplitude anomalies, which are particularly easy to notice on time slices. In RadExPro, we aim to interactively detect these post-migration amplitude anomalies by computing a qualitative boulder probability attribute which highlights the anomalies and applying a threshold to this attribute.

A time slice of a migrated 2D cube with example boulders highlighted with a white box.

The same time slice of the qualitative boulder probability, the boulders have the highest amplitudes of the attribute.

A 3D labeling algorithm identifies all the potential boulders (attribute above the threshold) in the cube and computes their spatial properties. These properties then are used to filter out those boulder candidates, which do not fit the a priori constraints on the boulder shape — too elongated, too flat, too big/small, etc.

The same time slice with detected boulders highlighted after most of false positives were filtered out.

The result is exported as a table of detected boulders with their locations, depths, sizes and aspect ratios.