OpendTect's library of trained Machine Learning models supports a set of powerful models for quickly enhancing post-stack 3D seismic data. For example, there are models for removing random...
OpendTect's library of trained Machine Learning models supports a set of powerful models for quickly enhancing post-stack 3D seismic data.
For example, there are models for removing random noise; suppressing horizontal multiples; predicting fault likelihood; interpolating missing traces and for improving the interpretability of seismic data.
These models are 3D models that have learned to transform blocks of 3D seismic data into a desired target response. The output volume is constructed from the overlapping blocks and blending the outputs in the overlapping zones.
As of OpendTect v7, it is possible to apply trained 3D models to 2D seismic data. In the video we show examples of 3 different models applied to an Ultra High Resolution 2D seismic line that was acquired for Windfarm development.
SimpleDenoise removes random noise; AJAX improves the interpretability of data by removing noise, enhancing the continuity of reflectors, and modifying the amplitude spectrum; SimpleHmult suppresses horizontal multiples. Read the post on LinkedIn
