Part 4 – Automatic Fault Plane Extraction
This is the fourth post in our series on fault plane interpretation in OpendTect. In Part 1, we looked at pre-processing and visualization of fault attributes. This was followed by manual interpretation on a plane-by-plane basis and by the standard workflow of picking sticks first and creating planes thereafter.
Now in Part 4, we showcase Automatic Fault Extraction (AFE). This functionality is available in the Faults & Fractures plugin. As input you need a Fault Likelihood (FL) volume with three components: Fault Likelihood (range 0 to 1), Fault strike and Fault Dip. Fault Likelihood is computed algorithmically in the Faults & Fractures plugin. You can also use Thinned Fault Likelihood (TFL) as input to AFE but we recommend using FL. A Fault Probability volume that is created by a Machine Learning model features only one component: the fault likelihood itself. Before we can use this volume as input for Automatic Fault Extraction, we must first compute the missing dip- and strike-components. This is done with the option “Dip/Strike from Fault Probability” in Faults & Fractures.
In the AFE process fault likelihood values are grouped into grids called fault skins. Sliders on skin size, dip and strike are used to control which skins are shown in the 3D scene. Subsequently, skins are selected for fault plane extraction (Left-click to select, Control-click to add to the selection, Control-A to select all). The smoothness of the extracted planes is controlled by the direction and sampling of fault sticks. Optionally, AFE can be used to filter out unwanted features from a TFL volume. For example, if a TFL volume contains many small features, you can compute skins, use the sliders to remove the smallest skins and save the remaining skins as a new TFL volume. As we have seen in Part 1, visualization of TFL can be improved by computing the RMS of TFL in the attribute engine.
And with that, we wrap up our series on fault-interpretation tools in OpendTect.
Thanks for joining, engaging, and sharing your thoughts!
For more information or to request a demo license, please reach out to us: