The Neural Network plugin supports supervised and unsupervised neural networks to combine multiple attributes into "meta-attributes". The main application of unsupervised networks is clustering of attributes and/or waveforms for seismic facies analysis. The supervised approach is used for more advanced seismic facies analysis and to create object "probability" cubes such as TheChimneyCube® and TheFaultCube®. For optimal results the neural network plugin should be combined with the "dip-steering" plugin.
"OpendTect provides state-of-the-art pattern recognition tools to complement your eyeball quantitatively." --Leon Thomsen, Principal, Delta Geophysics, former Senior Advisor and Principal Geophysicist, BP, and former S.E.G. President
Examples
Porosity prediction from Acoustic Impedance. The learning set is constructed from examples extracted along the well tracks. Random line from the porosity volume through the wells. Porosity logs are shown in yellow.
Combining FaultCube (grey) and ChimneyCube (yellow) for fault seal analysis. Yellow spots are related to high fluid flux zones and mud volcanoes. Note conducting and sealing faults and fluid migration at fault intersections.
Reservoir charge and leakage: Chimney detection highlights fluid migration paths from source to reservoir and beyond. Detailed interpretations can thus be used for prospect evaluation and ranking.
The unsupervised approach provides an unbiased view of the data. The 3D objects represent similar seismic attribute responses, which remain to be interpreted by the user. This example shows a Seismic facies cube.