Statistical inversion of variable salt velocity by neural network classification in the central Gulf of Mexico

Li, Hongyan (Schlumberger) | Hutson, Jarrod (Schlumberger) | Weir, Brian (Schlumberger) | Peng, Chuck (Schlumberger) | Koechner, Betty (WesternGeco)


Seismic velocity in salt domes in the Gulf of Mexico varies due to sediment inclusions and sutures. Typically, seismic velocity of salt is slower than clean salt velocity: the maximum slowdown can be more than 20%. Therefore, it is extremely important to build a velocity model with variable salt velocity to improve the base salt interpretation, subsalt imaging and better well ties at the subsalt level. Using different seismic attributes (envelope, root-mean-square amplitude, absolute amplitude, gradient, and others) as inputs, a neural network classifies a seismic image of a salt body into a set of classes. Different classes are mapped to different salt velocities. The scalars used in the mapping are adjusted using the sonic velocity inside of the salt as calibration and the base salt tie to the well markers on seismic images. This workflow was applied to a large reprocessing project in the central Gulf of Mexico; this method has been proven to be effective and efficient, resulting in improvement in base salt events and substantially improved subsalt imaging as well as base salt well tie, all while offering improved turnaround time.

Presentation Date: Wednesday, October 19, 2016

Start Time: 2:20:00 PM

Location: 143/149

Presentation Type: ORAL