Optimal transport (OT) distances have been recently proposed to mitigate the non-convexity of the L2 misfit function in FWI. However, OT is only applicable to positive and normalized data. To overcome this difficulty, we have proposed to rely on the interpretation of the discrete graph space of the data through OT. In this study, we compare this approach to the classical L2 misfit on the inversion of 3D OBC data from the Valhall field, using a visco-acoustic time-domain FWI algorithm. Starting from a crude 1D linear initial velocity model which generates heavy cycle-skipping when used with L2 misfit, the graph space optimal transport allows recovering an acceptable velocity model down to 2 km depth, while only marginally increasing the computational cost compared to standard L2 FWI.
Presentation Date: Tuesday, October 13, 2020
Session Start Time: 1:50 PM
Presentation Time: 3:05 PM
Location: 361F
Presentation Type: Oral