A Robust Multistage Full Waveform Inversion and Its Application

High-resolution velocity model building is crucial for depth migration. Inaccurate model will lead to poor image qualities. Full waveform inversion (FWI) is developed to generate more geologically realistic and higher resolution velocity model iteratively. However, it’s challenging to apply FWI on narrow azimuth acquisition (NAZ) data. Here we proposed a more robust multistage FWI workflow. For the first stage, we introduce dynamic warping preconditioned full waveform inversion (DWFWI) to mitigate the cycle-skipping problem, which creates large scale background updates. In the second stage, the conventional FWI with image-guided regularization (IGFWI) is introduced to overcome the footprint problem and help the convergence, which can add more detailed updates to the velocity model with much higher resolution with more reflection events included and inverted. Due to poor crossline sampling, FWI updates in the crossline direction always contain strong acquisition footprints and swing artifacts. Image guided regularization is powerful to remove the acquisition footprints as well as preserve the detailed update around the faults. This new workflow is applied on the Hoop Fault Complex data in the southwestern Barents Sea. A more geologically realistic and higher-resolution model of the Hoop Fault Complex is obtained with improved migration image and gathers significantly.