Velocity model building for variable-depth streamer data: Brunei case study

Abstract

The computation of velocity by tomography using Residual Move-Out (RMO) curves is widely used in the industry. The semblance-based, automatic dense picking of RMO has proved to be a very efficient technique for conventional data. However, in the case of variable-depth streamer acquisition, this technique is poorly adapted due to the presence of ghosts in the data. We present here a robust, automatic, high-order dense picking technique based on a cross-semblance criteria using the migrated and mirrormigrated common image gathers that eliminates the need for pre-stack deghosting prior to velocity update. This paper focuses on velocity model building using our method. It is applied to a complex 2D variable-depth streamer acquisition from Brunei, which is characterized by noisy data where faults and gas pockets coexist. We show that our method allows us to obtain an accurate velocity function conformable to structure that improves the focusing.