Extracting Fresnel Zones From Migrated Dip-Angle Gathers Using Multitask Learning for 3-D High SNR Imaging

Migrated dip-angle gathers provide a visible domain for estimating Fresnel zones of migration. However, it remains a challenge to extract Fresnel zones from the dip-angle gathers with higher precisions and less workloads, especially in 3-D cases. A commonly used approach is to yield a pair of 2-D dip-angle gathers by stacking the 3-D dip-angle gathers along the inline and crossline direction, then extract the two 1-D Fresnel zones from the resulting pair of 2-D dip-angle gathers, and finally incorporate the two 1-D Fresnel zones into 2-D one. The deep learning-based data-driven method has proven efficient in automatically extracting 1-D Fresnel zones from a 2-D dip-angle gather. However, unlike the current data-driven methods that treat the extraction of each 1-D Fresnel zone as an independent task, we propose a unified multitask learning (MTL) network based on the segmentation model to simultaneously extract a pair of 1-D Fresnel zones from the pair of 2-D dip-angle gathers. The network architecture consists of a shared encoder and multihead decoder modules, thus leveraging shared knowledge and representations across tasks while extracting task-specific features through end-to-end training. The proposed architecture significantly reduces the total number of training parameters and accelerates the training process compared to independent networks for each task [i.e., single-task learning (STL)]. Furthermore, the network’s generalizability is enhanced compared to STL due to the fact that multiple tasks act as regularizations for each other. We demonstrate the method using real 3-D field data. Higher signal-to-noise ratio (SNR) migration results are obtained.